id
stringlengths 10
11
| text
stringlengths 0
185k
| title
stringlengths 0
273
| date
stringlengths 0
10
| authors
stringlengths 0
356
| language
stringclasses 2
values |
---|---|---|---|---|---|
PMC4552518 | ORIGINAL RESEARCH
Unchanged content of oxidative enzymes in fast-twitch
muscle fibers and _VO2 kinetics after intensified training in
trained cyclists
Peter M. Christensen1,2, Thomas P. Gunnarsson1, Martin Thomassen1, Daryl P. Wilkerson3,
Jens Jung Nielsen1 & Jens Bangsbo1
1 Department of Nutrition, Exercise and Sports, Section of Integrated Physiology, University of Copenhagen, Copenhagen, Denmark
2 Team Danmark (Danish Elite Sport Organization), Copenhagen, Denmark
3 Sport and Health Sciences, St Luke’s Campus, University of Exeter, Exeter, UK
Keywords
_VO2 kinetics, high intensity training, interval
training, type I fibers, type II fibers.
Correspondence
Jens Bangsbo, Department of Nutrition,
Exercise and Sports, Section of Integrated
Physiology, University of Copenhagen,
August Krogh Building, Universitetsparken
13, 2100 KBH Ø, Denmark.
Tel: +45 35 32 16 23
Fax: +45 35 32 16 00
E-mail: [email protected]
Funding Information
The study was supported by Team Danmark
(Danish Elite Sport Organization).
Received: 8 May 2015; Accepted: 19 May
2015
doi: 10.14814/phy2.12428
Physiol Rep, 3 (7), 2015, e12428,
doi: 10.14814/phy2.12428
Abstract
The present study examined if high intensity training (HIT) could increase
the expression of oxidative enzymes in fast-twitch muscle fibers causing a fas-
ter oxygen uptake ( _VO2) response during intense (INT), but not moderate
(MOD), exercise and reduce the _VO2 slow component and muscle metabolic
perturbation during INT. Pulmonary _VO2 kinetics was determined in eight
trained male cyclists ( _VO2-max: 59 4 (means SD) mL min1 kg1) dur-
ing MOD (205 12 W ~65% _VO2-max) and INT (286 17 W ~85% _VO2-
max) exercise before and after a 7-week HIT period (30-sec sprints and 4-min
intervals) with a 50% reduction in volume. Both before and after HIT the
content in fast-twitch fibers of CS (P < 0.05) and COX-4 (P < 0.01) was
lower, whereas PFK was higher (P < 0.001) than in slow-twitch fibers. Con-
tent of CS, COX-4, and PFK in homogenate and fast-twitch fibers was
unchanged with HIT. Maximal activity (lmol g DW1 min1) of CS (56 8
post-HIT vs. 59 10 pre-HIT), HAD (27 6 vs. 29 3) and PFK
(340 69 vs. 318 105) and the capillary to fiber ratio (2.30 0.16 vs.
2.38 0.20) was unaltered following HIT. _VO2 kinetics was unchanged with
HIT and the speed of the primary response did not differ between MOD and
INT. Muscle creatine phosphate was lower (42 15 vs. 66 17 mmol kg
DW1) and muscle lactate was higher (40 18 vs. 14 5 mmol kg DW1)
at 6 min of INT (P < 0.05) after compared to before HIT. A period of inten-
sified training with a volume reduction did not increase the content of oxida-
tive enzymes in fast-twitch fibers, and did not change _VO2 kinetics.
Introduction
It has been known for decades that endurance training
results in a faster increase in the pulmonary oxygen
uptake ( _VO2) response in the initial phase of exercise
(Hickson et al. 1978). During constant load exercise at
intensities above the gas exchange threshold (GET) _VO2
continues to rise at a slow rate, with training reported to
reduce the magnitude of this ‘ _VO2 slow component’
(Jones et al. 2011). These alterations may be due to ele-
vated muscle oxidative enzyme capacity and greater oxy-
gen delivery (Jones and Poole 2005) as have been found
in training studies using untrained subjects with a maxi-
mal _VO2 ( _VO2-max) of ~50 mL min1 kg1 (Saltin et al.
1976; Phillips et al. 1995; Shoemaker et al. 1996; Krustrup
et al. 2004a). In the study of Krustrup et al. subjects per-
formed high intensity training (1 min intervals) and it
was reported that post-training leg _VO2 kinetics was fas-
ter during intense but not moderate exercise when com-
pared to pretraining, respectively (Krustrup et al. 2004a).
This may have been caused by adaptations in fast-twitch
(FT) muscle fibers recruited during training, since these
ª 2015 The Authors. Physiological Reports published by Wiley Periodicals, Inc. on behalf of
the American Physiological Society and The Physiological Society.
This is an open access article under the terms of the Creative Commons Attribution License,
which permits use, distribution and reproduction in any medium, provided the original work is properly cited.
2015 | Vol. 3 | Iss. 7 | e12428
Page 1
Physiological Reports ISSN 2051-817X
fibers have been reported to have lower oxidative maxi-
mal enzyme activity than slow-twitch (ST) muscle fibers
(Essen et al. 1975; Essen-Gustavsson and Henriksson
1984; Schantz and Henriksson 1987). In support for
training causing fiber type-specific adaptations, maximal
activity of oxidative enzymes in FT muscle fibers can
reach the same level as ST fibers in highly trained athletes
( _VO2-max: ~70 mL min1 kg1) (Jansson and Kaijser
1977; Chi et al. 1983) and a training study with untrained
subjects observed an increase in maximal activity in pools
of FT fibers following intense training (Henriksson and
Reitman 1976).
Training
studies
using
untrained
subjects
typically
result in several adaptations. Therefore, it cannot be
determined if one adaptation (e.g., increase in oxidative
enzyme capacity) is of importance for the faster rise in
_VO2 at the onset of exercise and the alterations in the
_VO2 slow component when other adaptations also occur
in response to training (e.g., increase in bulk oxygen
delivery or reduced mean transit time at the capillary level
due to increased capillary density). Instead, studying the
effect of an altered training regime of trained subjects
could potentially provide insight into what factors are
important to improve the _VO2 kinetics since fewer mus-
cular and vascular adaptations are expected to occur. Fur-
thermore, just a few studies have been conducted with
trained athletes and some (Dufour et al. 2006; Christen-
sen et al. 2011), but not all studies (Norris and Petersen
1998; Demarle et al. 2001), reported that _VO2 kinetics
remained unaltered. However, comparison is difficult due
to various designs used ranging from high-volume train-
ing for 8 weeks at moderate intensity at the start of the
season in cyclists (Norris and Petersen 1998), 6 weeks in
runners encompassing both moderate and intense bouts
(~90% _VO2-max in hypoxia or normoxia) (Dufour et al.
2006), to intense training (<5 min) either with a reduc-
tion in volume in soccer players for 2 weeks at the end of
the season (Christensen et al. 2011) or a maintained
training volume in runners for 8 weeks (Demarle et al.
2001).
Recruitment of FT fibers has been implicated in the
development of the _VO2 slow component (Barstow et al.
1996; Jones et al. 2011). Thus, it may be that an increased
content of oxidative enzymes in FT fibers could reduce
the _VO2 slow component in trained athletes since a high
proportion of FT fibers has been associated with a large
slow component (Barstow et al. 1996) and since FT fibers
appear to have a lower maximal oxidative enzyme activity
than ST fibers (Essen et al. 1975; Essen-Gustavsson and
Henriksson 1984; Schantz and Henriksson 1987) unless
vigorous training has been performed (Jansson and Kaij-
ser 1977; Chi et al. 1983). Moreover, since oxidative
enzymes are thought to be implicated in the rate of the
rise in _VO2 at the onset of exercise (Jones and Poole
2005), it may be that intense training targeting both ST
and FT fibers is essential for eliciting faster _VO2 kinetics
in trained athletes. This argument is supported by an aug-
mented signaling for cascades involved in the synthesis of
oxidative enzymes in well-trained cyclists ( _VO2-max:
68 mL min1 kg1) following repeated 30-sec sprinting
compared with less intense exercise (Psilander et al.
2010). Also, in less trained subjects repeated 30-sec
sprinting activated signal cascades involved in the synthe-
sis of oxidative enzymes to a similar extent as long dura-
tion low intensity exercise (Little et al. 2010, 2011). The
response of these signaling cascades increases in an inten-
sity-dependent manner (Egan et al. 2010; Nordsborg et al.
2010), which could be related to recruitment of more
muscle fibers. In runners subjected to intense training in
combination with a marked lowering of low- and moder-
ate
intensity
training
(~25–50%)
maximal
oxidative
enzyme activity in muscle homogenate was maintained
(Bangsbo et al. 2009; Iaia et al. 2009) which could be due
to adaptations in FT fibers since detraining is known to
lower activity of oxidative enzymes (Chi et al. 1983;
Christensen et al. 2011). Thus, repeated 30-sec sprinting
separated by ~4 min of rest may be an effective training
regime to cause increases in oxidative enzymes in FT
fibers. The majority of studies in untrained subjects
(Jacobs et al. 1987; MacDougall et al. 1998; Burgomaster
et al. 2005, 2008; Gibala et al. 2006) did not evaluate fiber
type-specific adaptations in oxidative enzymes, but recent
studies do suggest that repeated 20–30 sec sprints can
increase markers of oxidative capacity in both fiber types
in untrained subjects (Shepherd et al. 2013; Scribbans
et al. 2014). However, such evaluation has not been per-
formed on trained individuals following intense training
with 30-sec sprints (Bangsbo et al. 2009; Iaia et al. 2009;
Christensen et al. 2011; Gunnarsson et al. 2012). Oxida-
tive capacity is higher in ST than FT fibers (Essen et al.
1975; Essen-Gustavsson and Henriksson 1984; Schantz
and Henriksson 1987, Shepherd), which means that the
potential for improvements likely is greater in the FT
fibers supported by similar capacity in the two fiber types
in highly trained subjects (Jansson and Kaijser 1977; Chi
et al. 1983).
A faster _VO2 response is associated with reductions in
the anaerobic contribution to the total energy turnover at
the onset of exercise. This has been determined from
muscle metabolites and blood lactate either as a result of
performing prior exercise (Bangsbo et al. 2001; Krustrup
et al. 2001) or performing exercise training either with
low intensity and high volume (Phillips et al. 1995; Green
et al. 2009) or high intensity and low volume (Burgomas-
ter et al. 2006). However, limited knowledge exists about
muscle anaerobic energy turnover after a period of high
2015 | Vol. 3 | Iss. 7 | e12428
Page 2
ª 2015 The Authors. Physiological Reports published by Wiley Periodicals, Inc. on behalf of
the American Physiological Society and The Physiological Society.
Metabolic and Muscle Adaptations to HIT
P. M. Christensen et al.
intensity training of trained subjects. Venous lactate when
running at intensities below
_VO2-max was unchanged
after 4 weeks of repeated 30-sec sprint training in moder-
ately trained runners ( _VO2-max: 55 mL min1 kg1)
(Iaia et al. 2009) and after 6–9 weeks of combined aero-
bic high intensity training and repeated 30-sec sprinting
in
trained
runners
( _VO2-max:
61 mL
min1 kg1)
(Bangsbo et al. 2009), with both studies encompassing a
reduction
in
weekly
training
distance.
Changes
in
blood lactate likely reflect changes at the muscular level
(Green et al. 2009) but neither of the studies evaluated
changes in muscle lactate and creatine phosphate during
exercise. Following 3 weeks with high-intensity training
(8 9 5 min) added to the normal training, cyclists ( _VO2-
max: 65 mL min1 kg1) had lower muscle lactate but
unchanged
creatine
phosphate
content
directly
after
intense exercise (Clark et al. 2004). Therefore, it is
presently unknown if high intensity training encompass-
ing
repeated 30-sec
sprint
intervals,
in
combination
with reduction in volume, changes muscle anaerobic
energy turnover during submaximal exercise in trained
subjects.
Thus, the aims of the present study, using a training
intervention consisting of repeated 30-sec sprinting and
aerobic high intensity training for a group of trained
cyclists, were to examine (1) whether intensified training
increases the amount of oxidative enzymes in FT muscle
fibers as well as speeds _VO2 kinetics and reduces the _VO2
slow component during intense exercise; (2) if the anaer-
obic energy turnover during intense exercise was reduced
by training. We hypothesized that the intensified training
would increase the oxidative capacity of FT fibers result-
ing in faster _VO2 kinetics during intense exercise, but not
during moderate intensity exercise, as well as reduce the
_VO2 slow component and anaerobic energy turnover dur-
ing intense exercise.
Methods
Subjects
Eight trained male cyclists with an average (SD) age,
weight, and _VO2-max of 33 8 years, 81 8 kg, and
4.8 0.3 L min1 or 59 4 mL min1 kg1, respec-
tively, were recruited for the study. Prior to participating
in the study, the subjects had trained/competed ~three to
five times each week for at least 3 years. The study proce-
dures were approved by the local ethical committee of the
capital region of Copenhagen (Region Hovedstaden) and
all subjects received written and oral information about
the study procedures and gave their written informed
consent to participate in the study in accordance with the
Helsinki declaration.
Experimental design
The subjects carried out a 7-week high-intensity training
(HIT) intervention (see later) from October to December
just after the season had finished, hence subjects were
expected to be fit and in a physical stable condition. Both
before and after HIT the subjects carried out two main
experiments (EXP1 & EXP2) to evaluate changes in the met-
abolic response and _VO2 kinetics during submaximal exer-
cise (< _VO2-max). The subjects and training intervention
were the same as in a study focusing on adaptations of ion
transport proteins, ion kinetics, and performance during
repeated high intensity exercise (Gunnarsson et al. 2013).
Training
Subjects performed four supervised training sessions per
week on their own bikes on public roads. Training was
performed as 12 9 30-sec uphill (~6% gradient) maximal
sprints interspersed with 4–5 min low intensity recovery
(SPR; 2.59 week1) resulting in a 1:8–10 work rest ratio,
and 5 9 ~4 min aerobic high intensity intervals separated
by ~2 min of rest (AEH; 1.59 week1) on a flat 2.5 km
course with a work rest ratio of 2:1. In a training week
day 1 was recovery, AEH was performed on day 2, SPR on
day 3, recovery on day 4, SPR on day 5, recovery on day
6, and finally SPR or AEH on day 7 in alternate weeks. To
ensure maximal effort during training drafting was not
allowed and both SPR (1 vs. 1) and AEH (mass start) was
performed in a competitive manner with the objective of
finishing first. Heart rate (HR) was measured in 5-sec
intervals during training (Polar Team Edition, Finland).
Peak-HR during each SPR interval was 90 4% of HR-
max and average-HR during each AEH interval was
89 2% of HR-max. The physiological response from
one of the subjects during a SPR and an AEH training ses-
sion is shown in Fig. 1. Weekly volume was ~240 min
during HIT (15 min SET [~6%], 30 min AEH [~12%],
135 min low intensity recovery between intervals [56%]
and ~60 min moderate intensity [25%] as transport to
and from training). This amounted to a ~50% reduction
of the training volume being 472 153 min week1
before HIT (0–60% HR-max [29%], 60–70% HR-max
[24%], 70–80% HR-max [19%], 80–90% HR-max [21%],
90–95% HR-max [6%], 95–100% HR-max [1%]).
Exercise testing
All testing was performed on a mechanically braked
ergometer bike (Monark 839E, Varberg, Sweden) with the
subjects using their own pedals and specific geometric
setup which was maintained throughout the study. Sub-
jects were instructed not to perform any training the day
ª 2015 The Authors. Physiological Reports published by Wiley Periodicals, Inc. on behalf of
the American Physiological Society and The Physiological Society.
2015 | Vol. 3 | Iss. 7 | e12428
Page 3
P. M. Christensen et al.
Metabolic and Muscle Adaptations to HIT
before a testing session, and maintain the same food
intake and abstain from intake of caffeine on days of test-
ing.
During the first visit to the laboratory during the com-
petitive season subjects performed an incremental test
starting out at 100 W with increments of 25 W min1
until exhaustion. Pulmonary _VO2 was measured breath by
breath (Oxycon Pro, Viasys Healthcare, CareFusion, Rolle,
Switzerland) and VO2-max and HR-max was determined
as the highest value over a 30-sec period. In addition incre-
mental test peak power output was calculated as:
iPPO ¼ Power output (W) at the last completed stage
þ sec at the stage leading to task failure 60 sec1
25 W:
iPPO was 409 24 W. Before and after HIT changes in
pulmonary _VO2 kinetics was investigated using repeated
exercise transitions. Subjects performed both moderate
(MOD: 50% iPPO, 205 12 W) and intense (INT: 70%
iPPO, 286 17 W) exercise with 2 min at 20 W pre-
ceding all intervals. The absolute exercise intensity was
maintained throughout the study. Both prior to and fol-
lowing HIT subjects performed four transitions with
MOD each lasting 6 min (EXP1, EXP2, and two addi-
tional transitions on separate days) and three transitions
with INT each lasting 6 min (EXP1, EXP2, and an addi-
tional transition on a separate testing day) as well as two
transitions with INT lasting 3 min (EXP1, EXP2). Pul-
monary _VO2 was measured breath by breath. To deter-
mine _VO2 kinetics, errant breaths, defined as any value
lying more than 4 SDs away from the local mean (e.g.,
due to swallowing and coughing) were initially removed.
Then the _VO2 responses in each intensity domain were
linearly interpolated to give 1-sec values, and then aver-
aged. The initial cardiodynamic component was ignored
by eliminating the first 20 sec of data after the onset of
exercise.
MOD was modeled via a mono exponential function:
_VO2ðtÞ ¼ Baseline þ APð1 eðtTdP=sPÞÞ
INT was modeled via a bi-exponential function:
_VO2ðtÞ ¼Baseline þ APð1 eðtTdP=sPÞÞ
þ ASð1 eðtTdS=sSÞÞ
with _VO2 (t) being _VO2 to a given time (sec). _VO2 base-
line was calculated as average _VO2 from 30 to 90 sec of
0
4
8
12
16
20
1
2
3
4
5
0
4
8
12
16
20
1
2
3
4
5
6
7
8
9
10
11
12
Venous lactate
(mmol/L)
500
650
800
950
Power
(W)
300
325
350
375
400
0
1000
2000
3000
4000
5000
6000
6
12
18
24
30
Power
(W)
Time (min)
INTERVAL #
(12 x 30 sec; 5 min recovery)
INTERVAL #
(5 x 4 min; 2 min recovery)
VO2
(mL/min)
Venous lactate
(mmol/L)
A
B
C
D
Figure 1. The physiological response during a training session with 12 9 30-sec sprint intervals separated by 5 min of recovery (A & B) and a
session with 5 9 4 min intervals separated by 2 min of recovery (C & D) for one subject having a _VO2-max of 5.2 L min1 (hatched line). Peak
power (dotted bars), mean power (gray bars), and pulmonary _VO2 (full line) is shown for each interval (top) together with lactate (bottom)
from an antecubital vein before (open bars) and after (full bars) intervals. Each of the two training sessions was performed indoor using the
bike ergometer and _VO2-system described in the methods section.
2015 | Vol. 3 | Iss. 7 | e12428
Page 4
ª 2015 The Authors. Physiological Reports published by Wiley Periodicals, Inc. on behalf of
the American Physiological Society and The Physiological Society.
Metabolic and Muscle Adaptations to HIT
P. M. Christensen et al.
the 120 sec baseline cycling at 20 W. AP, TdP, and sP are
the amplitude, time delay, and time constant, respectively,
for the primary (P) response. AS, TdS, and sS are the
truncated amplitude, time delay, and time constant,
respectively, for the slow component (S). An iterative
process was used to determine the best fit of the curve.
The relative VO2 slow component was determined as the
ratio between AS and average VO2 during the last 2 min
of exercise.
Mechanical gross efficiency (GE) was calculated for
both MOD and INT during the last 2 min of exercise
using the formula
GE ¼ external bike load ðkJ min1Þ
energy turnover ðkJ min1Þ 100%
Energy
turnover
was
estimated
as
_VO2
(L
min1) 9 energetic value of oxygen (kJ L1) with the lat-
ter
being
calculated
from
the
measured
respiratory
exchange ratio (RER) thus taking into account the differ-
ent energy yield from oxidation of carbohydrate and fat.
In the event of RER exceeding 1.0 a value of 1.0 was used
in the calculation.
Main experiments
Two experimental days (EXP1 & EXP2) were performed
both prior to and after HIT to quantify muscle metabo-
lites during INT. Subjects arrived to the laboratory in the
morning after consuming a light breakfast.
On both EXP1 and EXP2 a biopsy at rest was collected
from m. vastus lateralis under local anesthesia using a
Bergstrom needle with suction. One part was frozen
immediately in liquid nitrogen within ~10 sec for analysis
of metabolites, enzyme activity, and protein content.
Another part of the biopsy was embedded in tissue tec
(Sakura Finetek, Netherlands) for histochemical analysis.
Initially, 6 min of MOD was performed followed by rest
for 30 min. To evaluate changes in muscle metabolism in
response to the HIT-period a muscle biopsy was obtained
following 6 min of INT on EXP1 and following 3 min of
INT on EXP2. Following 60 min of rest, 3 min of INT
was performed on EXP1 and following 30 min of rest,
6 min of INT was performed on EXP2.
Muscle analysis
All muscle samples were stored at 80°C until analyzed.
Muscle metabolites
The muscle biopsies taken at rest and after 3 and 6 min
of INT were analyzed for levels of lactate and creatine
phosphate (CP) using fluorometric methods (Lowry and
Passonneau 1972).
Maximal enzyme activity
In part of the muscle biopsy obtained at rest maximal
enzyme activity of citrate synthase (CS), 3-Hydroxyacyl
CoA dehydrogenase (HAD), phosphofructokinase (PFK),
and lactate dehydrogenase (LDH) was quantified in mus-
cle homogenates after freeze drying and removal of fat
and connective tissue using fluorometric methods (Flu-
oroscan Ascent, Thermo Scientific, Waltham, MA) (Lowry
and Passonneau 1972).
Protein expression in muscle homogenate lysates
Approximately 3 mg freeze dried muscle tissue was split
in two for double protein determination to increase mea-
suring sensitivity and then homogenized and centrifuged
to
exclude
non
dissolved
structures,
as
previously
described (Bangsbo et al. 2009). Total protein concentra-
tions were determined in each sample using BSA stan-
dards (Pierce, IL) and the lysates were then diluted in 69
Laemmli buffer and ddH2O to reach equal protein con-
centration before protein expression of CS, cytochrome c
oxidase complex 4 (COX-4) and PFK were determined by
western blotting. For subsequent analysis the average
value was calculated from the two samples.
Protein expression in segments of human single
muscle fibers
The determination of fiber type-specific changes in pro-
tein expression for CS, COX-4, and PFK were performed
as previously described Thomassen et al. (2013) with
minor changes. After freeze drying the muscle tissue sam-
ples (7–10 mg dry weight, n = 16) for 48 h, segments of
single fibers were dissected under a microscope and
stored in single microfuge tubes. The average size of the
segments collected were roughly determined by measuring
the
lengths
of
the
fiber
under
a
microscope
(1.4 0.3 mm, mean SD, n = 398). Before SDS-PAGE
18 lL 69 Laemmli buffer (0.7 mL 0.5 mol L1 Tris-base,
3 mL glycerol, 0.93 g DTT, 1 g SDS and 1.2 mg brom-
ophenol blue) diluted (1:3, v:v) in ddH2O was added to
each fiber and incubated for 1 h at room temperature.
In order to have equal number of fibers in the different
groups, all single fiber segments were first fiber typed
before the analysis of the protein of interest. About 5 lL
of the samples were loaded on a 26 well Tris-Tricine 4–
15% Criterion gels (Bio-Rad Laboratories, Solna, Sweden)
and by western blotting characterized as either slow
twitch (ST) or fast twitch (FT) muscle fibers by use of
ª 2015 The Authors. Physiological Reports published by Wiley Periodicals, Inc. on behalf of
the American Physiological Society and The Physiological Society.
2015 | Vol. 3 | Iss. 7 | e12428
Page 5
P. M. Christensen et al.
Metabolic and Muscle Adaptations to HIT
antibodies specific for myosin heavy chain (MHC) type I
(ST fibers) and MHC type II (FT fibers) as well as the
FT-specific SERCA1 protein (Thomassen et al. 2013).
Antibodies used were ST fibers: 0.5 lg∙mL1, mouse
monoclonal IgM, A4.840, Developmental Studies Hybrid-
oma Bank (DSHB), University of Iowa, USA; FT fibers:
2 lg∙mL1, mouse monoclonal IgG, A4.74, DSHB, both
developed by Dr Blau, and SERCA1: 0.1 lg∙mL1, mouse
monoclonal, MA3-912, Thermo Scientific. From this prefi-
ber typing, 160 fibers were selected (1.5 0.3 mm) to the
final analyses, including 20 from each subject with five
fibers from each of the four groups: ST pre, ST post, FT
pre, and FT post. The remaining 13 lL of the given selected
fiber segments were then loaded onto additional gels.
Western blotting
For both segments of single muscle fibers and muscle
homogenate lysates proteins were separated by SDS-PAGE
(55 mA per gel and maximum 150 V) for ~120 min and
then semi-dry transferred to a PVDF membrane (Milli-
pore A/S, Copenhagen, Denmark) for 120 min at 70 mA
per gel and maximum 25 V. After protein transfer, gels
including single fibers were incubated for 1 h in Coomas-
sie stain, including 0.3% Coomassie Brilliant Blue R
(Sigma-Aldrich, Copenhagen, Denmark), 40% ethanol
(96%), 10% Acetic acid (Merck, Copenhagen, Denmark)
and 49.7% ddH2O. Gels were then destained in ddH2O
overnight and imaged using a ChemiDoc MP Imaging
System (Bio-Rad Laboratories). These Coomassie-stained
posttransferred gels were used to determine the amount of
protein in each lane, based on the MHC (~200 kDa)
bands (Murphy et al. 2006). Membranes were blocked in
Tris-buffered saline including 0.1% Tween-20 (TBST) with
either 2% skimmed milk or 3% BSA for 1 h and then
incubated with primary antibodies over night. After 2
washes in TBST, horseradish peroxidase-conjugated sec-
ondary antibody (DAKO, Glostrup, Denmark) diluted 1
to 5000 in TBST with addition of either 2% skimmed milk
or 3% BSA was added, following which the membranes
were washed in TBST (3 9 15 min). Bands were visual-
ized using chemiluminescent detection (single fiber analy-
sis using Super Signal West Femto Maximum Sensitivity
Substrate, Thermo Scientific, – muscle lysates using ECL,
Millipore) and images were collected on a ChemiDoc MP
Imaging System. For further analyses, the membrane was
kept in TBST and re-incubated in a new primary antibody
overnight, giving the opportunity to determine the expres-
sion of several proteins with different molecular weights
on the same segment of fibers (Thomassen et al. 2013).
The membranes for single fiber analyses were divided
into four pieces by cutting over 250 kDa, right below the
150 kDa and 75 kDa, above the 25 kDa and below the
10 kDa markers (All Blue and Dual Color, Bio-Rad Labo-
ratories). The first and upper part was used to confirm the
predetermined fiber type (ST and FT 200 kDa), the second
part used for PFK (85 kDa) and SERCA1 (100 kDa), the
third for CS (48 kDa) and Actin (42 kDa), and the fourth
and lower part used for COX-4 (14 kDa).
Antibody details: Other antibodies used for protein
expression determination were: PFK: 0.2 lg∙mL1, mouse
monoclonal, Sc166722; Santa Cruz Biotechnology, Dallas,
TX;
CS:
0.33 lg∙mL1,
rabbit
polyclonal,
ab96600;
Abcam, Cambridge, UK; COX-4: 0.2 lg∙mL1, mouse
monoclonal, Sc58648; Santa Cruz Biotechnology.
Data treatment
In total, 160 segments of human skeletal muscle single
fibers from vastus lateralis were used for the final protein
expression determination in ST and FT fibers. On each
gel 5 ST and 5 FT fibers from a resting pre-HIT muscle
and 5 ST and 5 FT fibers from a resting post-HIT muscle
from the same individual were loaded. Given the small
size of segments of individual fibers, it was not possible
also to determine the total protein concentration in each
sample prior to sample loading. Consequently, different
amounts of protein were loaded in each well. In order to
compare the specific protein expression between fiber
samples, MHC on the post-transferred gel was quantified
and it was deemed that an equal proportion of total pro-
tein was transferred to the membrane independent of
total amount loaded. Thus, the signal on the posttrans-
ferred Coomassie stained gel was used for normalization
of the densities of the protein of interest, as previously
demonstrated as a reliable measure (Murphy et al. 2006;
Thomassen et al. 2013).
The signal intensity for each protein of interest was first
normalized to the mean intensity of all single human fiber
bands for that protein on the gel. Afterward data were
normalized to the total amount of protein in each sample
determined by Coomassie staining of the remaining MHC.
In order to compare fibers loaded on different gels single
values were normalized to the mean of ST pre-HIT in the
single fiber analysis and pre-HIT in the muscle homoge-
nate analysis. Finally, in order to have a normal distribu-
tion of the data, the ratios were log transformed before the
statistical analysis. For clarity the graphical presentation of
the results are displayed as ratios relative to ST pre-HIT
for single fiber data and from the backtransformed log val-
ues relative to pre-HIT for muscle homogenate.
Capillary density
Capillarization was analyzed using fluorescence micros-
copy. Transverse sections of the muscle biopsies were cut
2015 | Vol. 3 | Iss. 7 | e12428
Page 6
ª 2015 The Authors. Physiological Reports published by Wiley Periodicals, Inc. on behalf of
the American Physiological Society and The Physiological Society.
Metabolic and Muscle Adaptations to HIT
P. M. Christensen et al.
at a thickness of 8 lm and placed onto glass slides and
first treated with Biotinylated Ulex Europaeus Agglutinin
I (Vector Laboratories, Burlingame, CA) and later with
Streptavidin 1:200 (Dako, Glostrup, Denmark). Subse-
quently pictures were taken of the muscle samples and
analyzed on a computer (ImageJ) for capillary to fiber
ratio.
Statistics
Before and after HIT changes in _VO2 kinetics, capillary to
fiber ratio, muscle metabolites at rest and after 3 and
6 min of exercise as well as enzyme activity and protein
level in homogenates were examined with a Student’s
paired t-test. Also a paired t-test was used to evaluate if s
of the primary phase differed between MOD and INT
both pre- and post-HIT. Changes in segments of single
muscle fibers were evaluated using a two-way ANOVA
General Linear Model (ST vs. FT fibers and pre vs. post-
HIT as factors). The five ST fibers were averaged to yield
one value for each subject both pre- and post-HIT and
the same approach was made for the five FT fibers. If sig-
nificant main effects or an interaction were observed, then
a student Newman-Keuls post hoc analysis was performed
to identify the specific differences in protein expression
within fiber types. Correlations between _VO2 kinetics (s
of the primary response during MOD and INT and the
relative VO2 slow component) and maximal aerobic
enzyme activity and the capillary to fiber ratio was evalu-
ated using a one-tailed test with the a-priory hypothesis
that fast _VO2 kinetics and a minor _VO2 slow component
would be associated with a high aerobic enzyme activity
and capillary to fiber ratio.
Results
Muscle adaptations
After HIT no overall effect of the intervention was
observed for the amount of CS (P = 0.12; Fig. 2A), COX-
4 (P = 0.20; Fig. 2B) and PFK (P = 0.70; Fig. 2C) in seg-
ments of ST and FT fibers. An overall effect was found
for fiber type for CS (P = 0.01), COX-4 (P = 0.007) and
PFK (P < 0.001) since protein content in ST fibers was
higher than FT fibers for CS (P = 0.016 pre-HIT and
P = 0.007 post-HIT) and COX-4 (P = 0.012 pre-HIT and
P = 0.01 post-HIT) and lower for PFK (P < 0.001 pre
and post-HIT). Before HIT the content in ST fibers of CS
and COX-4 was on average 53% and 41% higher than in
FT fibers whereas PFK was ~389% higher in FT fibers
with respective values after HIT being 34%, 15% and
~241%. In muscle
homogenates
both
CS (P = 0.07;
Fig. 2A) and COX-4 (P = 0.10; Fig. 2B) tended to be
lower after compared to before HIT with an average
decrease in protein content of 16% and 11%, respectively.
PFK remained unchanged (P = 0.45; Fig. 2C) with an
average decrease of 3%. Representative Western blots of
the proteins investigated are displayed in Fig. 3.
After HIT the maximal enzyme activity was not changed
relative to before the intervention for CS (56 8 vs.
59 10 lmol g DW1 min1; P = 0.10), HAD (27 6
vs. 29 3 lmol g DW1 min1; P = 0.41), LDH (131
27 vs. 113 27 lmol g DW1 min1; P = 0.14) and PFK
(340 69 vs. 318 105 lmol g DW1 min1; P = 0.49).
Following HIT no changes relative to before the inter-
vention were observed in the capillary to fiber ratio
(2.30 0.16 vs. 2.38 0.20; P = 0.13) (Fig. 4A).
_VO2 kinetics
No significant differences following HIT were observed in
the _VO2 response during MOD (Fig. 5) and in all modeling
parameters (Table 1) as s was unchanged (P = 0.67)
together with absolute _VO2 (P = 0.39), RER (P = 0.63)
and GE (P = 0.39) averaged over the last 2 min of exercise.
The _VO2 response during INT was also unaffected by
HIT (Fig. 5) as were all modeling parameters (Table 1)
including s of the primary response (P = 0.32), the abso-
lute (P = 0.26) and relative (P = 0.25) size of the _VO2
slow component of the secondary response together with
absolute
_VO2
(P = 0.52),
RER
(P = 0.13)
and
GE
(P = 0.46) averaged over the last 2 min of exercise.
s during MOD and INT was not different neither pre
(P = 0.12) nor post-HIT (P = 0.22).
Correlations
Maximal activity of CS (r2 = 0.002–0.18; P > 0.05) and
HAD (r2 = 0.0001–0.16; P > 0.05) did not correlate with s
of the primary response during MOD and INT both before
and after HIT. Neither did CS correlate with the relative
size of the _VO2 slow component (r2 = 0.38 and 0.30 before
and after HIT; P > 0.05) as was the case for HAD
(r2 = 0.12 and 0.30 before and after HIT; P > 0.05). The
capillary to fiber ratio was associated with s during MOD
before (r2 = 0.90; P < 0.001) but not after HIT (r2 = 0.03;
P > 0.05) and no association was present during INT
(r2 = 0.10–0.14; P > 0.05). An association between the cap-
illary to fiber ratio and the relative size of the _VO2 slow
component (Fig. 4B) was present following HIT (r2 = 0.39;
P < 0.05) but not before HIT (r2 = 0.38; P > 0.05).
Muscle metabolites
Muscle CP (n = 7) was not changed by HIT at rest
(~90 mmol g DW1 min1; P = 0.60) and after 3 min of
ª 2015 The Authors. Physiological Reports published by Wiley Periodicals, Inc. on behalf of
the American Physiological Society and The Physiological Society.
2015 | Vol. 3 | Iss. 7 | e12428
Page 7
P. M. Christensen et al.
Metabolic and Muscle Adaptations to HIT
INT (~60 mmol g DW1 min1; P = 0.41), but after HIT
it was lower at 6 min of exercise than before HIT (42
15 vs. 66 17 mmol kg DW1; P = 0.011) (Fig. 6A).
Muscle lactate (n = 7) was not changed by HIT at rest
(~7 mmol g DW1 min1; P = 0.18) and after 3 min of
INT (~17 mmol g DW1 min1; P = 0.95), but at 6 min
muscle lactate was higher after compared to before HIT
(40.2 18.4 vs. 14.1 5.5 mmol kg DW1; P < 0.012)
(Fig. 6B).
Discussion
The major findings in the present study were that a per-
iod of reduced and intensified training performed by
trained cyclists did not elevate the protein content of oxi-
dative enzymes in FT fibers. Furthermore, no components
of pulmonary _VO2 kinetics were changed, whereas higher
muscle lactate accumulation and lower level of CP were
observed during intense cycling after compared to before
the intervention period.
Contrary to our hypothesis the content of CS and
COX-4 in the FT fibers was not elevated after the inter-
vention period in the form of volume reduced and inten-
sified training. The high exercise intensity used in HIT, in
particular the repeated sprint training, was chosen in
order to activate all FT fibers with this type of exercise
being a potent stimulator of the signal cascades leading to
adaptations of the muscular oxidative system (Psilander
0
0,5
1
1,5
2
1.5
0
2
4
6
8
10
0,5
1
1,5
0
0,5
1
HOM
COX-4 signal intensity
(relative to PRE)
0
1
1,5
PFK signal intensity
(relative to PRE)
0
0,5
1
CS signal intensity
(relative to PRE)
1.5
1.0
0.5
0
1.0
0.5
0
1.0
0.5
0
1.5
A
B
C
#
CS signal intensity
(relative to ST PRE)
COX-4 signal intensity
(relative to ST PRE)
0.5
0
1.0
PFK signal intensity
(relative to ST PRE)
ST PRE ST POST FT PRE FT POST
HOM
# #
# # #
0.5
2.7
0
1.0
0
100
200
300
400
500
PRE
POST
PFK maximal activity
(µmol/g DW/min)
(µmol/g DW/min)
0
20
40
60
80
CS maximal activity
D
E
F
H
G
Figure 2. Enzyme content following 7 weeks of high intensity and reduced volume training in trained cyclists (n = 8). The average content
(thick lines) of CS (A), COX-4 (B), and PFK (C) before (PRE) and after (POST) the intervention are shown in slow-twitch (ST) and fast-twitch (FT)
fibers together with individual values (average of five fibers SEM at each time point) which have been normalized to ST PRE for all enzymes.
Average and individual protein content measured in homogenates (HOM) are also shown for CS (D), COX-4 (E), and PFK (F) as well as maximal
activity for CS (G) and PFK (H). For clarity the graphs A-F display ratio data but the statistical analysis was based on log data. #P < 0.05,
##P < 0.01, ###P < 0.001; significant difference between ST and FT fibers.
2015 | Vol. 3 | Iss. 7 | e12428
Page 8
ª 2015 The Authors. Physiological Reports published by Wiley Periodicals, Inc. on behalf of
the American Physiological Society and The Physiological Society.
Metabolic and Muscle Adaptations to HIT
P. M. Christensen et al.
et al. 2010; Little et al. 2011). Thus, it was unexpected
that no change in oxidative enzyme content was present
in FT fibers after the training period. One explanation
may be that the cyclists during their normal training and
competition prior to the intervention period, routinely
engaged most of their muscle fibers, and that “extra” FT
fibers activated during the training in the intervention
period were only fibers from the highest order motor
units. Then, the FT fibers in “lower” order motor units
were recruited less due to the reduction in training vol-
ume from ~470 to ~240 min week1. In support of this,
it has been shown that, even at 65% (Scribbans et al.
2014) and 80% (Krustrup et al. 2004b) of _VO2-max, a
significant number of FT fibers are recruited. In the pres-
ent study a considerable part of the training prior to
HIT was performed near 80% _VO2-max (~28% of total
0
5
10
15
20
25
30
Relative VO2 slow component
(% of end exercise VO2)
Capillary to fibre ratio
PRE
POST
Capillary to fibre ratio
1.9
2.1
2.3
2.5
2.7
1.9
2.1
2.3
2.5
2.7
A
B
Figure 4. Muscle capillary to fiber ratio (A) before (PRE; open bars) and after (POST; closed bars) 7 weeks of high intensity and reduced
volume training in trained cyclists (n = 8) with insert picture showing staining of capillaries in a representative subject. Values are means SD.
Association between the capillary to fiber ratio and the relative _VO2 slow component (B) before (open symbols, r2 = 0.38; P > 0.05) and after
(closed symbols, r2 = 0.39; P < 0.05) 7 weeks of high intensity and reduced volume training in trained cyclists (n = 8).
MHCI
200 kDa
PRE
POST
ST
ST
FT
FT
200 kDa
MHC Coomassie
CS
200 kDa
MHCII
SERCA1
PFK
COX-4
100 kDa
85 kDa
48 kDa
14 kDa
14 kDa
85 kDa
48 kDa
CS
PFK
COX-4
PRE
POST
HOM
Figure 3. Representative western blots of the proteins investigated in slow-twitch (ST), fast-twitch (FT), and muscle homogenate (HOM) before
(PRE) and after (POST) 7 weeks of high intensity and reduced volume training in trained cyclists. See methods section for details.
ª 2015 The Authors. Physiological Reports published by Wiley Periodicals, Inc. on behalf of
the American Physiological Society and The Physiological Society.
2015 | Vol. 3 | Iss. 7 | e12428
Page 9
P. M. Christensen et al.
Metabolic and Muscle Adaptations to HIT
training time or ~130 min week1 was carried out with a
heart rate above 80% of HR-max). Furthermore, detrain-
ing in trained athletes is known to reduce content and
maximal activity of oxidative enzymes (Chi et al. 1983;
Christensen et al. 2011). Thus, it may be that the protein
content of oxidative enzymes was reduced in the motor
units controlling FT fibers lower in the fiber hierarchy
due to the reduction in training. This may also explain
the overall tendency for a drop in protein content and
maximal activity of CS and COX-4 measured in muscle
homogenates.
The single fiber data of CS and COX-4 showed a large
variation in oxidative enzyme protein content within each
fiber type as evidenced by the high standard error for
many of the cyclists (Fig. 2A and B). This is in agreement
with the pioneering work by Lowry et al. (1978) showing
a large range in maximal enzyme activity in single fibers,
but in that study a separation between ST and FT fibers
was not made. The observed difference in the content of
muscle oxidative enzymes between individual fibers sup-
ports the proposed concept that motor units have a range
from very fast to very slow _VO2 kinetics as well as differ-
ent steady-state values (Koppo et al. 2004). Studies using
trained subjects (Jansson and Kaijser 1977; Chi et al.
1983) and a longitudinal study on untrained subjects
(Henriksson and Reitman 1976) have shown that the
maximal activity of oxidative enzymes in FT fibers can be
as high as in ST fibers. This was not the case in the pres-
ent study, which may be due to a higher aerobic training
status ( _VO2-max: ~70 mL min1 kg1) of the subjects in
the studies showing similar levels in FT and ST fibers
(Jansson and Kaijser 1977; Chi et al. 1983). Nevertheless,
the
combination
of
repeated
30-sec
sprinting
(2.5 9 week1)
and
aerobic
high
intensity
training
(1.5 9 week1) in the present study with a reduced train-
ing volume did not increase the oxidative enzyme content
in FT fibers in already trained athletes with a _VO2-max of
~60 mL min1 kg1. On the other hand, it has been
Table 1. Pulmonary
_VO2 kinetics modeling parameters during
moderate (MOD) and intense (INT) cycling before (PRE) and after
(POST) 7 weeks of high intensity and reduced volume training in
trained cyclists (n = 8). Changes in pre and post were evaluated
with a paired t-test.
PRE
POST
MOD
Baseline (mL min1)
865 96
849 60
TdP (sec)
17.5 3.1
18.3 2.9
sP (sec)
15.9 2.4
15.4 2.5
AP (mL min1)
2209 119
2249 144
_VO2 4–6 min (mL min1)
3083 172
3110 165
RER 4–6 min
0.95 0.02
0.95 0.01
GE 4–6 min (%)
19.1 0.7
18.9 0.7
Cadence (rounds min1)
92 9
92 8
INT
Baseline (mL min1)
870 65
824 53
TdP (sec)
15.2 1.5
15.3 2.3
sP (sec)
18.2 2.5
17.2 2.9
AP (mL min1)
2931 240
2990 186
TdS (sec)
103 30
95 56
sS (sec)
141 60
181 111
AS (mL min1)
479 264
539 257
_VO2 4–6 min (mL min1)
4153 234
4180 184
RER 4–6 min
1.03 0.04
1.01 0.02
GE 4–6 min (%)
19.6 0.9
19.5 0.7
Cadence (rounds min1)
93 6
94 6
Values are means SD. Baseline ( _VO2 before onset of exercise).
Time delay (Td), time constant (s), amplitude (A) for the primary
response (P), and the slow component (S). Absolute oxygen uptake
( _VO2), respiratory exchange ratio (RER), and gross efficiency (GE)
in the last 2 min of exercise from 4–6 min.
0
1500
3000
4500
–120 –60
0
60
120 180 240 300 360
Time (sec)
0
1500
3000
4500
–120 –60
0
60
120 180 240 300 360
VO2
(mL/min)
Time (sec)
A
B
0
1500
3000
4500
–120 –60
0
60
120 180 240 300 360
Time (sec)
C
Figure 5. Pulmonary oxygen uptake ( _VO2) following 7 weeks of high intensity and reduced volume training in trained cyclists (n = 8) during
moderate (MOD; A) and intense (INT; B) cycling with the modeled responses shown (C) before (PRE; open symbols and hatched lines) and after
(POST; filled symbols and solid lines) the intervention. The arrow in panel C indicates the onset of the _VO2 slow component during INT (103
and 95 sec on average PRE and POST HIT) being superimposed on the primary _VO2 response.
2015 | Vol. 3 | Iss. 7 | e12428
Page 10
ª 2015 The Authors. Physiological Reports published by Wiley Periodicals, Inc. on behalf of
the American Physiological Society and The Physiological Society.
Metabolic and Muscle Adaptations to HIT
P. M. Christensen et al.
shown that adding two weekly aerobic high intensity ses-
sions (6 9 ~3 min with ~1.5 min recovery), at the speed
eliciting task failure in an incremental test, increased the
content of LDH in the FT fibers after 6 weeks of training
with a maintained training volume in well-trained run-
ners ( _VO2-max: 67 mL min1 kg1) (Kohn et al. 2011).
Therefore, it seems plausible that the training reduction
in the present study can explain the lack of increase in
oxidative protein content in FT fibers. Alternatively, train-
ing with repeated sprints may be less potent than aerobic
high intensity training (3–5 min intervals at ~90–100%
_VO2-max)
to
increase
protein
content
of
oxidative
enzymes in FT fibers in trained individuals, since the for-
mer type of training dominated in the present study.
Based on changes in blood and muscle lactate in
trained individuals it does appear that different types of
high intensity training can yield different outcomes.
Accordingly, aerobic high intensity training (< _VO2-max;
e.g., 3–5 min intervals) and a maintained training volume
appears to lower muscle (Clark et al. 2004) and blood
lactate (Acevedo and Goldfarb 1989; Kohn et al. 2011)
during intense exercise (< _VO2-max), whereas training
interventions with repeated 30-sec sprints in combination
with a volume reduction does not lower blood lactate
(Bangsbo et al. 2009; Iaia et al. 2009) and in the present
study muscle lactate was higher after training. Future
studies are needed to evaluate if the outcome on oxidative
adaptations in FT fibers differs between different types of
high intensity training which appears to be the case with
regards to lactate levels. The role of the total training vol-
ume is considered of interest, thus either adding SET on
top of the normal training or having SET substitute less
intense training seems relevant.
Pulmonary _VO2 kinetics was not changed with HIT
(Fig. 5 and Table 1). This finding differs from the faster
_VO2 kinetics observed after a period of intense training
in untrained subjects at both the muscular (Krustrup
et al. 2004a) and pulmonary level (Bailey et al. 2009) with
the latter study also reporting a reduced _VO2 slow com-
ponent. Despite the tendencies for a reduction in the con-
tent of CS and COX-4 as well as the maximal activity of
CS
in
the
present
study,
_VO2
kinetics
remained
unchanged. These findings suggest that the level of oxida-
tive enzymes does not limit the muscle oxygen utilization
in the initial part of exercise, nor the development of the
_VO2 slow component. It should, however, be considered
that the content and maximal activity of the oxidative
enzymes was not significantly reduced, and it may be that
endurance athletes have an excessive oxidative enzyme
capacity allowing a modest drop without having an effect
on _VO2 kinetics. The CS activity measured in muscle
homogenate was ~60 lmol g DW1 min1, which is
about twice as high as in untrained subjects (Krustrup
et al. 2004a; Burgomaster et al. 2005, 2008) and some-
what higher than previous observations in trained endur-
ance athletes with a similar _VO2-max as in the present
study (Yeo et al. 2008; Bangsbo et al. 2009). In a study of
trained soccer players 2 weeks without training resulted
in a decrease in activity and content of oxidative enzymes,
which was associated with slower _VO2 kinetics (Christen-
sen et al. 2011). Thus, it cannot be excluded that the level
of oxidative enzymes under some circumstances may
become limiting for _VO2 kinetics in trained individuals.
Unlike previous findings in trained individuals (Koppo
et al. 2004) the speed of the primary _VO2 response was
not significantly different between MOD (s~15.7 sec) and
INT (s~17.7 sec) although five of eight subjects had larger
s in INT than in MOD both pre- and post-HIT. Thus,
despite an expected larger recruitment of FT fibers in INT
relative to MOD (Krustrup et al. 2004b) and the fact that
FT
fibers
had
markedly
lower
content
of
oxidative
enzymes than ST fibers in the present study (Fig. 2A and
B), the _VO2 kinetics of the primary response in INT was
not slower than MOD. This in turn suggests that at least
in ST fibers there is an excess capacity of oxidative
enzymes that does not impact on the speed of the VO2
response. Maximal oxidative enzyme activity of CS and
HAD were poor predictors of fast _VO2 kinetics and the
relative size of the _VO2 slow component showing that in
trained individuals these enzymes appear to be of minor
importance and other muscular variables needs to investi-
gated.
Of
interest
was
the
association
between
the
0
10
20
30
40
50
60
70
Muscle lactate
(mmol/kg DW)
0
20
40
60
80
100
120
Creatine phosphate
(mmol/kg DW)
A
B
*
*
PRE POST
Rest
PRE POST
3 min
PRE POST
6 min
Figure 6. Muscle creatine phosphate (A) and lactate (B) at rest and
after 3 and 6 min of intense cycling before (PRE; open bars) and
after (POST; closed bars) 7 weeks of high intensity and reduced
volume training in trained cyclists (n = 7). Values are meansSD.
*P < 0.05; significant difference between PRE- and POST-values.
ª 2015 The Authors. Physiological Reports published by Wiley Periodicals, Inc. on behalf of
the American Physiological Society and The Physiological Society.
2015 | Vol. 3 | Iss. 7 | e12428
Page 11
P. M. Christensen et al.
Metabolic and Muscle Adaptations to HIT
capillary to fiber ratio and the relative size of the _VO2
slow component after the HIT intervention and the
marked tendency for an association before HIT (Fig. 4).
However, the association was mainly due to one subject
having a low capillary to fiber ratio. Nevertheless, during
intense exercise eliciting a slow component – likely due to
recruitment of FT fibers (Jones et al. 2011) – a high capil-
lary to fiber ratio could be speculated to lower blood
mean transit time optimizing conditions for diffusion of
oxygen which is supported by the finding of a reduced
_VO2 slow component during exercise inhaling an hyper-
oxic gas (Wilkerson et al. 2006).
The decrease in muscle CP and the increase in mus-
cle lactate from 3 to 6 min in INT (~85% _VO2-max)
were larger after the HIT period (Fig. 6). These findings
suggest a larger anaerobic energy turnover in the last
phase of INT. The activity of PFK was not elevated
with the intervention period, neither when expressed as
maximal activity or content, so it cannot explain the
apparently greater rate of glycolysis (Spriet et al. 2000).
The larger drop in CP indicates a higher accumulation
of muscle ADP, which may have elevated the rate of
glycolysis leading to a greater lactate production, but a
larger ADP concentration would also be expected to
stimulate the respiration but that was not the case in
light of the unchanged _VO2 response. Alternatively, the
tendency for a lower content and maximal activity of
muscle oxidative enzymes (CS and COX-4) after the
HIT period may have reduced the mitochondrial utili-
zation of the produced pyruvate (Brooks 2000), thereby
enhancing
the
lactate
production
catalyzed
by LDH
(Spriet et al. 2000). It is, however, unclear, why such
changes did not occur in the first phase (0–3 min) of
INT (Fig. 6B). Nevertheless, it appears that the anaero-
bic energy flux during the last phase (3–6 min) of INT
was higher after the intervention period, and thus, total
energy
turnover,
as
pulmonary
VO2
was
unaltered.
Alternatively, this finding may reflect a greater imbal-
ance between muscle CP and creatine and a change in
the ratio between muscle lactate and pyruvate during
exercise due to altered regulation. In moderately trained
subjects (VO2-max: 49 mL min1 kg1) repeated sprint
training, as used in the present study, resulted in lower
muscle lactate and ATP concentrations after exercise at
an intensity corresponding to 90% of _VO2-max. How-
ever, there were major differences between the present
study and the one by Burgomaster and co-workers
including a lower training status, and an increase in
training volume and a higher maximal aerobic enzyme
activity after the training period (Burgomaster et al.
2006) which may in part explain the different change
in muscle lactate accumulation during intense exercise.
The findings in the present study are in contrast to
observations in a study also using well-trained cyclists
( _VO2-max: ~65 mL O2 min1 kg1) who for a 3-week
period added aerobic high intensity training (8 9 5 min
~85%
_VO2-max) three times weekly to their normal
training volume. Following the training period muscle
lactate
accumulation
during
intense
exercise
(~85%
_VO2-max)
was
reduced
and
during
more
moderate
exercise (65% _VO2-max) fat and carbohydrate oxidation
was larger and lower, respectively (Clark et al. 2004).
Content
and
maximal
activity
of
oxidative
enzymes
were
not
reported,
but
was
unchanged
in
trained
cyclists in another study using the same type of training
(Yeo et al. 2008). Taken together these findings suggest
that the reduced amount of training in the present
study is the major cause of the elevated anaerobic
energy production during the intense submaximal work.
Furthermore, the present study shows that anaerobic
metabolism can be altered without a change in
_VO2
kinetics. Such dissociation between anaerobic and aero-
bic metabolism has also been reported in the exercise
transient in studies using hyperoxia where CP utiliza-
tion has been observed to be reduced (Vanhatalo et al.
2010) and the primary
_VO2 response appears to be
unaffected
(Wilkerson
et al.
2006).
Likewise,
during
steady-state conditions with moderate exercise higher
CP and lower muscle lactate have been observed in hy-
peroxia
(Stellingwerff
et al.
2006)
despite
the
_VO2
response being similar in both the exercise transient
and in the stable phase of exercise (Wilkerson et al.
2006). Further studies are needed to evaluate how dif-
ferent training regimes influence anaerobic energy turn-
over during submaximal exercise.
The functional significance of the training period in the
present study has been reported previously with regard to
high intensity
exercise
performance
in
the
form
of
improved
performance
in
a
repeated
sprint
test
(6 9 20 sec) and in an exhaustive test lasting ~4 min
with the latter test being preceded by a 2-min preload
with high intensity in which muscle lactate at the end of
the preload also was elevated after HIT (Gunnarsson et al.
2013). This indicates that the apparently larger anaerobic
muscle perturbation with HIT does not lower perfor-
mance during intense exercise.
Each subject had five ST and FT fibers analyzed both
pre and post-HIT (total of 20 fibers) and the average
value of the five fibers for each time point (pre- and
post-HIT) was used for further analysis. Such a low num-
ber may seem limiting, but a statistical difference was
present between ST and FT fibers for all enzymes investi-
gated in line with previous reports using pooled groups
of ST and FT fibers to determine maximal enzyme activ-
ity (Essen et al. 1975; Essen-Gustavsson and Henriksson
1984; Schantz and Henriksson 1987) or content (Thomas-
2015 | Vol. 3 | Iss. 7 | e12428
Page 12
ª 2015 The Authors. Physiological Reports published by Wiley Periodicals, Inc. on behalf of
the American Physiological Society and The Physiological Society.
Metabolic and Muscle Adaptations to HIT
P. M. Christensen et al.
sen et al. 2013). Furthermore, the tendencies for a drop
in homogenate protein content and maximal activity for
CS and COX-4 and unchanged PFK levels were mirrored
by the single fiber data. In addition the same three sub-
jects who had a higher CS content in FT fibers after HIT
also had elevated content of COX-4 (Symbols □, ■ and
▲on Fig. 2). Taken together this suggests that the method
is sensitive enough to detect possible changes in response
to a training intervention.
In
summary,
the
present
study
showed
that
the
amount of CS and COX-4 in FT fibers and muscle
homogenate as well as maximal activity of CS was not
changed after a 7-week period of intensified training with
a reduced volume in already trained cyclists. Further-
more,
no
change
in
pulmonary
_VO2
kinetics
was
observed during moderate (~65% _VO2-max) and intense
(~85% _VO2-max) submaximal exercise. During intense
cycling muscle CP levels were reduced and muscle lactate
accumulation was elevated to a greater extent in the later
part of exercise (3–6 min) following the training inter-
vention without an altered
_VO2. This could be inter-
preted to reflect a larger total energy turnover following
the training intervention due to a greater anaerobic
energy contribution to exercise or an altered regulation
of anaerobic muscle metabolism.
Acknowledgments
Lars Ropke, Rasmus Vilsgaard, Lars Nybo, Nikolai Nords-
borg for assistance during the experiments, Rafael Casuso
and Karina Olsen for assistance with muscle analysis and
the subjects for their great effort and determination dur-
ing training and testing.
Conflict of Interest
No conflict of interest is reported.
References
Acevedo, E. O., and A. H. Goldfarb. 1989. Increased training
intensity effects on plasma lactate, ventilatory threshold, and
endurance. Med. Sci. Sports Exerc. 21:563–568.
Bailey, S. J., D. P. Wilkerson, F. J. Dimenna, and A. M. Jones.
2009. Influence of repeated sprint training on pulmonary
O2 uptake and muscle deoxygenation kinetics in humans. J.
Appl. Physiol. 106:1875–1887.
Bangsbo, J., P. Krustrup, J. Gonzalez-Alonso, and B. Saltin.
2001. ATP production and efficiency of human skeletal
muscle during intense exercise: effect of previous exercise.
Am. J. Physiol. Endocrinol. Metab. 280:E956–E964.
Bangsbo, J., T. P. Gunnarsson, J. Wendell, L. Nybo, and M.
Thomassen. 2009. Reduced volume and increased training
intensity elevate muscle Na+-K+ pump alpha2-subunit
expression as well as short- and long-term work capacity in
humans. J. Appl. Physiol. 107:1771–1780.
Barstow, T. J., A. M. Jones, P. H. Nguyen, and R. Casaburi.
1996. Influence of muscle fiber type and pedal frequency on
oxygen uptake kinetics of heavy exercise. J. Appl. Physiol.
81:1642–1650.
Brooks, G. A. 2000. Intra- and extra-cellular lactate shuttles.
Med. Sci. Sports Exerc. 32:790–799.
Burgomaster, K. A., S. C. Hughes, G. J. Heigenhauser, S. N.
Bradwell, and M. J. Gibala. 2005. Six sessions of sprint
interval training increases muscle oxidative potential and
cycle endurance capacity in humans. J. Appl. Physiol.
98:1985–1990.
Burgomaster, K. A., G. J. Heigenhauser, and M. J. Gibala.
2006. Effect of short-term sprint interval training on human
skeletal muscle carbohydrate metabolism during exercise
and time-trial performance. J. Appl. Physiol. 100:2041–2047.
Burgomaster, K. A., K. R. Howarth, S. M. Phillips, M.
Rakobowchuk, M. J. MacDonald, S. L. McGee, et al. 2008.
Similar metabolic adaptations during exercise after low
volume sprint interval and traditional endurance training in
humans. J. Physiol. 586:151–160.
Chi, M. M., C. S. Hintz, E. F. Coyle, W. H. III Martin, J. L.
Ivy, P. M. Nemeth, et al. 1983. Effects of detraining on
enzymes of energy metabolism in individual human muscle
fibers. Am. J. Physiol. 244:C276–C287.
Christensen, P. M., P. Krustrup, T. P. Gunnarsson, K.
Kiilerich, L. Nybo, and J. Bangsbo. 2011. VO2 kinetics and
performance in soccer players after intense training and
inactivity. Med. Sci. Sports Exerc. 43:1716–1724.
Clark, S. A., Z. P. Chen, K. T. Murphy, R. J. Aughey, M. J.
McKenna, B. E. Kemp, et al. 2004. Intensified exercise
training does not alter AMPK signaling in human skeletal
muscle. Am. J. Physiol. Endocrinol. Metab. 286:E737–E743.
Demarle, A. P., J. J. Slawinski, L. P. Laffite, V. G. Bocquet, J.
P. Koralsztein, and V. L. Billat. 2001. Decrease of O(2)
deficit is a potential factor in increased time to exhaustion
after specific endurance training. J. Appl. Physiol. 90:
947–953.
Dufour, S. P., E. Ponsot, J. Zoll, S. Doutreleau, E. Lonsdorfer-
Wolf, B. Geny, et al. 2006. Exercise training in normobaric
hypoxia in endurance runners. I. Improvement in aerobic
performance capacity. J. Appl. Physiol. 100:1238–1248.
Egan, B., B. P. Carson, P. M. Garcia-Roves, A. V. Chibalin, F.
M. Sarsfield, N. Barron, et al. 2010. Exercise intensity-
dependent regulation of peroxisome proliferator-activated
receptor coactivator-1 mRNA abundance is associated with
differential activation of upstream signalling kinases in
human skeletal muscle. J. Physiol. 588:1779–1790.
Essen, B., E. Jansson, J. Henriksson, A. W. Taylor, and B.
Saltin. 1975. Metabolic characteristics of fibre types in
human skeletal muscle. Acta Physiol. Scand. 95:153–165.
ª 2015 The Authors. Physiological Reports published by Wiley Periodicals, Inc. on behalf of
the American Physiological Society and The Physiological Society.
2015 | Vol. 3 | Iss. 7 | e12428
Page 13
P. M. Christensen et al.
Metabolic and Muscle Adaptations to HIT
Essen-Gustavsson, B., and J. Henriksson. 1984. Enzyme levels
in pools of microdissected human muscle fibres of identified
type. Adaptive response to exercise. Acta Physiol. Scand.
120:505–515.
Gibala, M. J., J. P. Little, E. M. van, G. P. Wilkin, K. A.
Burgomaster, A. Safdar, et al. 2006. Short-term sprint
interval versus traditional endurance training: similar initial
adaptations in human skeletal muscle and exercise
performance. J. Physiol. 575:901–911.
Green, H. J., E. Bombardier, M. E. Burnett, I. C. Smith, S. M.
Tupling, and D. A. Ranney. 2009. Time-dependent effects of
short-term training on muscle metabolism during the early
phase of exercise. Am. J. Physiol. Regul. Integr. Comp.
Physiol. 297:R1383–R1391.
Gunnarsson, T. P., P. M. Christensen, K. Holse, D.
Christiansen, and J. Bangsbo. 2012. Effect of additional
speed endurance training on performance and muscle
adaptations. Med. Sci. Sports Exerc. 44:1942–1948.
Gunnarsson, T. P., P. M. Christensen, M. Thomassen, L. R.
Nielsen, and J. Bangsbo. 2013. Effect of intensified training
on muscle ion kinetics, fatigue development and repeated
short term performance in endurance trained cyclists. Am. J.
Physiol. Regul. Integr. Comp. Physiol. 305:R811–R821.
Henriksson, J., and J. S. Reitman. 1976. Quantitative measures
of enzyme activities in type I and type II muscle fibres of
man after training. Acta Physiol. Scand. 97:392–397.
Hickson, R. C., H. A. Bomze, and J. O. Hollozy. 1978. Faster
adjustment of O2 uptake to the energy requirement of
exercise in the trained state. J. Appl. Physiol. 44:877–881.
Iaia, F. M., Y. Hellsten, J. J. Nielsen, M. Fernstrom, K. Sahlin,
and J. Bangsbo. 2009. Four weeks of speed endurance
training reduces energy expenditure during exercise and
maintains muscle oxidative capacity despite a reduction in
training volume. J. Appl. Physiol. 106:73–80.
Jacobs, I., M. Esbjornsson, C. Sylven, I. Holm, and E. Jansson.
1987. Sprint training effects on muscle myoglobin, enzymes,
fiber types, and blood lactate. Med. Sci. Sports Exerc.
19:368–374.
Jansson, E., and L. Kaijser. 1977. Muscle adaptation to extreme
endurance training in man. Acta Physiol. Scand. 100:315–324.
Jones, A. M., and D. C. Poole. 2005. Oxygen uptake dynamics:
from muscle to mouth–an introduction to the symposium.
Med. Sci. Sports Exerc. 37:1542–1550.
Jones, A. M., B. Grassi, P. M. Christensen, P. Krustrup, J.
Bangsbo, and D. C. Poole. 2011. Slow component of VO2
kinetics: mechanistic bases and practical applications. Med.
Sci. Sports Exerc. 43:2046–2062.
Kohn, T. A., B. Essen-Gustavsson, and K. H. Myburgh. 2011.
Specific muscle adaptations in type II fibers after high-
intensity interval training of well-trained runners. Scand. J.
Med. Sci. Sports 21:765–772.
Koppo, K., J. Bouckaert, and A. M. Jones. 2004. Effects of
training status and exercise intensity on phase II VO2
kinetics. Med. Sci. Sports Exerc. 36:225–232.
Krustrup, P., J. Gonzalez-Alonso, B. Quistorff, and J. Bangsbo.
2001. Muscle heat production and anaerobic energy
turnover during repeated intense dynamic exercise in
humans. J. Physiol. 536:947–956.
Krustrup, P., Y. Hellsten, and J. Bangsbo. 2004a. Intense
interval training enhances human skeletal muscle oxygen
uptake in the initial phase of dynamic exercise at high but
not at low intensities. J. Physiol. 559:335–345.
Krustrup, P., K. Soderlund, M. Mohr, and J. Bangsbo. 2004b.
The slow component of oxygen uptake during intense, sub-
maximal exercise in man is associated with additional fibre
recruitment. Pflugers Arch. 447:855–866.
Little, J. P., A. Safdar, N. Cermak, M. A. Tarnopolsky, and M.
J. Gibala. 2010. Acute endurance exercise increases the
nuclear abundance of PGC-1alpha in trained human skeletal
muscle. Am. J. Physiol. Regul. Integr. Comp. Physiol. 298:
R912–R917.
Little, J. P., A. Safdar, D. Bishop, M. A. Tarnopolsky, and M.
J. Gibala. 2011. An acute bout of high-intensity interval
training increases the nuclear abundance of PGC-1alpha and
activates mitochondrial biogenesis in human skeletal muscle.
Am. J. Physiol. Regul. Integr. Comp. Physiol. 300:
R1303–R1310.
Lowry, O. H., and J. V. Passonneau. 1972. Pp. 237–249. A
flexible system of enzymatic analysis. Academic, New York.
Lowry, C. V., J. S. Kimmey, S. Felder, M. M. Chi, K. K.
Kaiser, P. N. Passonneau, et al. 1978. Enzyme patterns in
single human muscle fibers. J. Biol. Chem. 8269–8277.
MacDougall, J. D., A. L. Hicks, J. R. MacDonald, R. S.
McKelvie, H. J. Green, and K. M. Smith. 1998. Muscle
performance and enzymatic adaptations to sprint interval
training. J. Appl. Physiol. 84:2138–2142.
Murphy, R. M., E. Verburg, and G. D. Lamb. 2006. Ca2+
activation of diffusible and bound pools of mu-calpain in
rat skeletal muscle. J. Physiol. 576:595–612.
Nordsborg, N. B., C. Lundby, L. Leick, and H. Pilegaard.
2010. Relative workload determines exercise-induced
increases in PGC-1alpha mRNA. Med. Sci. Sports Exerc.
42:1477–1484.
Norris, S. R., and S. R. Petersen. 1998. Effects of endurance
training on transient oxygen uptake responses in cyclists. J.
Sports Sci. 16:733–738.
Phillips, S. M., H. J. Green, M. J. MacDonald, and R. L.
Hughson. 1995. Progressive effect of endurance training on
VO2 kinetics at the onset of submaximal exercise. J. Appl.
Physiol. 79:1914–1920.
Psilander, N., L. Wang, J. Westergren, M. Tonkonogi, and K.
Sahlin. 2010. Mitochondrial gene expression in elite cyclists:
effects of high-intensity interval exercise. Eur. J. Appl.
Physiol. 110:597–606.
Saltin, B., K. Nazar, D. L. Costill, E. Stein, E. Jansson, B.
Essen, et al. 1976. The nature of the training response;
peripheral and central adaptations of one-legged exercise.
Acta Physiol. Scand. 96:289–305.
2015 | Vol. 3 | Iss. 7 | e12428
Page 14
ª 2015 The Authors. Physiological Reports published by Wiley Periodicals, Inc. on behalf of
the American Physiological Society and The Physiological Society.
Metabolic and Muscle Adaptations to HIT
P. M. Christensen et al.
Schantz, P. G., and J. Henriksson. 1987. Enzyme levels of the
NADH shuttle systems: measurements in isolated muscle
fibres from humans of differing physical activity. Acta
Physiol. Scand. 129:505–515.
Scribbans, T. D., B. A. Edgett, K. Vorobej, A. S. Mitchell, S. D.
Joanisse, J. B. Matusiak, et al. 2014. Fibre-specific responses
to endurance and low volume high intensity interval
training: striking similarities in acute and chronic
adaptation. PLoS ONE 9:e98119.
Shepherd, S. O., M. Cocks, K. D. Tipton, A. M. Ranasinghe,
T. A. Barker, J. G. Burniston, et al. 2013. Sprint interval and
traditional endurance training increase net intramuscular
triglyceride breakdown and expression of perilipin 2 and 5.
J. Physiol. 591(Pt 3):657–675.
Shoemaker, J. K., S. M. Phillips, H. J. Green, and R. L.
Hughson. 1996. Faster femoral artery blood velocity kinetics
at the onset of exercise following short-term training.
Cardiovasc. Res. 31:278–286.
Spriet, L. L., R. A. Howlett, and G. J. Heigenhauser. 2000. An
enzymatic approach to lactate production in human
skeletal muscle during exercise. Med. Sci. Sports Exerc.
32:756–763.
Stellingwerff, T., P. J. LeBlanc, M. G. Hollidge, G. J.
Heigenhauser, and L. L. Spriet. 2006. Hyperoxia decreases
muscle glycogenolysis, lactate production, and lactate efflux
during steady-state exercise. Am. J. Physiol. Endocrinol.
Metab. 290:E1180–E1190.
Thomassen, M., R. M. Murphy, and J. Bangsbo. 2013. Fibre
type-specific change in FXYD1 phosphorylation during
acute intense exercise in humans. J. Physiol. 591:1523–1533.
Vanhatalo, A., J. Fulford, F. J. Dimenna, and A. M. Jones.
2010. Influence of hyperoxia on muscle metabolic responses
and the power-duration relationship during severe-intensity
exercise in humans: a 31P magnetic resonance spectroscopy
study. Exp. Physiol. 95:528–540.
Wilkerson, D. P., N. J. Berger, and A. M. Jones. 2006.
Influence of hyperoxia on pulmonary O2 uptake kinetics
following the onset of exercise in humans. Respir. Physiol.
Neurobiol. 153:92–106.
Yeo, W. K., C. D. Paton, A. P. Garnham, L. M. Burke, A. L.
Carey, and J. A. Hawley. 2008. Skeletal muscle adaptation
and performance responses to once a day versus twice every
second day endurance training regimens. J. Appl. Physiol.
105:1462–1470.
ª 2015 The Authors. Physiological Reports published by Wiley Periodicals, Inc. on behalf of
the American Physiological Society and The Physiological Society.
2015 | Vol. 3 | Iss. 7 | e12428
Page 15
P. M. Christensen et al.
Metabolic and Muscle Adaptations to HIT
| Unchanged content of oxidative enzymes in fast-twitch muscle fibers and V˙O2 kinetics after intensified training in trained cyclists. | [] | Christensen, Peter M,Gunnarsson, Thomas P,Thomassen, Martin,Wilkerson, Daryl P,Nielsen, Jens Jung,Bangsbo, Jens | eng |
PMC8639140 | 1
https://doi.org/10.11606/s1518-8787.2021055003903
Original Article
Rev Saude Publica. 2021;55:97
Running away from the jab: factors
associated with COVID-19 vaccine
hesitancy in Brazil
Marco Antonio Catussi PaschoalottoI
, Eduardo Polena Pacheco Araújo CostaI
, Sara Valente
de AlmeidaI,II
, Joana CimaIII
, Joana Gomes da CostaIV
, João Vasco SantosV,VI,VII
,
Pedro Pita BarrosI
, Claudia Souza PassadorVIII
, João Luiz PassadorVIII
I Universidade NOVA de Lisboa. Nova School of Business and Economics. Carcavelos, Portugal
II Imperial College London. Faculty of Health Sciences. London, England
III Universidade do Minho. Núcleo de Investigação em Políticas Económicas e Empresariais. Braga, Portugal
IV Universidade do Porto. Faculdade de Economia e Gestão. Porto, Portugal
V Universidade do Porto. Faculdade de Medicina. MEDCIDS – Departamento Medicina da Comunidade,
Informação e Decisão em Saúde. Porto, Portugal
VI Universidade do Porto. Faculdade de Medicina. Centro de Investigação em Tecnologias e Serviços de Saúde.
Porto, Portugal
VII ARS Norte. ACES Grande Porto VIII - Espinho/Gaia. Unidade de Saúde Pública. Vila Nova de Gaia, Portugal
VIII Universidade de São Paulo. Faculdade de Economia, Administração e Contabilidade de Ribeirão Preto.
Ribeirão Preto, São Paulo, Brasil
ABSTRACT
OBJECTIVE: To investigate how sociodemographic conditions, political factors, organizational
confidence, and non-pharmaceutical interventions compliance affect the COVID-19 vaccine
hesitancy in Brazil.
METHODS: Data collection took place between November 25th, 2020 and January 11th, 2021
using a nationwide online survey. Subsequently, the researches performed a descriptive analysis
on the main variables and used logistic regression models to investigate the factors associated
with COVID-19 vaccine hesitancy.
RESULTS: Less concern over vaccine side effects could improve the willingness to be vaccinated
(probability changed by 7.7 pp; p < 0.10). The current vaccine distrust espoused by the Brazilian
president is associated with vaccine hesitancy, among his voter base. Lower performance
perception (“Very Bad” with 10.7 pp; p < 0.01) or higher political opposition (left-oriented)
regarding the current presidency is associated with the willingness to be vaccinated. Higher
compliance with non-pharmaceutical interventions (NPIs) is usually positively associated with
the willingness to take the COVID-19 vaccine (+1 score to NPI compliance index is associated
with higher willingness to be vaccinated by 1.4 pp, p < 0.05).
CONCLUSION: Willingness to be vaccinated is strongly associated with political leaning,
perceived federal government performance, vaccine side effects, and compliance with
non-pharmaceutical interventions (NPIs).
DESCRIPTORS: COVID-19 Vaccines. Vaccination Refusal. Socioeconomic Factors. Political
Activism. Health Knowledge, Attitudes, Practice.
Correspondence:
Marco Antonio Catussi Paschoalotto
Travessa do Hospital, 18 - 2° andar
1150-187 Lisboa, Portugal
E-mail: [email protected]
Received: May 27, 2021
Approved: Jul 19, 2021
How to cite: Paschoalotto MAC,
Costa EPPA, Valente-de-Almeida
S, Cima J, Gomes-da-Costa
J, Santos JV, et al. Running
away from the jab: Factors
Associated with COVID-19
Vaccine Hesitancy in Brazil.
Rev Saude Publica. 2021;55:97.
https://doi.org/10.11606/s1518-
8787.2021055003903
Copyright: This is an open-access
article distributed under the
terms of the Creative Commons
Attribution License, which permits
unrestricted use, distribution, and
reproduction in any medium,
provided that the original author
and source are credited.
http://www.rsp.fsp.usp.br/
2
Factors associated with COVID-19 vaccine hesitancy
Paschoalotto MAC et al.
https://doi.org/10.11606/s1518-8787.2021055003903
INTRODUCTION
By July 2021, the COVID-19 pandemic had already resulted in more than 186 million cases
and 4 million deaths worldwide, with Brazil ranking third place in the number of cases and
second in the number of deaths1. In a global effort to contain the spread of the new virus,
countries adopted several non-pharmacological interventions (NPI) such as social
distancing2–4 and face mask use5. But despite the importance of such measures, the solution
to the pandemic rests on the success of vaccination programs6,7.
After the extraordinary efforts made to rapidly research and develop effective COVID-19
vaccines and their recent rollout, researchers and the media have pointed to a growing
concern regarding public confidence in the vaccination process. In fact, “anti-vaccine
movements” can foster vaccine hesitancy, reducing the population’s willingness
to be vaccinated6–13.
Several surveys have been used to characterize behaviours concerning vaccine hesitancy
and NPI compliance10–18. According to the existing literature, sociodemographic conditions
(e.g., education, age, or job occupation)10,18–20 and political and organizational trust
aspects5,8–10,12,13, can affect people’s willingness to be vaccinated.
One of the countries with the highest number of COVID-19 cases and deaths1,
Brazil has a population of diverse sociodemographic backgrounds20–23 and is governed
by a president with a long history of questioning scientific findings, including
vaccine efficacy and safety24–26. Reducing vaccine hesitancy will largely determine
Brazil’s – and other low-middle income countries (LMICs) – success in controlling the
current pandemic.
Given this context, this study investigates the factors associated with COVID-19 vaccine
hesitancy in Brazil. Using a nationwide online survey, we analyse how sociodemographic
conditions, political factors, organizational confidence, and non-pharmaceutical
interventions compliance influence the population’s willingness to be vaccinated.
METHODS
This study was approved by the Research Ethics Committee at NOVA School of Business
and Economics (Portugal) on November 23rd, 2020, via letter sent by the Scientific Council’s
president. Regarding Brazilian ethical standards, the research complied with the National
Health Council Resolution 466/12a. In its first page, the online survey highlighted the
research characteristics and information, anonymity assurance, data protection, and a
consent form.
Data
Data collection took place between November 25th, 2020 and January 11th, 2021, period
before the second COVID-19 wave, considered the deadliest so far, and before the first
COVID-19 vaccine (Coronavac – Sinovac/Butantã) was introduced. Using an online survey
built on Qualtrics software and disseminated on different social networks (Facebook,
Instagram, WhatsApp, and email groups), we sought to collect a diversified base of
responses from all Brazilian regions and different social sectors.
Table 1 describes the survey data and compares it to the national averages.
Our sample comprised 1,623 valid responsesb, collected from almost all Brazilian
states and capitals, but mainly from São Paulo (67%). While not representative of
the Brazilian population, the study sample comes close to some sociodemographic
characteristics, such as gender, age, residence area, number of households and
professional situation27.
a Ministério da Saúde (BR),
Conselho Nacional de
Saúde. Resolução Nº 466 de
12 de dezembro de 2012.
Aprova diretrizes e normas
regulamentadoras de pesquisas
envolvendo seres humanos.
Diário Oficial da União. 13 jun
2013; Seção 1: 59. Available
from: https://conselho.saude.gov.
br/resolucoes/2012/Reso466.pdf
b These responses include all
completed and submitted
responses recorded after
validation testing. Participants
were given the option not
to disclose their political
preferences, perception of
vaccine side effects, perception
of the federal government, and
compliance levels.
3
Factors associated with COVID-19 vaccine hesitancy
Paschoalotto MAC et al.
https://doi.org/10.11606/s1518-8787.2021055003903
Table 1. Sample characteristics (Survey Data) x National characteristics (National Data).
Variable
Survey Data
National Data
States and municipalities (number)a
States
24
27
Capitals
20
27
Municipalities
263
5,570
Gender (%)
Male
37.9
48.2
Female
61.7
51.8
Other/No answer
0.4
Age (%)
≤ 18 years
0.9
< 19 years
33.1
19 to 25 years
30.6
20 to 24 years
9.0
26 to 32 years
20.9
25 to 34 years
17.1
33 to 45 years
27.4
35 to 44 years
14.0
46 to 64 years
18.4
45 to 64 years
19.2
65 to 79 years
1.9
65 to 79 years
6.0
≥ 80 years
0.1
≥ 80 years
1.6
Education (%)
Elementary school
0.5
55.8
High school
14.4
30.1
University – Bachelor
40.5
14.1
University – MBAs and specializations
20.8
University – Master’s
14.1
University – Doctorate
9.7
Residence area (%)
Urban
97.4
84.4
Rural
2.6
15.6
Households (%; average number)
One/Live alone
7.9
30.9
Two
33.3
Three
26.9
30.4
Four
20.7
22.8
Five
7.2
10.0
More than five
4.0
5.9
Professional situation (%)
Retired
2.8
Out of the
workforce
37.2
Student
21.2
Unemployed
6.5
Unoccupied
6.6
Public server
17.2
Occupied
39.1
Worker – Own business
10.9
Worker – SME enterprises
15.4
Worker – Big enterprises
22.1
Other/No answer
3.9
Other
17.1
a Sample comprising 88.9% of the Brazilian states, 74.7% of the Brazilian capitals, and 4.7% of the Brazilian
municipalities. More than 75% of the Brazilian municipalities are characterized as “small” (< 25,000 inhabitants),
reducing the likelihood of achieving a substantial representativeness for them (31).
4
Factors associated with COVID-19 vaccine hesitancy
Paschoalotto MAC et al.
https://doi.org/10.11606/s1518-8787.2021055003903
Beyond sociodemographic conditions, we also have collected data regarding political factors,
organizational confidence, NPI compliance, perception of vaccine side effects and vaccine
hesitancy (Appendix 1).
Respondents were asked to disclose their political leaning on a scale of 1 (Far Left) to 7 (Far
Right) and to qualitatively evaluate (Very Bad, Bad, Good and Very Good) their perception
of several institutions’ performance concerning the COVID-19 pandemic, including the
Federal Government.
Regarding NPI compliance – mandatory mask use, social distancing (1,5 meters), respiratory
etiquette, hand washing and staying at home –, respondents were asked to disclose their
agreement level using a 5-point scale (disagree – agree), and their compliance level (never,
rarely, frequently, and always) (Appendix 1). By means of a Principal Content Analysis (PCA),
we used these questions to create a composite indicator labelled as “Compliance Index,”
which represents 47.68% of the explanatory power of the total variables. Each measure
contributed to the compliance index with different weights: mandatory mask use – 19.86%;
social distancing (1.5 meters) – 21.84%; respiratory etiquette – 16.87%; hand washing –
20.49%; and staying at home, if possible – 20.94%.
As for vaccines, respondents were asked about their perception of vaccine side effects and
willingness to be vaccinated (no, maybe, and yes).
Data Analysis
We performed a set of bivariate analyses to understand the association between key
variables – NPI Compliance Index, Age (years), Gender, Schooling level, Vaccine side effect,
Political leaning and Government performance (Federal) – and willingness to take the
COVID-19 vaccine. Subsequently, we used logistic regression models to estimate COVID-19
vaccine hesitancy. Using willingness to be vaccinated (measured by no/maybe (0), and yes
(1)) as the dependent variable, the first model considers the baseline sociodemographic
conditions as independent variables; the second model, in turn, includes political
leaning, organizational confidence, non-pharmaceutical interventions compliance, and
vaccine side effects as independent variables. Results are presented as Odds Ratios (OR),
which indicate the odds of a dependent variable occurring in the presence or absence
of the reference group, and as marginal effects (dy/dx), which tells us, in percentage
points (pp), how a dependent variable changes when an explanatory variable changes,
ceteris paribus.
RESULTS
Descriptive Statistics
Regarding the willingness to take the COVID-19 vaccine, 70% of the sample showed to be
willing to take the COVID-19 shot, while almost 30% exhibited some degree of hesitancy
(“not” or “maybe”) (Figure A). Such willingness to be vaccinated assumes that a vaccine is
available for a given individual.
Plots 1B to 1H show the association between willingness to be vaccinated and the
independent variables. Divided into tertiles, the NPI Compliance Index (Figure B) ranges
from lower compliance (1) to higher compliance (3) levels, suggesting a possible association
between this variable and willingness to be vaccine, with a higher percentage of “Yes” at
the level “3”, than at the level “1.” Such findings may reflect the population’s level of concern:
more concerned individuals are willing to be vaccinated and show higher compliance with
sanitary measures.
As for the association between age and willingness to be vaccinated (Figure C), younger (less
than 25 years) and older (more than 65 years) individuals showed higher levels of hesitancy,
5
Factors associated with COVID-19 vaccine hesitancy
Paschoalotto MAC et al.
https://doi.org/10.11606/s1518-8787.2021055003903
Figure. Bivariate analysis plots (except for 1A), respectively: Willingness to be vaccinated (A); Willingness to be vaccinated and NPI
Compliance Index (B); Willingness to be vaccinated and Age (C); Willingness to be vaccinated and Gender (D); Willingness to be vaccinated
and Schooling level (E); Willingness to be vaccinated and Vaccine side effects (F); Willingness to be vaccinated and Political leaning (G);
Willingness to be vaccinated and Federal Government performance (H).
A
B
C
D
E
F
G
H
100
80
60
40
20
0
Willingness to COVID-19 Vaccine
No
Maybe
1
2
3
NPI Compliance Index
Gender
Age (years)
Female
Male
1
2
3
4
5
6
7
Political position
Government performance
Very bad
Bad
Average
Good
Very
good
University
High school
Elementary
school
Schooling
Disagree
Partially
disagree
Neutral
Partially
agree
Agree
Yes
Vaccine side effect
<18 yrs
19–25 yrs 26–32 yrs 33–45 yrs 46–65 yrs
>65 yrs
100
80
60
40
20
0
100
80
60
40
20
0
100
80
60
40
20
0
100
80
60
40
20
0
100
80
60
40
20
0
100
80
60
40
20
0
100
80
60
40
20
0
6
Factors associated with COVID-19 vaccine hesitancy
Paschoalotto MAC et al.
https://doi.org/10.11606/s1518-8787.2021055003903
while those between 26 and 65 years old were less hesitant. In our sample, women showed
greater hesitancy regarding the COVID-19 vaccine than men (Figure D).
As expected, the analysis found a strong association between schooling level and vaccine
hesitancy (Figure E): individuals with only elementary schooling show vaccine hesitancy
levels up to four times higher than those with higher schooling levels. Moreover, individuals
more concerned with vaccine side effects show greater hesitancy in their willingness to be
vaccinated (Figure F).
In our sample, right-wing individuals – generally more favorable to the current government –
show higher levels of vaccine hesitancy than left-wing individuals. Together with the previous
findings, this suggests that distrust in government is associated with higher compliance
and vaccine acceptance, possibly due to high levels of concern (Figure G). We observed
a similar inverse relationship between perception of government and willingness to be
vaccinated (Figure H): respondents who scored government action as “Very bad” showed
and 86% willingness to be vaccinated; among those who scored the government action as
“Very good”, in turn, this willingness drops to 38%.
Logistic Regression Models
In this study, we performed two regression models to estimate the factors associated
with the willingness to take the COVID-19 vaccine. While model 1 includes only
sociodemographic characteristics, model 2 considers the participants’ opinion on vaccine
side effects, political leaning, perception of federal government performance and the
compliance indexc. This section focuses on the marginal effects analysis, but full results
are shown below (Table 2).
In both models, age group does not seem to explain willingness to be vaccinated. Being
retired is associated with the probability of taking the COVID-19 vaccine by 17.9 pp
(p < 0.01) and by 14.5 pp (p < 0.05) in the first and second model, respectively, being the
only professional situation with significant impact on the dependent variable – relative to
being unemployed (baseline group). Although we found a positive impact associated with
being male in the first model, this loses significance once we control for opinion on vaccine
effects and compliance index. We observed similar results regarding educational variables
such as Master’s and PhD programs.
The second model shows a negative and statistically significant association between fear of
vaccine side effects and willingness to be vaccinated. Respondents who answered having
no concern over vaccine side effects show higher levels of willingness to be vaccinated,
with their probability changing by 7.7 pp (p < 0.10). On the other hand, individuals with
high levels of concern about side effects have lower willingness to be vaccinated, varying
by 34.4 pp (p < 0.01). Regarding political leaning, results show an association between being
left-oriented and willingness to take the vaccine. Rating the government’s performance
as “very bad” affects the probability of agreeing to be vaccinated by 10.7 pp (p < 0.01).
The compliance index, which gives us an indicator of the participants’ overall level of
compliance with all preventive measures, is in turn positively associated with willingness
to vaccinate. An extra score on the compliance index means a 1.4 pp (p < 0.05) change in
the probability of agreeing to vaccinate.
c Compliance Index explained in
detail in the methods section.
7
Factors associated with COVID-19 vaccine hesitancy
Paschoalotto MAC et al.
https://doi.org/10.11606/s1518-8787.2021055003903
Table 2. Logit models analyzing the explanatory capacity of the independent variables concerning the
willingness to be vaccinated.
(1)
(1)
(2)
(2)
OR
dydx
OR
dydx
Compliance Index
1.123b
0.014b
Age (years)
(baseline group ≤ 18)
19–25
0.912
-0.018
0.951
-0.006
26–32
1.014
0.003
0.841
-0.020
33–45
1.002
0.0004
0.877
-0.015
46–64
0.785
-0.048
0.597
-0.063
≥ 65
0.447
-0.174
0.619
-0.058
Gender
(baseline group: Female)
Male
1.324b
0.054b
1.218
0.023
Professional situation
(baseline group: Unemployed)
Retired
2.93b
0.179c
3.867a
0.145b
Student
1.391
0.066
1.012
0.002
Other
0.834
-0.039
1.080
0.010
Public server
1.424
0.070
1.405
0.042
Worker – Big enterprises
1.178
0.034
1.829
0.072
Worker – SME enterprises
1.122
0.024
1.266
0.029
Worker – Own business
0.936
-0.014
1.225
0.025
Schooling level
(baseline group: Elementary school)
High school
2.940
0.246
2.856
0.127
University – Bachelor
1.960
0.161
1.218
0.027
University – MBAs and specializations
2.827
0.238
1.700
0.069
University – Master
4.747a
0.328
2.692
0.121
University – PhD
5.103a
0.338∗
2.049
0.091
Vaccine side effects
(baseline group: do not disagree or agree)
Fully disagree
3.454a
0.077b
Partially disagree
2.346a
0.060b
Partially agree
0.503b
-0.077c
Fully agree
0.108c
-0.344c
Political leaning
(baseline group: Center)
1- Far left
0.896
-0.014
2
1.869b
0.072b
3
1.553a
0.053a
5
0.690
-0.050
6
0.476c
-0.104b
7 - Far right
0.388b
-0.136b
Federal government - Performance
(baseline group: Fair)
Very bad
2.355c
0.107c
Bad
1.337
0.039
Good
0.699
-0.052
Very good
0.532
-0.095
N
1,623
1,623
1,261
1,261
a, b, c: indicate significance at 10%, 5% and 1% level, respectively.
Note: We also ran ordered logit models, which presented the same significative results.
8
Factors associated with COVID-19 vaccine hesitancy
Paschoalotto MAC et al.
https://doi.org/10.11606/s1518-8787.2021055003903
DISCUSSION
This study investigated the association between social characteristics, political factors,
and organizational performance and vaccine hesitancy in Brazil, contributing to
understanding vaccine hesitancy factors in a LMIC context.
Our main finding suggests a negative association between positive perception of the federal
government’s performance and willingness to be vaccinated, similar to previous studies
on the likelihood of getting vaccinated in Brazil26. It also corroborates a North-American
study, conducted during the Trump Administration, which suggested higher vaccine
hesitancy among Trump supporters18. This phenomenon can be explained by the current
Brazilian president’s negationist remarks regarding the COVID-19 pandemic and his
position against compliance with NPIs and being vaccinated – a political scenario similar
to the Trump administration24,25,28.
Regarding political leaning, our results show that espousing far-right ideology is positively
associated with vaccine hesitancy, while being centre-left is associated with vaccine
acceptance. This finding corroborates other studies on anti-vaccine movements and
ideological isolation11–13,26) and reinforces the importance of political leadership in promoting
compliance and public trust during crisis.
The NPI compliance index also provided interesting results, showing a positive association
with willingness to be vaccinated. Such index is an innovative approach already used in
previous studies4,10,18 and our results are in agreement with the literature5,11,13. We found
a similar association regarding vaccine side effects, with more concerned individuals
showing a positive association with willingness to be vaccinated. Such results highlight
the importance of public communication about NPIs and vaccines.
This research has two major limitations. First, the method of data collection prevented
us from obtaining a representative sample, particularly regarding the vulnerable
population, which was underrepresented. Research shows that the most vulnerable
individuals (with low schooling levels and high poverty levels) may express least
willingness to be vaccinated10,18–20. If we transpose this scenario to the Brazilian context,
then our vaccine hesitancy estimates should be interpreted as a lower bound. Like
previous studies with convenient sampling methods17,18, however, the present study
can still be used to derive significant policies. Even if the sample is not representative
of the entire population, it can be for particular groups. Second, some respondents
were not comfortable disclosing their political leanings, thus reducing the number of
observations available in the second model. If such respondents are not distributed
randomly, then the results may be biased.
Overall, the study contributes to a better understanding of vaccine hesitancy factors in a
low-to-middle income country. Vaccine hesitancy is associated with multiple factors, such
as NPIs compliance, sociodemographic and employment characteristics, political leaning,
and public perception of government performance. Willingness to be vaccinated in Brazil
is strongly associated with political leaning, perceived federal government performance,
vaccine side effects, and compliance with non-pharmaceutical interventions. We found a
strong association between vaccine hesitancy and being right-wing and positive perception
of government performance. These findings suggest that the current distrust shown by the
Brazilian president regarding vaccines contributes to vaccine hesitancy among his voter
base. Individuals who oppose the current government, in turn, show higher willingness to
be vaccinated.
REFERENCES
1. Johns Hopkins University & Medicine, Johns Hopkins Coronavirus Resource Center. Baltimore,
MD: CRC; 2021 [cited 2021 May 18]. Available from: https://coronavirus.jhu.edu/
9
Factors associated with COVID-19 vaccine hesitancy
Paschoalotto MAC et al.
https://doi.org/10.11606/s1518-8787.2021055003903
2. Hughes RP, Hughes DA. Impact of relaxing COVID-19 social distancing measures
on rural North Wales: a simulation analysis. Front Public Health. 2020;8:562473.
https://doi.org/10.3389/fpubh.2020.562473
3. Duczmal LH, Almeida ACL, Duczmal DB, Alves CRL, Magalhães FCO, Lima MS, et al. Vertical
social distancing policy is ineffective to contain the COVID-19 pandemic. Cad Saude Publica.
2020;36(5):e00084420. https://doi.org/10.1590/0102-311X00084420
4. Almeida SV, Costa E, Lopes FV, Santos JV, Barros PP. Concerns and adjustments:
how the Portuguese population met COVID-19. PLoS One. 2020;15(10):e0240500.
https://doi.org/10.1371/journal.pone.0240500
5. Chernozhukov V, Kasahara H, Schrimpf P. Causal impact of masks, policies,
behavior on early COVID-19 pandemic in the U.S. J Econom. 2021;220(1):23-62.
https://doi.org/10.1016/j.jeconom.2020.09.003
6. Chou WYS, Budenz A. Considering emotion in COVID-19 vaccine communication: addressing
vaccine hesitancy and fostering vaccine confidence. Health Commun. 2020;35(14):1718-22.
https://doi.org/10.1080/10410236.2020.1838096
7. Harrison EA, Wu JW. Vaccine confidence in the time of COVID-19. Eur J Epidemiol.
2020;35(4):325-30. https://doi.org/10.1007/s10654-020-00634-3
8. Brunson EK, Schoch-Spana M. A social and behavioral research agenda to facilitate
COVID-19 vaccine uptake in the United States. Health Secur. 2020;18(4):338-44.
https://doi.org/10.1089/hs.2020.0106
9. Kasstan B. Vaccines and vitriol: an anthropological commentary on vaccine hesitancy,
decision-making and interventionism among religious minorities. Anthropol Med.
2020 Nov 13:1-9. https://doi.org/10.1080/13648470.2020.1825618. Epub ahead of print.
10. Lin Y, Hu Z, Zhao Q, Alias H, Danaee M, Wong LP. Understanding COVID-19 vaccine
demand and hesitancy: a nationwide online survey in China. PLoS Negl Trop Dis.
2020;14(12):e0008961. https://doi.org/10.1371/journal.pntd.0008961
11. Puri N, Coomes EA, Haghbayan H, Gunaratne K. Social media and vaccine hesitancy: new
updates for the era of COVID-19 and globalized infectious diseases. Hum Vaccin Immunother.
2020;16(11):2586-93. https://doi.org/10.1080/21645515.2020.1780846
12. Ward JK, Alleaume C, Peretti-Watel P, Seror V, Cortaredona S, Launay O, et al. The French
public’s attitudes to a future COVID-19 vaccine: the politicization of a public health issue.
Soc Sci Med. 202;265:113414. https://doi.org/10.1016/j.socscimed.2020.113414
13. Callaghan T, Moghtaderi A, Lueck JA, Hotez P, Strych U, Dor A, et al. Correlates and
disparities of intention to vaccinate against COVID-19. Soc Sci Med. 2021;272:113638.
https://doi.org/10.1016/j.socscimed.2020.113638
14. Lima-Costa MF, Macinko J, Andrade FB, Souza Júnior PRB, Vasconcellos MTL,
Oliveira CM. ELSI-COVID-19 initiative: methodology of the telephone survey on coronavirus
in the Brazilian Longitudinal Study of Aging. Cad Saude Publica. 2020;36 Suppl 3:e00183120.
https://doi.org/10.1590/0102-311X00183120
15. Wouters OJ, Shadlen KC, Salcher-Konrad M, Pollard AJ, Larson HJ, Teerawattananon
Y, et al. Challenges in ensuring global access to COVID-19 vaccines: production,
affordability, allocation, and deployment. Lancet. 2021;397(10278):1023-34.
https://doi.org/10.1016/S0140-6736(21)00306-8
16. Murphy J, Vallières F, Bentall RP, Shevlin M, McBride O, Hartman TK, et al.
Psychological characteristics associated with COVID-19 vaccine hesitancy and
resistance in Ireland and the United Kingdom. Nat Commun. 2021;12(1):29.
https://doi.org/10.1038/s41467-020-20226-9
17. Latkin CA, Dayton L, Yi G, Konstantopoulos A, Boodram B. Trust in a COVID-19
vaccine in the U.S.: a social-ecological perspective. Soc Sci Med. 2021;270:113684.
https://doi.org/10.1016/j.socscimed.2021.113684
18. Lazarus JV, Ratzan SC, Palayew A, Gostin LO, Larson HJ, Rabin K, et al. A global
survey of potential acceptance of a COVID-19 vaccine. Nat Med. 2021;27(2):225-8.
https://doi.org/10.1038/s41591-020-1124-9
19. Silva LLS, Lima AFR, Polli DA, Razia PFS, Pavão LFA, Cavalcanti MAFH, et al.
Social distancing measures in the fight against COVID-19 in Brazil: description
and epidemiological analysis by state. Cad Saude Publica. 2020;36(9):e00185020.
https://doi.org/10.1590/0102-311X00185020
10
Factors associated with COVID-19 vaccine hesitancy
Paschoalotto MAC et al.
https://doi.org/10.11606/s1518-8787.2021055003903
20. Ranzani OT, Bastos LSL, Gelli JGM, Marchesi JF, Baião F, Hamacher S, et al.
Characterisation of the first 250 000 hospital admissions for COVID-19 in Brazil:
a retrospective analysis of nationwide data. Lancet Respir Med. 2021;9(4):407-18.
https://doi.org/10.1016/S2213-2600(20)30560-9
21. Alves JA, Gibson CL. States and capitals of health: multilevel health governance in Brazil.
Lat Am Polit Soc. 2019;61(1):54-77. https://doi.org/10.1017/lap.2018.59
22. Penna GO, Silva JAA, Cerbino Neto J, Temporão JG, Pinto LF. PNAD COVID-19: a powerful
new tool for public health surveillance in Brazil. Cienc Saude Coletiva. 2020;25(9):3567-71.
https://doi.org/10.1590/1413-81232020259.2400
23. Castro MC, Kim S, Barberia L, Ribeiro AF, Gurzenda S, Ribeiro KB, et al.
Spatiotemporal pattern of COVID-19 spread in Brazil. Science. 2021;272(6544):821-6.
https://doi.org/10.1126/science.abh1558
24. Orellana JDY, Cunha GM, Marrero L, Moreira RI, Leite IC, et al. [Excess deaths during the
COVID-19 pandemic: underreporting and regional inequalities in Brazil]. Cad Saude Publica.
2020;36(1):e00259120. Portuguese. https://doi.org/10.1590/0102-311X00259120
25. Cotrin P, Moura W, Gambardela-Tkacz CM, Pelloso FC, Santos L, Carvalho MDB, et al.
Healthcare workers in Brazil during the COVID-19 pandemic: a cross-sectional online survey.
Inquiry. 2020;57:46958020963711. https://doi.org/10.1177/0046958020963711
26. Gramacho WG, Turgeon M. When politics collides with public health: COVID-19 vaccine
country of origin and vaccination acceptance in Brazil. Vaccine. 2021;39(19):2608-12.
https://doi.org/10.1016/j.vaccine.2021.03.080
27. Lazarus JV, Wyka K, Rauh L, Rabin K, Ratzan S, Gostin LO, et al. Hesitant or not?
The association of age, gender, and education with potential acceptance of a
COVID-19 vaccine: a country-level analysis. J Health Commun. 2020;25(10):799-807.
https://doi.org/10.1080/10810730.2020.1868630
28. Troiano G, Nardi A. Vaccine hesitancy in the era of COVID-19. Public Health.
2021;194:245-51. https://doi.org/10.1016/j.puhe.2021.02.025
29. Paul E, Steptoe A, Fancourt D. Attitudes towards vaccines and intention to vaccinate
against COVID-19: implications for public health communications. Lancet Reg Health Eur.
2021;1:100012. https://doi.org/10.1016/j.lanepe.2020.100012
30. Figueiredo A, Simas C, Karafillakis E, Paterson P, Larson HJ. Mapping global trends
in vaccine confidence and investigating barriers to vaccine uptake: a large-scale
retrospective temporal modelling study. Lancet. 2020;26;396(10255):898-908.
https://doi.org/10.1016/S0140-6736(20)31558-0
Funding: Eduardo Costa was funded by Fundação para a Ciência e a Tecnologia (FCT) under PhD grant number
BD128545/2017. Joana Gomes da Costa was funded by Fundação para a Ciência e a Tecnologia (FCT) under PhD
grant number SFRH/BD/140727/2018. The remaining authors have no financial relationships relevant to this
article to disclose.
Authors’ Contribution: Study design and planning: MACP, EPPAC, SVA, JC, JGC, JVS, PPB, CSP, JLP. Data
collection, analysis and interpretation: MACP, EPPAC, SVA, JC, JGC, JVS, PPB, CSP, JLP. Manuscript drafting or
review: MACP, EPPAC, SVA, JC, JGC, JVS, PPB, CSP, JLP. Approval of the final version: MACP, EPPAC, SVA, JC, JGC,
JVS, PPB, CSP, JLP. Public responsibility for the content of the article: MACP, EPPAC, SVA, JC, JGC, JVS, PPB, CSP, JLP.
Conflict of Interests: The authors declare no conflict of interest.
| Running away from the jab: factors associated with COVID-19 vaccine hesitancy in Brazil. | 11-26-2021 | Paschoalotto, Marco Antonio Catussi,Costa, Eduardo Polena Pacheco Araújo,Almeida, Sara Valente de,Cima, Joana,Costa, Joana Gomes da,Santos, João Vasco,Barros, Pedro Pita,Passador, Claudia Souza,Passador, João Luiz | eng |
PMC6211760 | RESEARCH ARTICLE
Cardiorespiratory and metabolic responses
and reference equation validation to predict
peak oxygen uptake for the incremental
shuttle waking test in adolescent boys
Andreza L. Gomes1☯, Vanessa A. Mendonc¸a1☯, Tatiane dos Santos Silva1‡, Crislaine K.
V. Pires1‡, Liliana P. Lima1‡, Alcilene M. Silva1‡, Ana Cristina R. Camargos1,2‡, Camila D.
C. Neves1‡, Ana C. R. Lacerda1‡, He´rcules R. LeiteID1☯*
1 Programa de Po´s-Graduac¸ão em Reabilitac¸ão e Desempenho Funcional, Departamento de Fisioterapia,
Universidade Federal dos Vales do Jequitinhonha e Mucuri (UFVJM), Campus JK, Alto da Jacuba,
Diamantina, Minas Gerais, Brazil, 2 Escola de Educac¸ão Fı´sica, Fisioterapia e Terapia Ocupacional
(EEFFTO), Departamento de Fisioterapia, Universidade Federal de Minas Gerais (UFMG), Diamantina,
Minas Gerais, Brazil
☯ These authors contributed equally to this work.
‡ These authors also contributed equally to this work.
* [email protected]
Abstract
Background
Previous studies speculated that the Incremental Shuttle Walking Test (ISWT) is a maximal
test in children and adolescents, however comparison between ISWT with cardiopulmonary
exercise test has not yet performed. Furthermore, there is no regression equation available
in the current literature to predict oxygen peak consumption (VO2 peak) in this population.
This study aimed to assesses and correlate the cardiorespiratory responses of the ISWT
with the cardiopulmonary exercise (CEPT) and to develop and validate a regression equa-
tion to predict VO2 peak in healthy sedentary adolescent boys.
Methods
Forty-one participants were included in the study. In the first stage, the VO2 peak, respira-
tory exchange ratio (R peak), heart rate max (HR max) and percentage of predicted HR max
(% predicted HR max) were evaluated in CEPT and ISWT (n = 26). Second, an equation
was developed (n = 29) to predict VO2 peak. In both phases, the VO2 peak, respiratory
exchange ratio R and hearth rate (HR) were evaluated. In the third stage, the validation
equation was performed by another 12 participants.
Results
Similar results in VO2 peak (P>0.05), R peak (P>0.05) and predicted maximum HR
(P>0.05) were obtained between the ISWT and CEPT. Both tests showed moderate signifi-
cant correlations of VO2 peak (r = 0.44, P = 0.002) e R peak (r = -0.53, P < 0.01), as well as
PLOS ONE | https://doi.org/10.1371/journal.pone.0206867
November 1, 2018
1 / 11
a1111111111
a1111111111
a1111111111
a1111111111
a1111111111
OPEN ACCESS
Citation: Gomes AL, Mendonc¸a VA, Santos Silva
Td, Pires CKV, P. Lima L, Silva AM, et al. (2018)
Cardiorespiratory and metabolic responses and
reference equation validation to predict peak
oxygen uptake for the incremental shuttle waking
test in adolescent boys. PLoS ONE 13(11):
e0206867. https://doi.org/10.1371/journal.
pone.0206867
Editor: Gustavo Batista Menezes, UFMG, BRAZIL
Received: August 17, 2018
Accepted: October 19, 2018
Published: November 1, 2018
Copyright: © 2018 Gomes et al. This is an open
access article distributed under the terms of the
Creative Commons Attribution License, which
permits unrestricted use, distribution, and
reproduction in any medium, provided the original
author and source are credited.
Data Availability Statement: All relevant data are
within the manuscript and its supporting
information files.
Funding: The authors are grateful to Brazilian
agencies CNPq, CAPES and FAPEMIG for financial
support.
Competing interests: The authors have declared
that no competing interests exist.
the agreement of these measurements by Bland-Altman analysis (VO2 peak, bias = -0.13; R
peak, bias = 0.0). Distance walked was the variable that explained 42.5% (R2 = 0.425, p =
0.0001) of the variance in VO2 peak. The equation was VO2 peak (predicted) = 20.94 +
(0.02 x distance walked). The results obtained by the equation were not significantly differ-
ent compared to the values obtained by the gas analyzer and the Bland-Altman analysis
showed agreement (bias = 1.6).
Conclusion
The ISWT produced maximal cardiorespiratory responses comparable to the CEPT, and
the developed equation showed viability for the prediction of VO2 peak in healthy sedentary
adolescent boys.
Introduction
Assessment of functional capacity or cardiorespiratory fitness (CRF) is defined as the ability to
perform a moderate to high intensity exercise involving large muscle groups over a period of
time [1,2]. It is an important component of health related physical fitness, which reflects the
functional capacities of the respiratory, cardiovascular and musculoskeletal systems [1]. The
CRF assessment has been widely used in clinical practice and research aiming to provide
parameters for physical activity prescription and to evaluate reduced exercise tolerance in sev-
eral health conditions [3–5].
The performance of a cardiopulmonary exercise testing (CEPT) followed by the measure-
ment of peak oxygen consumption (VO2 peak) through the direct analysis of exhaled gases is
the most commonly reported procedure in the literature for the evaluation of CRF [6]. How-
ever, this measurement is often infeasible because its require high-cost equipment, specialized
laboratory and trained professionals [7]. Thus, field test and prediction equation to indirect
estimate VO2 peak in clinical practice has been widely implemented [2]. Among the field test,
we highlight the Incremental Shuttle Walking Test (ISWT) developed by Sing et al., [8] which
comprises as a simple incremental walk test with pace dictated by external stimulus which
asses CRF based on distance walked. Despite being developed initially for individuals with
chronic obstructive pulmonary disease [8], it has been used recently in different health condi-
tions and age groups [5,9,10].
Some studies have used ISWT to asses CRF in children and adolescents with asthma [11],
scoliosis [5] and very low premature newborn [10]. However, the application and intensity of
this test in a healthy population is scarce. Lanza et al., [9] developed an equation to predict dis-
tance walked and also showed that the ISWT is a maximal test in a children and adolescent
population determined indirectly by means of the maximum heart rate (HR max) achieved at
the end of the test. On the other hand, Coelho et al. [12] demonstrated that a healthy control
group of children and adolescents showed submaximal values of HR max in the ISWT. How-
ever, these authors fail to confirm the cardiorespiratory responses with the completion and
comparison with the CEPT.
Taken together, there is a gap in the current literature regarding the ISWT intensity valida-
tion, as well as an equation to predict VO2 peak in the adolescent population. Thus, the present
study aims to evaluate and correlate the cardiorespiratory outcomes during the ISWT and a
CEPT, in order to classify the intensity of ISWT, and to develop and validate an equation to
Incremental shuttle walking test in healthy sedentary adolescent boys
PLOS ONE | https://doi.org/10.1371/journal.pone.0206867
November 1, 2018
2 / 11
predict VO2 peak in healthy adolescent boys. We postulate that the ISWT would promote
maximal cardiorespiratory responses in agreement with the CEPT, and the regression equa-
tion would be feasible for predicting VO2 peak in healthy adolescent boys.
Materials and methods
This was a cross-sectional study that included 41 healthy adolescent boys. They were recruited
by convenience from private and public schools. The protocol began in July 2016 and ended in
November 2017. All measurements were obtained in the physiology of exercise laboratory
(UFVJM) by trained investigators. The parents were asked to report the health history of the
subject (i.e., prematurity birth and current physical activity engagement and comorbidities).
Thus, the inclusion criteria were as follows: male boys, ages 12–18 years old, absence of chronic
or acute diseases, physical activity engagement less than three times a week, preterm birth and
parents sign the consent form. The volunteers were excluded if they were unable to understand
the test. To meet the objectives, this study was divided into three stages. The first stage aimed
to evaluate the intensity of the ISWT; the second stage aimed to develop a regression for the
prediction of VO2 peak; and the third stage to validate the prediction equation. This study fol-
lowed the declaration of Helsinki. The Ethics and Research Committee of Universidade Fed-
eral dos Vales do Jequitinhonha e Mucuri (UFVJM), Brazil, approved this study (Protocol:
52980816.4.0000.5108). The following protocol description reproduces information already
reported elsewhere [2].
First stage procedures
In the first stage, 26 volunteers went to the laboratory on three consecutive days at the same
time period each day. On the first day, the body composition was assessed (weight, height,
BMI) and familiarization was performed. The weight and height were measured on an anthro-
pometric mechanical scale, the BMI was calculated as the weight divided by height squared
[3]. Familiarization consisted of testing that would be performed on consecutive days to
reduce the effect of learning. On the second and third days, the ISWT and CEPT were applied.
The testing order was randomized and balanced. The subjects were instructed to avoid physi-
cal activity and any intake of caffeine in the 24 hours prior to testing, to get at least 8 hours of
sleep the night before, to eat a light meal and to ingest 500 ml of water in two hours before the
tests. On the days of testing subjects were asked about their compliance with the recommenda-
tions above and about possible complications or changes in their daily routines [2].
To perform the ISWT, the participants were instructed to walk a distance of 10 meters
around a marking between two cones, placed 0.5m from each endpoint [8]. The walking speed
at which the participant should walk (or run) was dictated by a sound played from a CD that
was originally generated by a microcomputer. Each minute the walking speed increased by
0.17m/s. The test was finished when the volunteer was not able to maintain the required speed
(more than 0.5m from the cone), at the request of the volunteer, or for some other reported
symptom (dyspnea, dizziness, vertigo, angina). The original protocol consisted of 12 levels
(1020m); however, as suggested by the literature, we used a protocol of 15 levels (1500m) to
evaluate healthy participants, in order to prevent the ceiling effect [13]. Additionally, during
the testing, the laps were recorded to calculate the distance and gait speed reached at the last
full level. The ISWT were performed twice with at least 30 min of rest between them. The best
test (i.e., the longest distance walked) was considered for analysis. The maximum difference
between the tests should be 40 m [14]. A third test was performed when the difference was
greater than this. A trained professional conducted the tests. Before and after the test, heart
Incremental shuttle walking test in healthy sedentary adolescent boys
PLOS ONE | https://doi.org/10.1371/journal.pone.0206867
November 1, 2018
3 / 11
rate (HR, measured by a heart rate monitor) and blood pressure (measured by a mercury
sphygmomanometer cuff and a stethoscope) were measured.
The CEPT was performed on a treadmill using a protocol based on the progression of the
ISWT. This protocol consisted of 1-minute stages, with speed increasing every minute without
increasing the incline of the treadmill. The initial speed was 0.5 m/s, and it increased by 0.17
m/s at each stage. Before, during and after the test, heart rate and blood pressure were mea-
sured as described above. The criteria for stopping the test was as follows systolic blood
pressure (SBP) greater than 210 mmHg; diastolic blood pressure greater than 120 mmHg; sus-
tained decrease in SBP, angina dyspnea, cyanosis; nausea, dizziness or by the request of the
volunteer [15].
Second stage procedures
In the second stage 29 volunteers went to the laboratory at two different days. On the first day,
the body composition measurements were obtained as described in the first stage. On the sec-
ond day, the participants went to the laboratory for two ISWT with an interval of 30 minutes
between them. Completion of two ISWTs with this interval had been suggested to reduce
the effects of the learning test [13,16]. For the data analysis, the results of the test in which the
volunteer obtained the greatest distance covered were used. As with the first stage, the entire
procedure took place during a single day shift: the subjects were instructed to follow all the rec-
ommendations for the practice of physical tests, and prior to completion of the tests.
Cardiorespiratory and metabolic responses
During the tests of the two stages of this study, the exhaled gases were collected using a gas
analyzer via the portable telemetry system (k4b2, Cosmed, Rome, Italy). Among other vari-
ables, oxygen uptake (VO2), respiratory quotient (R) and HR breath-by-breath were moni-
tored. The absolute VO2 peak rate (mL/min) was expressed as relative rate defined as VO2
peak (mL/kg/min) and R peak the highest value of these measures at peak effort [17] and maxi-
mum heart rate (HRmax) as the highest HR value recorded during the test [2]. The maximum
predicted HR was calculated as 208 (0.7 age) [18].
Validation of the reference equation
To validate the equation, a different group of healthy males, composed of 12 individuals, was
selected according to the same inclusion criteria of the study. This group completed the ISWT
as described in the preceding stages. Likewise, the VO2 peak was predicted by the reference
equation.
Statistical analysis
The statistical analysis was performed using the statistical packages SPSS 22.0 (Inc., USA) and
GraphPad Prism 4 (Inc., USA). In the first stage, the normality of data was checked by the Sha-
piro–Wilk test and the differences among measured variables were determined by paired-t-
test for variables with normal distribution or the Wilcoxon test for variables with non-normal
distribution. Pearson’s coefficient of correlation was performed to study the correlation
between variables and the agreement between tests was assessed by Bland-Altman analysis.
The sample size was calculated based on the study by Neves et al [2] and was identified at least
10 participants. In the second stage, the normality of data was checked by Kolmogorov-Smir-
nov test and for compiling the reference equation, the linear multiple regression analysis was
performed to identify the predictors of the dependent variable. Multicollinearity was measured
Incremental shuttle walking test in healthy sedentary adolescent boys
PLOS ONE | https://doi.org/10.1371/journal.pone.0206867
November 1, 2018
4 / 11
by variance inflation factors (VIF). In this stage, the sample size was estimated on GPower
Software version 3.1 and based on the relationship between the numbers of variables to be
included in the multiple regression analysis and the minimum number of observations
required, indicating at least 29 participants in order to develop a linear model containing up to
three variables. At the end of the regression analysis, the paired t-test was utilized to compare
the means of the results obtained by the reference equation with the measured values of VO2
peak obtained using the gas analyzer. Moreover, the validation of the reference equation was
evaluated in an additional group of 12 volunteers: the values of VO2 peak obtained by the ref-
erence equation were compared with the measured values of VO2 peak obtained by the gas
analyzer using the paired t-test. The level of statistical significance was P<0.05.
Results
A total of 336 subjects were screened, but 186 did not return the baseline questionnaire. From
the 150 eligible participants, 28 reported any chronic, acute illness or reported premature
birth, 49 decline and 32 subjects were excluded for other reasons. The final sample was 41
male adolescents.
First stage
The general characteristics of the participants of first and second stage and their performance
on ISWT are showed in Table 1. The cardiorespiratory responses obtained at the end of the
ISWT and CEPT are presented in Table 2. Similar results in VO2 peak, R peak, and predicted
HRmax were found. Moderate and significant correlations in VO2 peak (r = 0.44, P = 0.02)
and R peak (r = -0.53, P<0.01) were found between the tests. The Bland-Altman analysis also
showed agreement between the results for VO2 peak (bias = -0.13) and R peak (bias = 0.00) on
the ISWT and CEPT (Fig 1A and 1B).
Table 2. Comparison between the results of cardiorespiratory variables at the end of the test, obtained in ISWT
and CEPT.
Outcome
Tests
Comparison between tests
ISWT (n = 26)
CEPT (n = 26)
P-value
VO2 peak (mL/kg/min)
44.02 (8.2)
44.2 (6.2)
0.93a
R peak
1.1 (0.2)
1.1 (0.1)
0.28b
HR max (% predicted)
98.8 (6.3)
96.6 (3.5)
0.63b
The data is presented as mean (SD). ISWT = incremental shuttle walking test; CEPT = cardiopulmonary exercise
testing; VO2 = oxygen uptake; R = respiratory exchange ratio; HR = heart rate.
aPaired-t test,
bWilcoxon test.
https://doi.org/10.1371/journal.pone.0206867.t002
Table 1. Characteristics of participants of the first, second and third stage.
Characteristics of the participants
(n = 41)
First phase
(n = 26)
Second phase
(n = 29)
Third phase
(n = 12)
Age (yr)
14.2 (1.8)
14.3 (1.8)
14.3 (1.9)
BMI (kg/m2)
19.5 (2.4)
19.5 (2.4)
20.5 (2.8)
Gait speed (m/s)
2.2 (0.3)
2.2 (0.3)
2.2 (0.4)
Distance walked (m)
923.9 (249.4)
938.0 (250.2)
915 (309.5)
The data is presented as mean (SD). BMI = body mass index.
https://doi.org/10.1371/journal.pone.0206867.t001
Incremental shuttle walking test in healthy sedentary adolescent boys
PLOS ONE | https://doi.org/10.1371/journal.pone.0206867
November 1, 2018
5 / 11
Fig 1. Agreement between VO2 (mL/kg/min) peak and R peak obtained in the ISWT and CEPT. (A) Bland-Altman plot of the
difference between the VO2 peak of the ISWT and CEPT plotted against the mean VO2 peak of the ISWT and CEPT; (B) Difference R
peak of the ISWT and CEPT plotted against the mean R peak of the ISWT and CEPT. ISWT = Incremental Shuttle Walking Test;
CEPT = cardiopulmonary exercise testing; VO2 = oxygen uptake; R = respiratory exchange ratio.
https://doi.org/10.1371/journal.pone.0206867.g001
Incremental shuttle walking test in healthy sedentary adolescent boys
PLOS ONE | https://doi.org/10.1371/journal.pone.0206867
November 1, 2018
6 / 11
Second stage
The characteristics of the participants of the second stage are showed in Table 1. Considering
the best ISWT, age, BMI and distance walked were the demographic, anthropometric and
physical performance variables selected for the preparation of the reference equation, respec-
tively. The univariate analysis showed that the VO2 peak correlated significantly with age
(r = 0.38, p = 0.04), and distance (r = 0.67, p = 0.0001). There was no significant correlation
with BMI (r = -0.24, p = 0.22). A model of stepwise linear multiple regressions showed that dis-
tance walked explained 42.5% (R2 adjusted = 0.425, p = 0.0001) of the variance in VO2 peak.
The 95% Confidence Interval for unstandardized coefficients were the constants (11.12 to
30.77) and distance (0.01 to 0.03). The reference equation for the VO2 peak in the ISWT was:
VO2 peakðpredictedÞ ¼ 20:94 þ ð0:02 x distance walkedÞ
Validation of the reference equation
The characteristics of the volunteers who attended in the equation validation stage were pres-
ent in Table 1. The results obtained by the equation of VO2 peak with the values obtained by
the gas analyzer, showed no significant difference between them (VO2 peak [predicted] =
39.24 ± 6.1 mL/kg/min; VO2 peak [gas analyzer] = 40.87 ± 5.4 mL/kg/min, P = 0.1776). It was
possible to verify the agreement between these measures by the Bland-Altman method, in
which a bias of 1.6 was showed, representing a difference of 4.4% in the VO2 peak (Fig 2). Fur-
thermore, there was no statistically significant difference between the participants of equation
elaboration and validation for age (p = 0.7978), weight (p = 0.5498), height (p = 0.0650), BMI
(p = 0.2480), distance walked (p = 0.9213) and walking speed (p = 0.0.6212).
Discussion
The present study describes the comparison of CRF between the ISWT with CEPT in healthy
sedentary adolescent boys. In the ISWT, the adolescent boys reached values of HRmax > 90%
and R peak > 1.1, thus classifying the ISWT as a maximal effort test [19,20]. Furthermore,
results showed a moderate and significant correlations as well as agreement between VO2 peak
and R peak by both tests. Our results are corroborated by the results of Lanza et al. [9]. These
authors showed that ISWT is a maximal test in children and adolescent by registering higher
Fig 2. Bland-Altman agreement of VO2 peak in the validation of the reference equation.
https://doi.org/10.1371/journal.pone.0206867.g002
Incremental shuttle walking test in healthy sedentary adolescent boys
PLOS ONE | https://doi.org/10.1371/journal.pone.0206867
November 1, 2018
7 / 11
HR values (>90%) at end of the test. Our research group also showed previously in healthy
men [2] and women (data not published) that cardiorespiratory outcomes (VO2 peak and R
peak) collected during ISWT are comparable to CEPT test, as well as both tests showed agree-
ment and high correlations between VO2 and R peak between ISTW and CEPT [2]. In the
other hand, previous studies showed lower HR max in healthy control children at end of
the test compared to our data, such as 69% [11] and 55% [10]. However, it’s important to high-
light that our participants were allowed to run which can explain the higher HRmax found. To
the best of our knowledge there are no studies evaluating CRF between ISWT and CEPT in
healthy sedentary adolescent boys. Taken together, our data support that ISWT can be consid-
ered as a valid measure to assess CRF in this population as a maximal effort test.
Additionally, our study is the first one to develop an equation to predict VO2 peak in the
ISWT in this population. Despite of have including anthropometric variables in the multivari-
ate analysis, only distance walked explained the variance of VO2 (43%) peak in our population.
Distance walked as one of the major determinant of VO2 peak was also observed in previous
studies that developed reference equations for the prediction of VO2 peak during the applica-
tion of the ISWT in healthy adults [21,22] and during the six-minute walking test in obese ado-
lescents [23].
Although the age was significantly correlated to VO2 in the linear analysis, this correlation
was not strong the sufficient for explained the variance of VO2 peak. Similar result was
observed by Tsiaras et al. (2010), which shown that the addiction of age did not further
improve the prediction accuracy of the equation for prediction of VO2 peak from a maximal
treadmill test in 12–18 year-old active male adolescents [24]. This absence of influence of age
seems to be related to the stabilization of aerobic performance in youth when compared to
childhood. In fact, previous studies showed that the performance of adolescents improved lin-
early with increase of age, it increased up to 12–13 years, and after (aged 14–19 years) tended
to achieve a plateau [25,26]. As with age, BMI did not influence the prediction of VO2 peak.
The probable reason for this seems to be related to homogeneity of sample of present study. It
is noted that participants of present study showed normal BMI. Thus, given that the CRF is
lower in adolescents who are overweight than in those of normal weight, the normal body
composition did not was correlated to VO2 peak [25,27]. Finally, it’s important to highlight
that distance walked is a feasible variable in clinical practice and have to take into account
when developing a regression equation [28].
Although the prediction equation proposed in the present study might be explained by
moderate variance, the VO2 peak data collected by the gas analyzer and the developed equation
showed agreement. Moreover, the reference values from the current literature that classify
CRF (i.e. very week to excellent) vary approximately 7mL/kg/min among the age ranges. Thus,
the variation found in the present study (4%) is less likely to change the individuals CRF classi-
fication [17]. Finally, the VO2 peak mean reached by the male adolescents in our study (~ 44.0
mL/kg/min) was smaller than previous study reporting VO2 reference for trained men with
age ranging from 15 to 24 (53.3 mL/kg/min) [17] or 10–14 years old ( 52.3 mL/kg/min) [6],
which classifies our population as sedentary [17].
The results pointed here raise important advancing scientific knowledge regarding the level
of ISW in healthy sedentary adolescent boys. The results found in this study contribute to the
process of measurement of peak VO2 becomes more accessible to clinical practice so that the
prescription and elaboration of exercise programs happen in a more informed and assertive
way, as well as ISWT can be used as a maximal effort test in replacement of submaximal
field test available (e.g. six minute walking test) [29]. Moreover, clinicians should considerer
ISWT instead of other field test (e.g. 9-minute walk / run test, 1-mile walk / run test and
the 20 m Shuttle Run Test) [30–32] because these tests are influenced by external factors
Incremental shuttle walking test in healthy sedentary adolescent boys
PLOS ONE | https://doi.org/10.1371/journal.pone.0206867
November 1, 2018
8 / 11
(e.g. motivation and self-paced) which can lead to great variability and compromising the
application in randomized controlled trials. Lastly, our prediction equation could be used in
clinical studies aiming to investigate CRF in disable adolescent boys population avoiding to
use control groups for comparing theirs results [5,9,10]. However, due to restrictions of fund-
ing and time, no further experiments such as assessing girls and children with age under 12
years old were conducted. Further studies are necessary to address this population.
Conclusion
In a conclusive way, the VO2 peak values found in our study allow us to affirm that the ISWT
was in fact a maximum intensity test in healthy sedentary adolescent boys assessed by direct
gas analyzer. Furthermore, the regression equation was feasible and might be useful for clini-
cians for predicting VO2 peak in this population.
Author Contributions
Conceptualization: Andreza L. Gomes, Vanessa A. Mendonc¸a, Alcilene M. Silva, Ana Cristina
R. Camargos, Camila D. C. Neves, Ana C. R. Lacerda, He´rcules R. Leite.
Data curation: Tatiane dos Santos Silva, Crislaine K. V. Pires, Ana Cristina R. Camargos.
Formal analysis: Andreza L. Gomes, Vanessa A. Mendonc¸a, Ana Cristina R. Camargos, He´r-
cules R. Leite.
Funding acquisition: Vanessa A. Mendonc¸a, He´rcules R. Leite.
Investigation: Vanessa A. Mendonc¸a, Crislaine K. V. Pires, Liliana P. Lima, He´rcules R. Leite.
Methodology: Andreza L. Gomes, Tatiane dos Santos Silva, Ana Cristina R. Camargos, He´rcu-
les R. Leite.
Project administration: Ana C. R. Lacerda, He´rcules R. Leite.
Resources: He´rcules R. Leite.
Supervision: Ana C. R. Lacerda, He´rcules R. Leite.
Validation: He´rcules R. Leite.
Writing – original draft: He´rcules R. Leite.
Writing – review & editing: Vanessa A. Mendonc¸a, Liliana P. Lima, Camila D. C. Neves, He´r-
cules R. Leite.
References
1.
Haskell WL, Lee I-M, Pate RR, Powell KE, Blair SN, et al. (2007) Physical activity and public health:
updated recommendation for adults from the American College of Sports Medicine and the American
Heart Association. Circulation 116: 1081. https://doi.org/10.1161/CIRCULATIONAHA.107.185649
PMID: 17671237
2.
Neves CD, Lacerda ACR, Lage VK, Lima LP, Fonseca SF, et al. (2015) Cardiorespiratory responses
and prediction of peak oxygen uptake during the shuttle walking test in healthy sedentary adult men.
PloS one 10: e0117563. https://doi.org/10.1371/journal.pone.0117563 PMID: 25659094
3.
Nici L, Donner C, Wouters E, Zuwallack R, Ambrosino N, et al. (2006) American thoracic society/Euro-
pean respiratory society statement on pulmonary rehabilitation. American journal of respiratory and criti-
cal care medicine 173: 1390–1413. https://doi.org/10.1164/rccm.200508-1211ST PMID: 16760357
4.
Villa F, Castro APBM, Pastorino AC, Santare´m JM, Martins MA, et al. (2011) Aerobic capacity and skel-
etal muscle function in children with asthma. Archives of disease in childhood: archdischild212431.
Incremental shuttle walking test in healthy sedentary adolescent boys
PLOS ONE | https://doi.org/10.1371/journal.pone.0206867
November 1, 2018
9 / 11
5.
Sperandio EF, Vidotto MC, Alexandre AS, Yi LC, Gotfryd AO, et al. (2015) Functional exercise capacity,
lung function and chest wall deformity in patients with adolescent idiopathic scoliosis. Fisioterapia em
Movimento 28: 563–572.
6.
Rodrigues AN, Perez AJ, Carletti L, Bissoli NS, Abreu GR (2006) Maximum oxygen uptake in adoles-
cents as measured by cardiopulmonary exercise testing: a classification proposal. Jornal de Pediatria
82: 426–430. https://doi.org/10.2223/JPED.1533 PMID: 17003945
7.
Probst VS, Hernandes NA, Teixeira DC, Felcar JM, Mesquita RB, et al. (2012) Reference values for the
incremental shuttle walking test. Respiratory medicine 106: 243–248. https://doi.org/10.1016/j.rmed.
2011.07.023 PMID: 21865021
8.
Singh SJ, Morgan M, Scott S, Walters D, Hardman AE (1992) Development of a shuttle walking test of
disability in patients with chronic airways obstruction. Thorax 47: 1019–1024. PMID: 1494764
9.
de Cordoba Lanza F, do Prado Zagatto E, Silva JC, Selman JPR, Imperatori TBG, et al. (2015) Refer-
ence equation for the incremental shuttle walk test in children and adolescents. The Journal of pediat-
rics 167: 1057–1061. https://doi.org/10.1016/j.jpeds.2015.07.068 PMID: 26323195
10.
Tsopanoglou SP, Davidson J, Goulart AL, de Moraes Barros MC, dos Santos AMN (2014) Functional
capacity during exercise in very-low-birth-weight premature children. Pediatric pulmonology 49: 91–98.
https://doi.org/10.1002/ppul.22754 PMID: 23359551
11.
Gomes E´ LD, Sampaio LMM, Costa IP, Dias FD, Ferneda VS, et al. (2013) Analysis of autonomic modu-
lation during maximal and submaximal work rate and functional capacity in asthmatic children. Journal
of Asthma 50: 613–618. https://doi.org/10.3109/02770903.2013.793707 PMID: 23574110
12.
Coelho CC, Aquino EdS, Almeida DCd, Oliveira GC, Pinto RdC, et al. (2007) Comparative analysis and
reproducibility of the modified shuttle walk test in normal children and in children with cystic fibrosis. Jor-
nal Brasileiro de Pneumologia 33: 168–174. PMID: 17724536
13.
Dourado VZ, Guerra RLF (2013) Reliability and validity of heart rate variability threshold
assessment during an incremental shuttle-walk test in middle-aged and older adults. Brazilian
Journal of Medical and Biological Research 46: 194–199. https://doi.org/10.1590/1414-431X20122376
PMID: 23369974
14.
Bradley J, Howard J, Wallace E, Elborn S (2000) Reliability, repeatability, and sensitivity of the modified
shuttle test in adult cystic fibrosis. Chest 117: 1666–1671. PMID: 10858400
15.
American College of Sports Medicine (2003) Diretrizes do ACSM para os testes de esforc¸o e sua pre-
scric¸ão. Rio de Janeiro: Guanabara Koogan. 704 p. 25057689.
16.
Ju¨rgensen SP, de Oliveira Antunes LC, Tanni SE, Banov MC, Lucheta PA, et al. (2011) The incremental
shuttle walk test in older Brazilian adults. Respiration 81: 223–228. https://doi.org/10.1159/000319037
PMID: 20639622
17.
Herdy AH, Caixeta A (2016) Brazilian cardiorespiratory fitness classification based on maximum oxygen
consumption. Arquivos brasileiros de cardiologia 106: 389–395. https://doi.org/10.5935/abc.20160070
PMID: 27305285
18.
Tanaka H, Monahan KD, Seals DR (2001) Age-predicted maximal heart rate revisited. Journal of Ameri-
can College of Cardiology 37: 153–6.
19.
Robergs RA, Dwyer D, Astorino T (2010) Recommendations for improved data processing from expired
gas analysis indirect calorimetry. Sports Medicine 40: 95–111. https://doi.org/10.2165/11319670-
000000000-00000 PMID: 20092364
20.
Sawyer BJ, Blessinger JR, Irving BA, Weltman A, Patrie JT, et al. (2010) Walking and running economy:
inverse association with peak oxygen uptake. Medicine and science in sports and exercise 42: 2122.
https://doi.org/10.1249/MSS.0b013e3181de2da7 PMID: 20351592
21.
Dourado VZ, Banov M, Marino M, De Souza V, Antunes LdO, et al. (2010) A simple approach to assess
VT during a field walk test. International journal of sports medicine 31: 698–703. https://doi.org/10.
1055/s-0030-1255110 PMID: 20617483
22.
Dourado VZ, Guerra RLF, Tanni SE, Antunes LCdO, Godoy I (2013) Reference values for the incre-
mental shuttle walk test in healthy subjects: from the walk distance to physiological responses. Jornal
Brasileiro de Pneumologia 39: 190–197. https://doi.org/10.1590/S1806-37132013000200010 PMID:
23670504
23.
Vanhelst J, Fardy PS, Salleron J, Be´ghin L (2013) The six-minute walk test in obese youth: reproducibil-
ity, validity, and prediction equation to assess aerobic power. Disability and rehabilitation 35: 479–482.
https://doi.org/10.3109/09638288.2012.699581 PMID: 22779759
24.
Tsiaras V, Zafeiridis A, Dipla K, Patras K, Georgoulis A, et al. (2010) Prediction of peak oxygen uptake
from a maximal treadmill test in 12-to 18-year-old active male adolescents. Pediatric exercise science
22: 624–637. PMID: 21242610
Incremental shuttle walking test in healthy sedentary adolescent boys
PLOS ONE | https://doi.org/10.1371/journal.pone.0206867
November 1, 2018
10 / 11
25.
Lipman TH (2007) Cardiorespiratory Fitness Levels Among US Youth 12 to 19 Years of Age: Findings
From the 1999–2002 National Health and Nutrition Examination Survey. MCN: The American Journal
of Maternal/Child Nursing 32: 197.
26.
Olds T, Tomkinson G, Le´ger L, Cazorla G (2006) Worldwide variation in the performance of children
and adolescents: an analysis of 109 studies of the 20-m shuttle run test in 37 countries. Journal of sports
sciences 24: 1025–1038. https://doi.org/10.1080/02640410500432193 PMID: 17115514
27.
Castro-Piñero J, Ortega F, Keating XD, Gonza´lez-Montesinos J, Sjo¨strom M, et al. (2011) Percentile
values for aerobic performance running/walking field tests in children aged 6 to 17 years; influence of
weight status. Nutricio´n hospitalaria 26.
28.
Salbach NM, O’Brien KK, Brooks D, Irvin E, Martino R, et al. (2015) Reference values for standardized
tests of walking speed and distance: a systematic review. Gait & posture 41: 341–360.
29.
Cacau LdAP, Carvalho VO, dos Santos Pin A, Daniel CRA, Ykeda DS, et al. (2017) Reference Values
for the 6-min Walk Distance (6MWT) in Healthy Children Aged 7 to 12 Years in Brazil: Main Results of
the TC6minBRASIL Multi-Center Study. Respiratory care: respcare. 05686.
30.
Leger LA, Lambert J (1982) A maximal multistage 20-m shuttle run test to predict $ $\dot V $ $ O2 max.
European journal of applied physiology and occupational physiology 49: 1–12. PMID: 7201922
31.
Paludo AC, Batista MB, Serassuelo Ju´nior H, Cyrino ES, Ronque ERV (2012) Estimation of cardiorespi-
ratory fitness in adolescents with the 9-minute run/walk test. Revista Brasileira de Cineantropometria &
Desempenho Humano 14: 401–408.
32.
Cureton KJ, Sloniger MA, O’Bannon JP, Black DM, McCormack WP (1995) A generalized equation for
prediction of VO2peak from 1-mile run/walk performance. Med Sci Sports Exerc 27: 445–451. PMID:
7752874
Incremental shuttle walking test in healthy sedentary adolescent boys
PLOS ONE | https://doi.org/10.1371/journal.pone.0206867
November 1, 2018
11 / 11
| Cardiorespiratory and metabolic responses and reference equation validation to predict peak oxygen uptake for the incremental shuttle waking test in adolescent boys. | 11-01-2018 | Gomes, Andreza L,Mendonça, Vanessa A,Dos Santos Silva, Tatiane,Pires, Crislaine K V,Lima, Liliana P,Gomes, Alcilene M,Camargos, Ana Cristina R,Neves, Camila D C,Lacerda, Ana C R,Leite, Hércules R | eng |
PMC9653753 | Citation: Jost, Z.; Tomczyk, M.;
Chroboczek, M.; Calder, P.C.;
Laskowski, R. Improved Oxygen
Uptake Efficiency Parameters Are
Not Correlated with VO2peak or
Running Economy and Are Not
Affected by Omega-3 Fatty Acid
Supplementation in Endurance
Runners. Int. J. Environ. Res. Public
Health 2022, 19, 14043. https://doi.
org/10.3390/ijerph192114043
Academic Editors: Nicolas Berger
and Russ Best
Received: 14 October 2022
Accepted: 25 October 2022
Published: 28 October 2022
Publisher’s Note: MDPI stays neutral
with regard to jurisdictional claims in
published maps and institutional affil-
iations.
Copyright:
© 2022 by the authors.
Licensee MDPI, Basel, Switzerland.
This article is an open access article
distributed
under
the
terms
and
conditions of the Creative Commons
Attribution (CC BY) license (https://
creativecommons.org/licenses/by/
4.0/).
International Journal of
Environmental Research
and Public Health
Article
Improved Oxygen Uptake Efficiency Parameters Are Not
Correlated with VO2peak or Running Economy and Are Not
Affected by Omega-3 Fatty Acid Supplementation in
Endurance Runners
Zbigniew Jost 1,*
, Maja Tomczyk 1, Maciej Chroboczek 2, Philip C. Calder 3,4
and Radosław Laskowski 2,*
1
Department of Biochemistry, Gdansk University of Physical Education and Sport, 80-336 Gdansk, Poland
2
Department of Physiology, Gdansk University of Physical Education and Sport, 80-336 Gdansk, Poland
3
Faculty of Medicine, School of Human Development and Health, University of Southampton,
Southampton SO16 6YD, UK
4
NIHR Southampton Biomedical Research Centre, University Hospital Southampton NHS Foundation Trust
and University of Southampton, Southampton SO16 6YD, UK
*
Correspondence: [email protected] (Z.J.); [email protected] (R.L.)
Abstract: Peak oxygen uptake (VO2peak) is one of the most reliable parameters of exercise capacity;
however, maximum effort is required to achieve this. Therefore, alternative, and repeatable sub-
maximal parameters, such as running economy (RE), are needed. Thus, we evaluated the suitability
of oxygen uptake efficiency (OUE), oxygen uptake efficiency plateau (OUEP) and oxygen uptake
efficiency at the ventilatory anaerobic threshold (OUE@VAT) as alternatives for VO2peak and RE.
Moreover, we evaluated how these parameters are affected by endurance training and supplementa-
tion with omega-3 fatty acids. A total of 26 amateur male runners completed a 12-week endurance
program combined with omega-3 fatty acid supplementation or medium-chain triglycerides as a
placebo. Before and after the intervention, the participants were subjected to a treadmill test to
determine VO2peak, RE, OUE, OUEP and OUE@VAT. Blood was collected at the same timepoints to
determine eicosapentaenoic acid (EPA) and docosahexaenoic acid (DHA) in erythrocytes. OUE corre-
lated moderately or weakly with VO2peak (R2 = 0.338, p = 0.002) and (R2 = 0.226, p = 0.014) before and
after the intervention, respectively. There was a weak or no correlation between OUEP, OUE@VAT,
VO2peak and RE despite steeper OUE, increased OUEP and OUE@VAT values in all participants.
OUE parameters cannot be treated as alternative parameters for VO2peak or RE and did not show
changes following supplementation with omega-3 fatty acids in male amateur endurance runners.
Keywords: peak oxygen uptake; oxygen uptake efficiency plateau; running economy; omega-3 fatty
acids; endurance runners
1. Introduction
There are many cardiopulmonary exercise tests (CPETs) that aim to assess parameters
related to human performance, such as peak oxygen uptake (VO2peak) or maximal oxygen
uptake (VO2max). VO2max is considered the best indicator of potential in endurance events,
being a ‘gold standard’ measurement of integrated cardiopulmonary-muscle oxidative
function [1–3]. Although heart rate (HR), respiratory exchange ratio (RER), and minute
ventilation (Ve) are considered cardiovascular, respiratory, and pulmonary parameters,
respectively, their comprehensive function is often difficult to evaluate. Therefore, there is
a need to identify alternative validated and reliable parameters for assessing cardiorespira-
tory fitness.
Sun and co-authors [4] determined the relationship between oxygen uptake (VO2)
and Ve, called oxygen uptake efficiency (OUE). They noted that OUE increases linearly
with time during early exercise, but becomes non-linear as Ve increases faster than VO2.
Int. J. Environ. Res. Public Health 2022, 19, 14043. https://doi.org/10.3390/ijerph192114043
https://www.mdpi.com/journal/ijerph
Int. J. Environ. Res. Public Health 2022, 19, 14043
2 of 10
This curvilinear relationship during an exercise test is not as appropriate for assessing
aerobic capacity as VO2peak. Thus, the authors described other physiological parameters
that can be determined from respiratory gases during CPET, i.e., oxygen uptake efficiency
at the ventilatory anaerobic threshold (OUE@VAT) and oxygen uptake efficiency plateau
(OUEP) in healthy subjects. It was observed that both OUE@VAT and OUEP are simple
measurements that do not require maximum effort. Moreover, they are also easy to visualise,
recognise and calculate [4], making them potentially robust parameters for assessing
physical fitness. It is worth noting that there is still scarce evidence of improvements in OUE
parameters after physical training and no evidence of improvements after supplementation
with bioactive compounds such as the omega-3 fatty acids (eicosapentaenoic acid (EPA)
and docosahexaenoic acid (DHA)).
Although some studies show no improvements in cardiopulmonary-muscle oxida-
tive function following supplementation with fish oil containing omega-3 fatty acids [5,6],
several studies do indicate a positive effect. For example, long-term EPA and DHA supple-
mentation may contribute to the improvement in VO2max [7] or to the reduction in the cost
of aerobic exercise in trained cyclists [8,9]. Moreover, our recent study showed that 12-week
supplementation with omega-3 fatty acids improved running economy (RE) in amateur
runners [10]. These studies focus mainly on VO2max and RE, and other submaximal oxygen
kinetics parameters need to be further explored.
The aim of our study was to determine whether OUE, OUEP and OUE@VAT can
be considered as a robust measurements of endurance capacity. Moreover, we verify if
those parameters are sensitive to changes after omega-3 fatty acid supplementation. The
main hypothesis of this research was that OUE will be sensitive to changes in VO2peak. We
also hypothesised that OUE@VAT and OUEP can be used as non-invasive, submaximal
parameters of oxygen kinetics replacing VO2peak and RE. We also evaluated whether twelve-
week of endurance training combined with omega-3 fatty acid supplementation can alter
these parameters in male amateur endurance runners.
2. Materials and Methods
2.1. Participants
A total of 26 male amateur runners (37 ± 3 years old; 77 ± 9 kg body weight; VO2peak:
54.2 ± 6 mL*kg−1*min−1) completed the 12-week experimental study as previously de-
scribed [10], which tested the effect of supplementation with omega-3 fatty acids on exercise
capacity in male amateur endurance runners. Participants were not taking medication and
all were in good health, as confirmed by a medical check. The study was approved by
the Bioethical Committee of Regional Medical Society in Gda´nsk (NKBBN/628/2019) and
conducted according to the Declaration of Helsinki (2013). All participants provided their
written informed consent prior participating in the study. Detailed participant characteris-
tics and project design are shown in Table 1 and Figure 1, respectively.
Table 1. Characteristics of participants.
Variable
MCT
(n = 12)
Mean ± SD
OMEGA
(n = 14)
Mean ± SD
Age [y]
37 ± 4
37 ± 3
Body mass [kg]
78.0 ± 8
76.3 ± 11
Height [cm]
180 ± 4
181 ± 7
EPA [% of total erythrocyte fatty acids]
Pre
1.2 ± 0.3
1.1 ± 0.4
Post
1.2 ± 0.3
4.9 ± 1.1 *,ˆ
DHA [% of total erythrocyte fatty acids]
Pre
4.4 ± 1.1
4.7 ± 1.0
Post
4.5 ± 0.8
6.7 ± 0.8 *,ˆ
Int. J. Environ. Res. Public Health 2022, 19, 14043
3 of 10
Table 1. Cont.
Variable
MCT
(n = 12)
Mean ± SD
OMEGA
(n = 14)
Mean ± SD
HRmax [beats*min−1]
Pre
186 ± 9
190 ± 9
Post
184 ± 7
189 ± 9
VO2peak [mL*kg−1*min−1]
Pre
54.7 ± 6.8
53.6 ± 4.4
Post
56.4 ± 5.9
56.0 ± 3.7 *
RE [mL*kg−1*min−1]
Pre
47.7 ± 3.3
47.6 ± 1.8
Post
48.7 ± 2.9
46.5 ± 2.4 ˆ
EPA—eicosapentaenoic acid; DHA—docosahexaenoic acid; HRmax—maximal heart rate; RE—running economy;
data are presented as mean ± SD; * p < 0.05 for post vs. pre value ˆ p < 0.05 for MCT vs. OMEGA.
Int. J. Environ. Res. Public Health 2022, 19, x FOR PEER REVIEW
3 of 10
Post
1.2 ± 0.3
4.9 ± 1.1 *,^
DHA [% of total erythrocyte fatty acids]
Pre
4.4 ± 1.1
4.7 ± 1.0
Post
4.5 ± 0.8
6.7 ± 0.8 *,^
HRmax [beats*min−1]
Pre
186 ± 9
190 ± 9
Post
184 ± 7
189 ± 9
VO2peak [mL*kg−1*min−1]
Pre
54.7 ± 6.8
53.6 ± 4.4
Post
56.4 ± 5.9
56.0 ± 3.7 *
RE [mL*kg−1*min−1]
Pre
47.7 ± 3.3
47.6 ± 1.8
Post
48.7 ± 2.9
46.5 ± 2.4 ^
EPA—eicosapentaenoic acid; DHA—docosahexaenoic acid; HRmax—maximal heart rate; RE—run-
ning economy; data are presented as mean ± SD; * p < 0.05 for post vs. pre value ^ p < 0.05 for MCT
vs. OMEGA.
Figure 1. Procedure design.
2.2. Supplementation
Participants were randomly assigned to one of two groups with the final character-
istics as follows: OMEGA (37 ± 3 yr; 76.3 ± 11 kg body weight; VO2peak: 53.6 ± 4
mL*kg−1*min−1) or medium-chain triglycerides (MCT) as placebo (37 ± 4 yr; 78 ± 8 kg body
weight; VO2peak: 54.7 ± 7 mL*kg−1*min−1). The division of participants into two groups was
performed to check the difference in the response of OUE parameters to supplementation.
Hence, participants supplemented four capsules per day, providing a total of 2234 mg of
EPA + 916 mg of DHA (OMEGA group) or 4000 mg of MCT (MCT group). The capsules
were provided in coded, identical-looking packages to avoid a potential recognition. To
maintain the quality of supplements consisting of omega-3 fatty acids and their respective
dosages, materials adhering to the International Fish Oil Standard (IFOS) were used.
2.3. Treadmill Exercise Testing
Exercise tests were conducted under controlled environmental conditions (18–20 °C
and humidity 40–45%) and were performed at similar time of day ± 2 h. Before carrying
out the exercise tests, the participants performed a familiarization trial. The participants
were informed to refrain from strenuous exercise for 24 h and from caffeine or alcohol
consumption for 12 h prior to the tests. Before and after twelve weeks of the training pro-
gram, participants undertook a ramp exercise test to volitional exhaustion on a treadmill
(h/p Cosmos, Saturn, Nussdorf-Traunstein, Germany). First, participants stood on the
treadmill for 2 min to make sure the measuring equipment was ready and to measure the
resting parameters. Thereafter, runners walked for 5 min at 5 km/h speed and with a 1.5%
inclination as a warm-up prior to starting the test. Every next stage lasted 3 min, and the
Figure 1. Procedure design.
2.2. Supplementation
Participants were randomly assigned to one of two groups with the final characteristics as
follows: OMEGA (37 ± 3 years; 76.3 ± 11 kg body weight; VO2peak: 53.6 ± 4 mL*kg−1*min−1)
or medium-chain triglycerides (MCT) as placebo (37 ± 4 years; 78 ± 8 kg body weight;
VO2peak: 54.7 ± 7 mL*kg−1*min−1). The division of participants into two groups was per-
formed to check the difference in the response of OUE parameters to supplementation. Hence,
participants supplemented four capsules per day, providing a total of 2234 mg of EPA +
916 mg of DHA (OMEGA group) or 4000 mg of MCT (MCT group). The capsules were
provided in coded, identical-looking packages to avoid a potential recognition. To maintain
the quality of supplements consisting of omega-3 fatty acids and their respective dosages,
materials adhering to the International Fish Oil Standard (IFOS) were used.
2.3. Treadmill Exercise Testing
Exercise tests were conducted under controlled environmental conditions (18–20 ◦C
and humidity 40–45%) and were performed at similar time of day ± 2 h. Before carrying
out the exercise tests, the participants performed a familiarization trial. The participants
were informed to refrain from strenuous exercise for 24 h and from caffeine or alcohol
consumption for 12 h prior to the tests. Before and after twelve weeks of the training
program, participants undertook a ramp exercise test to volitional exhaustion on a treadmill
(h/p Cosmos, Saturn, Nussdorf-Traunstein, Germany). First, participants stood on the
treadmill for 2 min to make sure the measuring equipment was ready and to measure the
resting parameters. Thereafter, runners walked for 5 min at 5 km/h speed and with a 1.5%
inclination as a warm-up prior to starting the test. Every next stage lasted 3 min, and the
treadmill belt was accelerated starting from 8 km/h by 1 km/h per stage up to 12 km/h.
Int. J. Environ. Res. Public Health 2022, 19, 14043
4 of 10
Then, the inclination of the treadmill was increased to 5%, 10% and 15% at 12 km/h
speed until volitional exhaustion, despite strong verbal encouragement. During both
tests, heart rate (HR) was monitored (Polar RS400, Kempele, Finland). RE was measured
as an oxygen cost from last 50 s of each stage to 12 km/h speed and was expressed
as mL*kg−1*min−1 [11].
2.4. Respiratory Gas Measurements
During both laboratory tests, the exhaled air was continuously measured using a
breath-by-breath analyser (Oxycon Pro, Jaeger, Hoechberg, Germany). Before the tests, the
analyser was calibrated in accordance with the manufacturer’s instructions. All measure-
ments were averaged to 10 s intervals and included: oxygen uptake (VO2), carbon dioxide
output (VCO2), minute ventilation (Ve), end-tidal pressure of oxygen (PETO2) and end-tidal
pressure of carbon dioxide (PETCO2).
2.5. Determination of Oxygen Uptake Efficiency and Ventilatory Thresholds
The OUE was individually determined for each participant by calculating the regres-
sion slope from the linear relationship of absolute VO2 (mL*min−1) plotted as a linear
function of Ve (L*min−1) (VO2 = Ve + b), as previously described by Sun et al. [4]. Af-
ter calculating the OUE individually for each participant from the formula, the OUE
was correlated with the true VO2peak and normalized, and the original OUE values (“b”)
were compared for the slope of the linear regression of the oxygen uptake efficiency.
OUEP was calculated as the 90 s average of the highest consecutive measurements of VO2
(mL*min−1)/Ve (L*min−1) and OUE at the ventilatory anaerobic threshold (OUE@VAT),
as the 60 s average of consecutive measurements at and immediately before the VAT ac-
cordingly to Sun et al. [4]. First, ventilatory threshold (VT1) was determined as increase
in both the ventilatory equivalent of oxygen (Ve/VO2) and end-tidal pressure of oxygen
(PETO2) with no concomitant increase in the ventilatory equivalent of carbon dioxide
(Ve/VCO2) [12]. The ventilatory anaerobic threshold (VAT) was measured by the V-slope
method [13]. Peak oxygen uptake (VO2peak) was obtained as the last 30 s oxygen uptake
mean value recorded during the test [14].
2.6. Training Program
All participants underwent 12 weeks of an endurance training program. The partici-
pants performed endurance training of varying intensity three times a week according to
Costa et al. [15] with slight modifications. Additionally, participants performed training
once a week, which aimed to strengthen the central stabilization muscles and to reduce
the risk of injury [16]. The training intensity was distributed among 3 heart-rate zones
(Z1-Z2-Z3). They were determined according to the first ventilatory threshold (VT1), ven-
tilatory anaerobic threshold (VAT) and the corresponding values of the heart rate [Z1:
≤HR@VT1 + 5 bpm; Z2: (>HR@VT1 + 5 bpm) to (≤HR@VAT-5 bpm); Z3: >HR@VAT-5
bpm]. Average training times spent in every mesocycle were (~80%-15%-5%) in zones
(Z1-Z2-Z3), respectively. In the last week, the training volume was reduced to reduce the
accumulated fatigue. All trainings were monitored by Polar M430 (Kempele, Finland)
wrist watches and H9 heart-rate chest sensor and the supervision over the participants was
carried out by a certified track and field coach.
2.7. Erythrocyte Fatty Acid Analysis
Sample collection and fatty acid determination were outlined elsewhere [10]. In brief,
blood samples were collected into 4 mL sodium citrate vacutainer tubes and centrifuged at
4 ◦C (4000× g for 10 min). After centrifugation, plasma was collected with a disposable
Pasteur pipette, transferred into separate Eppendorf probes and stored in a −80 ◦C freezer
until further analysis. Erythrocyte lipids were extracted into chloroform:methanol and fatty
acid methyl esters (representing the erythrocyte fatty acids) were formed by heating the
lipid extract with methanolic sulphuric acid. The fatty acid methyl esters were separated
Int. J. Environ. Res. Public Health 2022, 19, 14043
5 of 10
by gas chromatography on a Hewlett Packard 6890 gas chromatograph fitted with a BPX-70
column using the settings and run conditions described by Fisk et al. [17]. Fatty acid methyl
esters were identified by comparison with runtimes of authentic standards and data were
expressed as weight % of total fatty acids.
2.8. Statistical Analysis
Statistical analysis was performed using GraphPad Prism 7 (San Diego, CA, USA).
Arithmetic means, standard deviation (SD), and significance levels of differences between
means were calculated. Two-way analysis of variance (ANOVA), with repeated measures,
was used to investigate the significance of differences between groups and time. Significant
main effects were further analyzed using the Bonferroni corrected post hoc test. Correlations
between variables were evaluated using the Pearson and Spearman correlations coefficients.
All analyses used a significance level of p < 0.05.
3. Results
3.1. Predicted VO2peak from OUE Equation
Predicted VO2peak calculated from the OUE formula both before and after the supple-
mentation intervention was moderately correlated with peak oxygen uptake (R2 = 0.338,
p = 0.002; Figure 2A) for all participants before the study. Moreover, the results without
grouping also showed a correlation after 12 weeks of intervention (R2 = 0.226, p = 0.014;
Figure 2B), but the correlation was weak.
Int. J. Environ. Res. Public Health 2022, 19, x FOR PEER REVIEW
5 of 10
fatty acid methyl esters (representing the erythrocyte fatty acids) were formed by heating
the lipid extract with methanolic sulphuric acid. The fatty acid methyl esters were sepa-
rated by gas chromatography on a Hewlett Packard 6890 gas chromatograph fitted with
a BPX-70 column using the settings and run conditions described by Fisk et al. [17]. Fatty
acid methyl esters were identified by comparison with runtimes of authentic standards
and data were expressed as weight % of total fatty acids.
2.8. Statistical Analysis
Statistical analysis was performed using GraphPad Prism 7 (San Diego, CA, USA).
Arithmetic means, standard deviation (SD), and significance levels of differences between
means were calculated. Two-way analysis of variance (ANOVA), with repeated measures,
was used to investigate the significance of differences between groups and time. Signifi-
cant main effects were further analyzed using the Bonferroni corrected post hoc test. Cor-
relations between variables were evaluated using the Pearson and Spearman correlations
coefficients. All analyses used a significance level of p < 0.05.
3. Results
3.1. Predicted VO2peak from OUE Equation
Predicted VO2peak calculated from the OUE formula both before and after the supple-
mentation intervention was moderately correlated with peak oxygen uptake (R2 = 0.338, p
= 0.002; Figure 2A) for all participants before the study. Moreover, the results without
grouping also showed a correlation after 12 weeks of intervention (R2 = 0.226, p = 0.014;
Figure 2B), but the correlation was weak.
Figure 2. The linear relationship between VO2peak and predicted VO2peak before (A) and after (B)
twelve weeks of combined endurance training and supplementation (OMEGA and MCT groups; n
= 26).
3.2. Oxygen Uptake Efficiency Plateau
Pre-intervention OUEP values weakly correlated with VO2peak (R2 = 0.247, p = 0.01;
Figure 3A). After twelve weeks of intervention, no correlation was found between these
two indicators (R2 = 0.077, p = 0.17, Figure 3B).
Figure 2.
The linear relationship between VO2peak and predicted VO2peak before (A) and af-
ter (B) twelve weeks of combined endurance training and supplementation (OMEGA and MCT
groups; n = 26).
3.2. Oxygen Uptake Efficiency Plateau
Pre-intervention OUEP values weakly correlated with VO2peak (R2 = 0.247, p = 0.01;
Figure 3A). After twelve weeks of intervention, no correlation was found between these
two indicators (R2 = 0.077, p = 0.17, Figure 3B).
Int. J. Environ. Res. Public Health 2022, 19, x FOR PEER REVIEW
6 of 10
Figure 3. The linear relationship between VO2peak and OUEP before (A) and after (B) twelve weeks
of combined endurance training and supplementation (OMEGA and MCT groups; n = 26).
3.3. Oxygen Uptake Efficiency at the Ventilatory Anaerobic Threshold
OUE@VAT poorly correlated with the peak oxygen uptake (VO2peak) before the study
(R2 = 0.179, p = 0.031, Figure 4A) and there was no correlation after the 12-week interven-
tion (R2 = 0.082, p = 0.154, Figure 4B) in all participants.
Figure 3. The linear relationship between VO2peak and OUEP before (A) and after (B) twelve weeks
of combined endurance training and supplementation (OMEGA and MCT groups; n = 26).
Int. J. Environ. Res. Public Health 2022, 19, 14043
6 of 10
3.3. Oxygen Uptake Efficiency at the Ventilatory Anaerobic Threshold
OUE@VAT poorly correlated with the peak oxygen uptake (VO2peak) before the study
(R2 = 0.179, p = 0.031, Figure 4A) and there was no correlation after the 12-week intervention
(R2 = 0.082, p = 0.154, Figure 4B) in all participants.
Figure 3. The linear relationship between VO2peak and OUEP before (A) and after (B) twelve weeks
of combined endurance training and supplementation (OMEGA and MCT groups; n = 26).
3.3. Oxygen Uptake Efficiency at the Ventilatory Anaerobic Threshold
OUE@VAT poorly correlated with the peak oxygen uptake (VO2peak) before the study
(R2 = 0.179, p = 0.031, Figure 4A) and there was no correlation after the 12-week interven-
tion (R2 = 0.082, p = 0.154, Figure 4B) in all participants.
Figure 4. The linear relationship between VO2peak and OUE@VAT before (A) and after (B) twelve
weeks of combined endurance training and supplementation (OMEGA and MCT groups; n = 26).
3.4. Correlation between OUEP, OUE@VAT and RE
The changes observed in RE (presented as VO2 delta [%] at 12 km/h) were not corre-
lated with the change in OUEP (R2 = 0.018, p = 0.511; Figure 5A). Similar results were ob-
served in the correlation between RE and OUE@VAT (r = 0.079, p = 0.699; Figure 5B) in all
participants.
Figure 5. Correlation between changes in RE and OUEP (A) and OUE@VAT (B) after twelve weeks
of combined endurance training and supplementation (OMEGA and MCT groups; n = 26).
Figure 4. The linear relationship between VO2peak and OUE@VAT before (A) and after (B) twelve
weeks of combined endurance training and supplementation (OMEGA and MCT groups; n = 26).
3.4. Correlation between OUEP, OUE@VAT and RE
The changes observed in RE (presented as VO2 delta [%] at 12 km/h) were not
correlated with the change in OUEP (R2 = 0.018, p = 0.511; Figure 5A). Similar results were
observed in the correlation between RE and OUE@VAT (r = 0.079, p = 0.699; Figure 5B) in
all participants.
Figure 3. The linear relationship between VO2peak and OUEP before (A) and after (B) twelve weeks
of combined endurance training and supplementation (OMEGA and MCT groups; n = 26).
3.3. Oxygen Uptake Efficiency at the Ventilatory Anaerobic Threshold
OUE@VAT poorly correlated with the peak oxygen uptake (VO2peak) before the study
(R2 = 0.179, p = 0.031, Figure 4A) and there was no correlation after the 12-week interven-
tion (R2 = 0.082, p = 0.154, Figure 4B) in all participants.
Figure 4. The linear relationship between VO2peak and OUE@VAT before (A) and after (B) twelve
weeks of combined endurance training and supplementation (OMEGA and MCT groups; n = 26).
3.4. Correlation between OUEP, OUE@VAT and RE
The changes observed in RE (presented as VO2 delta [%] at 12 km/h) were not corre-
lated with the change in OUEP (R2 = 0.018, p = 0.511; Figure 5A). Similar results were ob-
served in the correlation between RE and OUE@VAT (r = 0.079, p = 0.699; Figure 5B) in all
participants.
Figure 5. Correlation between changes in RE and OUEP (A) and OUE@VAT (B) after twelve weeks
of combined endurance training and supplementation (OMEGA and MCT groups; n = 26).
Figure 5. Correlation between changes in RE and OUEP (A) and OUE@VAT (B) after twelve weeks
of combined endurance training and supplementation (OMEGA and MCT groups; n = 26).
3.5. Omega-3 Fatty Acids Supplementation
Baseline levels of EPA and DHA did not differ between the groups (OMEGA group:
1.1% EPA, 4.7% DHA; MCT group: 1.2% EPA, 4.4% DHA, both p > 0.999). Post-intervention
values of EPA and DHA increased in OMEGA group (4.9% EPA, 6.7% DHA, both p < 0.001).
Changes were not observed in MCT group (1.2% EPA, p > 0.999; 4.7% DHA, p = 0.551). All
results are provided in Table 1.
3.5.1. Oxygen Uptake Efficiency
At the end of the 12-week supplementation period, there was an increase in the slope
of oxygen uptake efficiency in the OMEGA group from 35.4 ± 3.3 to 37.6 ± 3.0 and in the
MCT group from 35.5 ± 3.7 to 37.2 ± 3.1; (both p < 0.001). OUE increased when groups
were combined from 35.5 ± 3.4 to 37.4 ± 3.0; (p < 0.001, Table 2).
Int. J. Environ. Res. Public Health 2022, 19, 14043
7 of 10
Table 2. Comparison of effects omega-3 fatty acid supplementation with placebo controlled on
cardiorespiratory fitness (CRF) parameters.
Variable
MCT
(n = 12)
Mean ± SD
OMEGA
(n = 14)
Mean ± SD
ALL
(n = 26)
Mean ± SD
Pre
Post
Pre
Post
Pre
Post
OUE [mL*L−1]
35.5 ± 3.7
37.2 ± 3.1 ***
35.4 ± 3.3
37.6 ± 3.1 ***
35.5 ± 3.4
37.4 ± 3.0 ***
OUEP [mL*L−1]
41.8 ± 5.2
42.9 ± 3.8
41.3 ± 4.6
43.6 ± 4.0 *
41.6 ± 4.8
43.2 ± 3.9 **
OUE@VAT [mL*L−1]
33.2 ± 3.8
35.4 ± 3.5 **
32.7 ± 3.6
35.9 ± 4.7 *
32.9 ± 3.7
35.7 ± 4.1 ***
Ve [L*min−1]
93.8 ± 11.6
90.7 ± 9.3 **
92.9 ± 20.4
87.4 ± 20.2 *
93.3 ± 16.4
88.9 ± 15.7 *
OUE—oxygen uptake efficiency; OUEP—oxygen uptake efficiency plateau; OUE@VAT—oxygen uptake efficiency
at the ventilatory anaerobic threshold; Ve—minute ventilation; * p < 0.05 for post to pre value; ** p < 0.01 for post
to pre value; *** p < 0.001 for post to pre value; data are presented as mean ± standard deviation (SD).
3.5.2. Oxygen Uptake Efficiency Plateau
Oxygen uptake efficiency plateau values increased in the OMEGA group from 41.3 ± 4.6
to 43.6 ± 4.0; (p = 0.017). There were no changes in the MCT group (p = 0.2). Moreover, the
analysis of the two groups together (regardless of the supplementation that was undertaken)
showed that OUEP increased from 41.6 ± 4.8 to 43.2 ± 3.9; (p = 0.007, Table 2).
3.5.3. Oxygen Uptake at Ventilatory Anaerobic Threshold
There was an increase in OUE@VAT in the OMEGA group from 32.7 ± 3.6 to 35.9 ± 4.7;
(p = 0.012) and in the MCT group from 33.2 ± 3.8 to 35.4 ± 3.5; (p = 0.003). The results,
regardless of the supplementation undertaken, showed that OUE@VAT increased from
32.9 ± 3.7 to 35.7 ± 4.1; (p < 0.001, Table 2).
4. Discussion
This is the first study to report the correlations between OUE, OUEP, OUE@VAT and
VO2peak as well as OUEP and OUE@VAT and RE. They were analyzed in terms of reliability
and repeatability, and whether they could be non-invasive substitute measurements for
VO2peak and RE. Additionally, we investigated whether these parameters were altered
following supplementation with omega-3 fatty acids.
The true VO2max value is mainly achievable during a laboratory progressive exercise
test to exhaustion where large muscle groups are involved. Simultaneously, the observed
kinetics of oxygen supply/utilization in the muscles must be without significant changes:
the so-called plateau [18]. It is known that this phenomenon occurs when a high intensity
is met, and the primary criteria for achieving this parameter (VO2max) during CPET are:
(1) reaching a VO2 plateau or (2) levelling-off the oxygen uptake (VO2) [19–21]. Thus, in
Sun and co-authors’ study, OUE, OUEP and OUE@VAT comprehensively reflected cardio-
vascular functions as an alternative for parameters assessing CRF without the need for
maximum effort [4]. A steeper OUE (VO2/Ve) and higher values of OUEP and OUE@VAT
show more efficient oxygen uptake and utilization in the working skeletal muscles. OUE
showed an improvement, but, for both groups, this occurred after 12 weeks of interven-
tion. Hence, it is believed that the increase in slope/higher OUE values was the result of
endurance training. Moreover, in our study, weak or no correlation was observed between
OUEP and peak oxygen uptake. In a study by Bongers et al. [22], in which 214 children
participated, OUEP was weak-to-moderately correlated with VO2peak (r = 0.646), which
is inconsistent with our results. However, children and adults respond differently to ex-
ercise, which might explain this difference. Another study also confirms that OUEP does
not accurately predict VO2max in male adolescents and should not replace VO2max, when
assessing CRF [19]. In our study, OUE@VAT also demonstrated no correlation with VO2peak
before and after 12 weeks of intervention. In contrast to our results, one study revealed
that ventilatory anaerobic threshold (VAT) strongly correlated with VO2peak (r = 0.831) [23].
However, there is a difference between the compared parameters, because OUE@VAT is the
Int. J. Environ. Res. Public Health 2022, 19, 14043
8 of 10
60-s average of consecutive measurements at and immediately before the VAT. On the other
hand, VAT is a single measurement and is not free from intra-observer and inter-observer
variability [24]. Hence, both OUEP and OUE@VAT may be more stable measurements than
VAT; however, the results of our study did not confirm this.
Endurance capacity also has a stable predictor in the form of RE [25]. However, as
earlier authors suggest, an accurate measurement of RE can be carried out with the use
of invasive lactate measurement, which is one of the disturbances in VO2 steady-state
indicators [26,27]. Therefore, in this study, an attempt was made to replace RE with OUEP
and OUE@VAT and to check whether they can be a solid, non-invasive predictor of RE
in recreational runners. Despite the increase in the efficiency of oxygen uptake in all
participants, the linear regression did not show any correlation between OUEP, OUE@VAT
and RE. Hence, the RE measurement should not be replaced with OUEP and OUE@VAT, as
they are not related.
The assessment of adaptive changes following supplementation with omega-3 fatty
acids is also not fully known. The health-promoting effects of n-3 PUFA supplementation
are well-established [28–30]. These effects are related to the incorporation of EPA and DHA
into the erythrocyte cell membrane [31], skeletal muscles [32] and heart [33]. Furthermore,
the systemic response to supplementation with omega-3 fatty acids as exemplified by
maximum oxygen uptake [7], exercise economy [9,10] or anaerobic endurance capacity [34]
is well-known. Nevertheless, in our study, for the first time, an attempt was made to link
the effect of supplemental EPA + DHA to changes in OUEP and OUE@VAT. However, the
OUE parameters increased in both groups. Therefore, changes in OUEP and OUE@VAT
following 12 weeks of intervention are dictated by adaptation to endurance training rather
than changes caused by EPA and DHA supplementation.
Limitations and Future Perspectives
Despite some valuable information coming from this study, there are some limitations.
First, the small number of participants could distort the estimate of correlations between
the variables. Second, this study was conducted in male runners only; therefore, these
findings cannot be generalized and extrapolated to females. Future studies should include
a larger number of participants and include females.
5. Conclusions
In conclusion, the results obtained in this study do not support the use of OUEP
and OUE@VAT as an alternative parameter to VO2peak and RE. Additionally, the 12-week
supplementation of omega-3 fatty acids at a dose of 2234 mg of EPA and 916 mg of DHA
daily did not reveal changes in OUEP and OUE@VAT. Hence, the suitability of using
OUEP and OUE@VAT as alternative, non-invasive CRF parameters for VO2peak and RE can
be questioned.
Author Contributions: Conceptualization, Z.J.; methodology, Z.J. and M.C.; software, Z.J., M.T. and
M.C.; validation, R.L.; formal analysis, Z.J. and M.C.; investigation, Z.J., M.T. and M.C.; resources,
M.T.; data curation, M.C.; writing—original draft preparation, Z.J., M.T., P.C.C. and R.L.; writing—
review and editing, Z.J., M.T., M.C., P.C.C. and R.L.; visualization, Z.J. and P.C.C.; supervision, P.C.C.
and R.L.; project administration, M.T.; funding acquisition, M.T. All authors have read and agreed to
the published version of the manuscript.
Funding: This research was funded by National Science Center (Poland), grant number 2018/31/N/N
Z7/02962.
Institutional Review Board Statement: The study was conducted in accordance with the Declaration
of Helsinki and approved by the Bioethical Committee of Regional Medical Society in Gda´nsk
(NKBBN/628/2019).
Informed Consent Statement: Informed consent was obtained from all subjects involved in the
study.
Int. J. Environ. Res. Public Health 2022, 19, 14043
9 of 10
Data Availability Statement: The data presented in this study are available on request from the
corresponding author.
Conflicts of Interest: The authors declare no conflict of interest.
References
1.
Levine, B.D. What Do We Know, and What Do We Still Need to Know? J. Physiol. 2008, 586, 25–34. [CrossRef] [PubMed]
2.
Taylor, H.L.; Buskirk, E.; Henschel, A. Maximal Oxygen Intake as an Objective Measure of Cardio-Respiratory Performance. J.
Appl. Physiol. 1955, 8, 73–80. [CrossRef]
3.
Poole, D.C.; Jones, A.M. Measurement of the Maximum Oxygen Uptake Vo2max: Vo2peak Is No Longer Acceptable. J. Appl.
Physiol. 2017, 122, 997–1002. [CrossRef]
4.
Sun, X.-G.; Hansen, J.E.; Stringer, W.W. Oxygen Uptake Efficiency Plateau: Physiology and Reference Values. Eur. J. Appl. Physiol.
2012, 112, 919–928. [CrossRef]
5.
Buckley, J.D.; Burgess, S.; Murphy, K.J.; Howe, P.R.C. DHA-Rich Fish Oil Lowers Heart Rate during Submaximal Exercise in Elite
Australian Rules Footballers. J. Sci. Med. Sport 2009, 12, 503–507. [CrossRef] [PubMed]
6.
Nieman, D.C.; Henson, D.A.; McAnulty, S.R.; Jin, F.; Maxwell, K.R. N-3 Polyunsaturated Fatty Acids Do Not Alter Immune and
Inflammation Measures in Endurance Athletes. Int. J. Sport Nutr. Exerc. Metab. 2009, 19, 536–546. [CrossRef] [PubMed]
7.
˙Zebrowska, A.; Mizia-Stec, K.; Mizia, M.; G ˛asior, Z.; Poprz˛ecki, S. Omega-3 Fatty Acids Supplementation Improves Endothelial
Function and Maximal Oxygen Uptake in Endurance-Trained Athletes. Eur. J. Sport Sci. 2015, 15, 305–314. [CrossRef] [PubMed]
8.
Peoples, G.E.; McLennan, P.L.; Howe, P.R.C.; Groeller, H. Fish Oil Reduces Heart Rate and Oxygen Consumption During Exercise.
J. Cardiovasc. Pharmacol. 2008, 52, 540–547. [CrossRef]
9.
Hingley, L.; Macartney, M.J.; Brown, M.A.; McLennan, P.L.; Peoples, G.E. DHA-Rich Fish Oil Increases the Omega-3 Index and
Lowers the Oxygen Cost of Physiologically Stressful Cycling in Trained Individuals. Int. J. Sport Nutr. Exerc. Metab. 2017, 27,
335–343. [CrossRef]
10.
Tomczyk, M.; Jost, Z.; Chroboczek, M.; Urba´nski, R.; Calder, P.C.; Fisk, H.L.; Sprengel, M.; Antosiewicz, J. Effects of 12 Weeks of
Omega-3 Fatty Acid Supplementation in Long-Distance Runners. Med. Sci. Sports Exerc. 2022; in press. [CrossRef] [PubMed]
11.
Jones, A.M.; Kirby, B.S.; Clark, I.E.; Rice, H.M.; Fulkerson, E.; Wylie, L.J.; Wilkerson, D.P.; Vanhatalo, A.; Wilkins, B.W. Physiological
Demands of Running at 2-Hour Marathon Race Pace. J. Appl. Physiol. 2021, 130, 369–379. [CrossRef]
12.
Pallarés, J.G.; Morán-Navarro, R.; Ortega, J.F.; Fernández-Elías, V.E.; Mora-Rodriguez, R. Validity and Reliability of Ventilatory
and Blood Lactate Thresholds in Well-Trained Cyclists. PLoS ONE 2016, 11, e0163389. [CrossRef] [PubMed]
13.
Beaver, W.L.; Wasserman, K.; Whipp, B.J. A New Method for Detecting Anaerobic Threshold by Gas Exchange. J. Appl. Physiol.
1986, 60, 2020–2027. [CrossRef] [PubMed]
14.
Day, J.R.; Rossiter, H.B.; Coats, E.M.; Skasick, A.; Whipp, B.J. The Maximally Attainable Vo2 during Exercise in Humans: The Peak
vs. Maximum Issue. J. Appl. Physiol. 2003, 95, 1901–1907. [CrossRef]
15.
Costa, P.; Simão, R.; Perez, A.; Gama, M.; Lanchtermacher, R.; Musialowski, R.; Braga, F.; de Mello Coelho, V.; Palma, A. A
Randomized Controlled Trial Investigating the Effects of Undulatory, Staggered, and Linear Load Manipulations in Aerobic
Training on Oxygen Supply, Muscle Injury, and Metabolism in Male Recreational Runners. Sports Med. Open 2019, 5, 32. [CrossRef]
16.
de Blaiser, C.; Roosen, P.; Willems, T.; Danneels, L.; vanden Bossche, L.; de Ridder, R. Is Core Stability a Risk Factor for Lower
Extremity Injuries in an Athletic Population? A Systematic Review. Phys. Ther. Sport 2018, 30, 48–56. [CrossRef] [PubMed]
17.
Fisk, H.L.; West, A.L.; Childs, C.E.; Burdge, G.C.; Calder, P.C. The Use of Gas Chromatography to Analyze Compositional
Changes of Fatty Acids in Rat Liver Tissue during Pregnancy. J. Vis. Exp. 2014, 85, e51445. [CrossRef]
18.
Joyner, M.J.; Coyle, E.F. Endurance Exercise Performance: The Physiology of Champions. J. Physiol. 2008, 586, 35–44. [CrossRef]
[PubMed]
19.
Sheridan, S.; McCarren, A.; Gray, C.; Murphy, R.P.; Harrison, M.; Wong, S.H.S.; Moyna, N.M. Maximal Oxygen Consumption and
Oxygen Uptake Efficiency in Adolescent Males. J. Exerc. Sci. Fit. 2021, 19, 75–80. [CrossRef] [PubMed]
20.
Niemeyer, M.; Knaier, R.; Beneke, R. The Oxygen Uptake Plateau—A Critical Review of the Frequently Misunderstood Phe-
nomenon. Sport. Med. 2021, 51, 1815–1834. [CrossRef]
21.
Martin-Rincon, M.; Calbet, J.A.L. Progress Update and Challenges on VO2max Testing and Interpretation. Front. Physiol. 2020, 11,
1070. [CrossRef]
22.
Bongers, B.C.; Hulzebos, E.H.; Helbing, W.A.; ten Harkel, A.D.; van Brussel, M.; Takken, T. Response Profiles of Oxygen Uptake
Efficiency during Exercise in Healthy Children. Eur. J. Prev. Cardiol. 2016, 23, 865–873. [CrossRef] [PubMed]
23.
Mourot, L.; Perrey, S.; Tordi, N.; Rouillon, J.D. Evaluation of Fitness Level by the Oxygen Uptake Efficiency Slope after a
Short-Term Intermittent Endurance Training. Int. J. Sports Med. 2004, 25, 85–91. [CrossRef]
24.
Yeh, M.P.; Gardner, R.M.; Adams, T.D.; Yanowitz, F.G.; Crapo, R.O. “Anaerobic Threshold”: Problems of Determination and
Validation. J. Appl. Physiol. 1983, 55, 1178–1186. [CrossRef] [PubMed]
25.
Saunders, P.U.; Pyne, D.B.; Telford, R.D.; Hawley, J.A. Factors Affecting Running Economy in Trained Distance Runners. Sport.
Med. 2004, 34, 465–485. [CrossRef]
26.
Hoff, J.; Støren, Ø.; Finstad, A.; Wang, E.; Helgerud, J. Increased Blood Lactate Level Deteriorates Running Economy in World
Class Endurance Athletes. J. Strength Cond. Res. 2016, 30, 1373–1378. [CrossRef] [PubMed]
Int. J. Environ. Res. Public Health 2022, 19, 14043
10 of 10
27.
Jones, A.M. The Physiology of the World Record Holder for the Women’s Marathon. Int. J. Sports Sci. Coach. 2006, 1, 101–116.
[CrossRef]
28.
Calder, P.C. N–3 Fatty Acids and Cardiovascular Disease: Evidence Explained and Mechanisms Explored. Clin. Sci. 2004, 107,
1–11. [CrossRef] [PubMed]
29.
Wang, C.; Harris, W.S.; Chung, M.; Lichtenstein, A.H.; Balk, E.M.; Kupelnick, B.; Jordan, H.S.; Lau, J. N−3 Fatty Acids from Fish or
Fish-Oil Supplements, but Not α-Linolenic Acid, Benefit Cardiovascular Disease Outcomes in Primary- and Secondary-Prevention
Studies: A Systematic Review. Am. J. Clin. Nutr. 2006, 84, 5–17. [CrossRef] [PubMed]
30.
Calder, P.C. Very Long-Chain n-3 Fatty Acids and Human Health: Fact, Fiction and the Future. Proc. Nutr. Soc. 2018, 77, 52–72.
[CrossRef] [PubMed]
31.
Katan, M.B.; Deslypere, J.P.; van Birgelen, A.P.; Penders, M.; Zegwaard, M. Kinetics of the Incorporation of Dietary Fatty Acids
into Serum Cholesteryl Esters, Erythrocyte Membranes, and Adipose Tissue: An 18-Month Controlled Study. J. Lipid. Res. 1997,
38, 2012–2022. [CrossRef]
32.
McGlory, C.; Galloway, S.D.R.; Hamilton, D.L.; McClintock, C.; Breen, L.; Dick, J.R.; Bell, J.G.; Tipton, K.D. Temporal Changes in
Human Skeletal Muscle and Blood Lipid Composition with Fish Oil Supplementation. Prostaglandins Leukot. Essent. Fat. Acids.
2014, 90, 199–206. [CrossRef] [PubMed]
33.
Harris, W.S.; von Schacky, C. The Omega-3 Index: A New Risk Factor for Death from Coronary Heart Disease? Prev. Med. 2004,
39, 212–220. [CrossRef] [PubMed]
34.
Gravina, L.; Brown, F.F.; Alexander, L.; Dick, J.; Bell, G.; Witard, O.C.; Galloway, S.D.R. N-3 Fatty Acid Supplementation During
4 Weeks of Training Leads to Improved Anaerobic Endurance Capacity, but Not Maximal Strength, Speed, or Power in Soccer
Players. Int. J. Sport Nutr. Exerc. Metab. 2017, 27, 305–313. [CrossRef] [PubMed]
| Improved Oxygen Uptake Efficiency Parameters Are Not Correlated with VO<sub>2peak</sub> or Running Economy and Are Not Affected by Omega-3 Fatty Acid Supplementation in Endurance Runners. | 10-28-2022 | Jost, Zbigniew,Tomczyk, Maja,Chroboczek, Maciej,Calder, Philip C,Laskowski, Radosław | eng |
PMC9794057 |
1
S8 Table. Consensus decision.
Results of the consensus decision of the steering committee, sorted by level of agreement.
Member
1
Member
2
Member
3
Member
4
Member
5
Level of
agreement (%)
Recovery speeda
Yes
Yes
Yes
Yes
Yes
100
Weight/ BMI
No
Yes
Yes
No
Yes
60
Tendon stiffness
Yes
No
Yes
No
Yes
60
Heat resistance
capacity
Yes
Yes
No
Yes
No
60
Altitude training
sensitivity
Yes
Yes
No
Yes
No
60
Angiogenesis
No
Yes
No
Yes
No
40
Muscle fibre
transformation
capacity
No
Yes
No
Yes
No
40
Healing function of
soft tissue
No
Yes
No
No
No
20
Risk of joint injuries
No
Yes
No
No
No
20
Risk of upper
respiratory tract
infections
No
Yes
No
No
No
20
Emotion regulation
No
No
No
Yes
No
20
Self-control
No
No
No
Yes
No
20
Resilience
No
Yes
No
No
No
20
a100% level of agreement and the factor therefore was included in the consensus report.
Yes = Factor should be included in consensus report.
No = Factor should not be included in consensus report.
| Factors associated with high-level endurance performance: An expert consensus derived via the Delphi technique. | 12-27-2022 | Konopka, Magdalena J,Zeegers, Maurice P,Solberg, Paul A,Delhaije, Louis,Meeusen, Romain,Ruigrok, Geert,Rietjens, Gerard,Sperlich, Billy | eng |
PMC9209328 | Hutchinson MJ, Goosey-Tolfrey VL, Rethinking aerobic exercise intensity prescription in
adults with spinal cord injury: time to end the use of “moderate to vigorous” intensity?
Supplementary Material 1: Dynamic model with lagged independent variable for RPE
and %V̇ O2peak
𝑥𝑥 = % V̇ O2peak
𝑦𝑦 = RPE
Z1 = 1, when i = 1 (i = measurement occasion)
Z2 = 1, when i > 1
TETRA = 1, if Group = TETRA
𝑦𝑦𝑖𝑖𝑖𝑖 = 𝛽𝛽1𝑖𝑖𝑍𝑍1𝑖𝑖𝑖𝑖 + 𝛽𝛽2𝑖𝑖𝑍𝑍1. 𝑥𝑥𝑖𝑖𝑖𝑖 + 𝛽𝛽3𝑖𝑖𝑍𝑍2𝑖𝑖𝑖𝑖 + 𝛽𝛽4𝑖𝑖𝑍𝑍2. 𝑥𝑥𝑖𝑖𝑖𝑖 + 𝛽𝛽5𝑖𝑖𝑍𝑍2. 𝑥𝑥𝑖𝑖−1𝑖𝑖 + 𝛽𝛽6𝑖𝑖𝑍𝑍2. 𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑖𝑖 + 𝑢𝑢0𝑖𝑖
𝛽𝛽1𝑖𝑖 = 𝛽𝛽1 + 𝑒𝑒1𝑖𝑖𝑖𝑖
𝛽𝛽2𝑖𝑖 = 𝛽𝛽2 + 𝑢𝑢2𝑖𝑖
𝛽𝛽3𝑖𝑖 = 𝛽𝛽3 + 𝑒𝑒3𝑖𝑖𝑖𝑖
𝛽𝛽4𝑖𝑖 = 𝛽𝛽4 + 𝑢𝑢4𝑖𝑖
𝛽𝛽5𝑖𝑖 = 𝛽𝛽5 + 𝑢𝑢5𝑖𝑖
𝛽𝛽6𝑖𝑖 = 𝛽𝛽6 + 𝑒𝑒6𝑖𝑖𝑖𝑖
⎝
⎜
⎛
𝑢𝑢0𝑖𝑖
𝑢𝑢2𝑖𝑖
𝑢𝑢4𝑖𝑖
𝑢𝑢5𝑖𝑖
⎠
⎟
⎞ ~𝑁𝑁(0, Ω𝑢𝑢): Ω𝑢𝑢 =
⎣
⎢
⎢
⎢
⎡ 𝜎𝜎𝑢𝑢0
2
𝜎𝜎𝑢𝑢02
𝜎𝜎𝑢𝑢2
2
𝜎𝜎𝑢𝑢04
0
𝜎𝜎𝑢𝑢4
2
𝜎𝜎𝑢𝑢05
0
𝜎𝜎𝑢𝑢45
𝜎𝜎𝑢𝑢5
2 ⎦
⎥
⎥
⎥
⎤
ቌ
𝑒𝑒1𝑖𝑖𝑖𝑖
𝑒𝑒3𝑖𝑖𝑖𝑖
𝑒𝑒6𝑖𝑖𝑖𝑖
ቍ ~𝑁𝑁(0, Ω𝑒𝑒): Ω𝑒𝑒 =
𝜎𝜎𝑒𝑒1
2
0
𝜎𝜎𝑒𝑒3
2
0
𝜎𝜎𝑒𝑒36
𝜎𝜎𝑒𝑒6
2
Hutchinson MJ, Goosey-Tolfrey VL, Rethinking aerobic exercise intensity prescription in
adults with spinal cord injury: time to end the use of “moderate to vigorous” intensity?
𝑦𝑦𝑖𝑖𝑖𝑖 = 5.133𝑍𝑍1𝑖𝑖𝑖𝑖 + 0.074𝑍𝑍1. 𝑥𝑥𝑖𝑖𝑖𝑖 + 3.411𝑍𝑍2𝑖𝑖𝑖𝑖 + 0.093𝑍𝑍2. 𝑥𝑥𝑖𝑖𝑖𝑖 + 0.074𝑍𝑍2. 𝑥𝑥𝑖𝑖−1𝑖𝑖
− 1.081𝑍𝑍2. 𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑖𝑖 + 𝑢𝑢0𝑖𝑖
−2𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑒𝑒𝑙𝑙𝑙𝑙ℎ𝑙𝑙𝑙𝑙𝑜𝑜 = 2744.893
⎝
⎜
⎛
𝑢𝑢0𝑖𝑖
𝑢𝑢2𝑖𝑖
𝑢𝑢4𝑖𝑖
𝑢𝑢5𝑖𝑖
⎠
⎟
⎞ ~𝑁𝑁(0, Ω𝑢𝑢): Ω𝑢𝑢 = ൦
3.524 (0.459)
−0.047 (0.009)
0.001 (0.000)
0.000
0
0.000
0.000
0
0.000
0.000
൪
ቌ
𝑒𝑒1𝑖𝑖𝑖𝑖
𝑒𝑒3𝑖𝑖𝑖𝑖
𝑒𝑒6𝑖𝑖𝑖𝑖
ቍ ~𝑁𝑁(0, Ω𝑒𝑒): Ω𝑒𝑒 =
0.768 (0.381)
0
1.044 (0.069)
0
0.212 (0.126)
0.000
Level 1 variance:
𝑍𝑍1 = 𝜎𝜎𝑒𝑒1
2 = 0.768
𝑍𝑍2 (𝑃𝑃𝑇𝑇𝑇𝑇𝑇𝑇) = 𝜎𝜎𝑒𝑒3
2 = 1.044
𝑍𝑍3 = 𝜎𝜎𝑒𝑒3
2 + 2𝜎𝜎36 + 𝜎𝜎𝑒𝑒6
2 = 1.469
Level 2 variance
𝑍𝑍1 = 𝜎𝜎𝑢𝑢0
2 + 2𝜎𝜎𝑢𝑢02 + 𝜎𝜎𝑢𝑢2
2 = 3.431
𝑍𝑍2 = 𝜎𝜎𝑢𝑢0
2 + 2𝜎𝜎𝑢𝑢04 + 𝜎𝜎𝑢𝑢4
2 + 2𝜎𝜎𝑢𝑢05 + 2𝜎𝜎𝑢𝑢45 + 𝜎𝜎𝑢𝑢5
2 = 3.524
Coefficient
Value
Standard error
P
𝛽𝛽1
5.113
0.473
< 0.0005
𝛽𝛽2
0.074
0.013
< 0.0005
𝛽𝛽3
3.411
0.242
< 0.0005
𝛽𝛽4
0.093
0.009
< 0.0005
𝛽𝛽5
0.074
0.010
< 0.0005
𝛽𝛽6
-1.081
0.419
0.009
Hutchinson MJ, Goosey-Tolfrey VL, Rethinking aerobic exercise intensity prescription in
adults with spinal cord injury: time to end the use of “moderate to vigorous” intensity?
Supplementary Material 2: Dynamic model with lagged independent variable for RPE
and %HRpeak
𝑥𝑥 = % HRpeak
𝑦𝑦 = RPE
Z1 = 1, when i = 1 (i = measurement occasion)
Z2 = 1, when i > 1
PARA = 1, if Group = PARA
𝑦𝑦𝑖𝑖𝑖𝑖 = 𝛽𝛽1𝑖𝑖𝑍𝑍1𝑖𝑖𝑖𝑖 + 𝛽𝛽2𝑍𝑍1. 𝑥𝑥𝑖𝑖𝑖𝑖 + 𝛽𝛽3𝑖𝑖𝑍𝑍2𝑖𝑖𝑖𝑖 + 𝛽𝛽4𝑍𝑍2. 𝑥𝑥𝑖𝑖𝑖𝑖 + 𝛽𝛽5𝑍𝑍2. 𝑥𝑥𝑖𝑖−1𝑖𝑖 + 𝛽𝛽6𝑖𝑖𝑃𝑃𝑇𝑇𝑇𝑇𝑇𝑇 + 𝑢𝑢0𝑖𝑖
𝛽𝛽1𝑖𝑖 = 𝛽𝛽1 + 𝑒𝑒1𝑖𝑖𝑖𝑖
𝛽𝛽3𝑖𝑖 = 𝛽𝛽3 + 𝑒𝑒3𝑖𝑖𝑖𝑖
൫𝑢𝑢0𝑖𝑖൯~𝑁𝑁(0, Ω𝑢𝑢): Ω𝑢𝑢 = [𝜎𝜎𝑢𝑢0
2 ]
൬𝑒𝑒1𝑖𝑖𝑖𝑖
𝑒𝑒3𝑖𝑖𝑖𝑖൰ ~𝑁𝑁(0, Ω𝑒𝑒): Ω𝑒𝑒 = ቈ𝜎𝜎𝑒𝑒1
2
0
𝜎𝜎𝑒𝑒3
2
𝑦𝑦𝑖𝑖𝑖𝑖 = −1.375𝑍𝑍1𝑖𝑖𝑖𝑖 + 0.160𝑍𝑍1. 𝑥𝑥𝑖𝑖𝑖𝑖 − 3.044𝑍𝑍2𝑖𝑖𝑖𝑖 + 0.168𝑍𝑍2. 𝑥𝑥𝑖𝑖𝑖𝑖 + 0.044𝑍𝑍2. 𝑥𝑥𝑖𝑖−1𝑖𝑖
+ 0.707𝑃𝑃𝑇𝑇𝑇𝑇𝑇𝑇 + 𝑢𝑢0𝑖𝑖
−2𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑒𝑒𝑙𝑙𝑙𝑙ℎ𝑙𝑙𝑙𝑙𝑜𝑜 = 2727.763
൫𝑢𝑢0𝑖𝑖൯~𝑁𝑁(0, Ω𝑢𝑢): Ω𝑢𝑢 = [2.929 (0.387)]
൬𝑒𝑒1𝑖𝑖𝑖𝑖
𝑒𝑒3𝑖𝑖𝑖𝑖൰ ~𝑁𝑁(0, Ω𝑒𝑒): Ω𝑒𝑒 = 1.471 (0.209)
0
1.182 (0.073)൨
Coefficient
Value
Standard error
P
𝛽𝛽1
-1.375
0.814
0.091
𝛽𝛽2
0.160
0.014
< 0.0005
𝛽𝛽3
-3.044
0.355
< 0.0005
𝛽𝛽4
0.168
0.010
< 0.0005
𝛽𝛽5
0.044
0.011
< 0.0005
𝛽𝛽6
0.707
0.324
0.029
| Rethinking aerobic exercise intensity prescription in adults with spinal cord injury: time to end the use of "moderate to vigorous" intensity? | 12-08-2021 | Hutchinson, Michael J,Goosey-Tolfrey, Victoria L | eng |
PMC6992743 | 1
Scientific RepoRtS | (2020) 10:1523 | https://doi.org/10.1038/s41598-020-58492-8
www.nature.com/scientificreports
estimation of energy consumed
by middle-aged recreational
marathoners during a marathon
using accelerometry-based devices
carlos Hernando1,2*, Carla Hernando3, Ignacio Martinez-navarro4,5, Eladio collado-Boira6,
nayara panizo6 & Barbara Hernando7
As long-distance races have substantially increased in popularity over the last few years, the
improvement of training programs has become a matter of concern to runners, coaches and health
professionals. triaxial accelerometers have been proposed as a one of the most accurate tools to
evaluate physical activity during free-living conditions. In this study, eighty-eight recreational
marathon runners, aged 30–45 years, completed a marathon wearing a GENEActiv accelerometer
on their non-dominant wrist. energy consumed by each runner during the marathon was estimated
based on both running speed and accelerometer output data, by applying the previously established
GENEActiv cut-points for discriminating the six relative-intensity activity levels. Since accelerometry
allowed to perform an individualized estimation of energy consumption, higher interpersonal
differences in the number of calories consumed by a runner were observed after applying the
accelerometry-based approach as compared to the speed-based method. Therefore, pacing analyses
should include information of effort intensity distribution in order to adjust race pacing appropriately
to achieve the marathon goal time. Several biomechanical and physiological parameters (maximum
oxygen uptake, energy cost of running and running economy) were also inferred from accelerometer
output data, which is of great value for coaches and doctors.
Running a marathon has rapidly become one of the most popular activities nowadays as shown by the number of
amateur participants with hundreds of marathons worldwide1,2. It is well-known that running a marathon is one
of the most challenging endurance competitions3,4. As a result of recent research focused on improving training
programs, which aimed to avoid soreness and prevent energy deficit during ultraendurance races5, the number
of runners crossing the finish (ultra)marathon line has significantly raised over the past few years6,7. For example,
a total of 3,388 runners more finished the Valencia Fundación Trinidad Alfonso EDP Marathon in 2018 as com-
pared to the 2016 edition (19,246 versus 15,858 finishers, respectively)8.
In their way towards the improvement of marathon time, recreational runners are surrounded by a wide range
of professionals in order to achieve their objectives9,10. Consequently, many studies has been focused on develop-
ing different methodologies to evaluate factors affecting running performance, such as the pacing strategy2,11, the
energy consumption12–14, the maximal oxygen uptake V
( O
)
2 max
15, the fraction of VO2 max
maintained (F)15, the
running speed16, the energy cost of running (Cr)17, and physical, biomechanical, metabolic, psychological and
social factors18.
Among all these factors, changes of running speed over race sections have been widely studied in order to
explain the running success of more efficient pacers – runners who are able to maintain their initial running pace
for more kilometers2. These more efficient pacers may avoid an excessive energy consumption while running the
first part of the marathon5.
1Sport Service, Jaume I University, Castellon, Spain. 2Department of Education and Specific Didactics, Jaume
I University, Castellon, Spain. 3Department of Mathematics, Carlos III University of Madrid, Madrid, Spain.
4Department of Physical Education and Sport, University of Valencia, Valencia, Spain. 5Sports Health Unit, Vithas-
Nisa 9 de Octubre Hospital, Valencia, Spain. 6Faculty of Health Sciences, Jaume I University, Castellon, Spain.
7Department of Medicine, Jaume I University, Castellon, Spain. *email: [email protected]
open
2
Scientific RepoRtS | (2020) 10:1523 | https://doi.org/10.1038/s41598-020-58492-8
www.nature.com/scientificreports
www.nature.com/scientificreports/
Therefore, measuring the energy expended by an individual while performing a specific activity has recently
been targeted by researchers. Ainsworth and colleagues published The Compendium of Physical Activities in
1993 (which was reviewed in 2000 and 2011), allowing to directly extrapolate the energy expenditure in Metabolic
Equivalent Task (METs), and thus in kilocalories (kcal), for running activities according to speed12,13,19.
Since the Compendium did not take into account interpersonal differences, the use of accelerometry-based
devices has been proposed to evaluate free-living physical activities performed by an individual, in terms of dura-
tion, frequency and intensity14,20,21. Therefore, using the cut-points recommended for a specific population and/
or activity, accelerometer output data can be applied to indirectly measure the energy expended by an individual
in METs22–24.
In this regard, our research group aimed to monitor middle-aged recreational marathoners during a marathon
using accelerometry-based devices. For this purpose, we previously established the GENEActiv cut-points that
dsicriminate the six relative-intensity activity levels in recreational marathoners25. This lab-based study was essen-
tial in order to delineate specific GENEActiv cut-points for a specific population who presents higher relative level
of fitness than the standard adult population. At this point, the main goal of the current study was to apply the
GENEActiv cut-points previously established for estimating the energy consumed by middle-aged recreational
marathoners during a marathon race (a free-living condition). Accelerometer output data allowed us to analyze
the effort distribution that runners followed to achieve their marathon time, by means of the time running at
each one of the six related-intensity levels (sedentary, light, moderate, vigorous, very vigorous and extremely vig-
orous activity) during the marathon. This information may be extremely valuable for both athletes and coaches.
Knowing the intensity, duration and energy cost of an activity is useful for designing training sessions because it
allows to objectively quantify and monitor training load. Energy consumption was also estimated based on run-
ning speed12, and results were compared with those obtained after using accelerometer data.
Results
A detailed description of individuals included in this study is summarized in Table 1.
The accelerometer output data allowed us to analyse the effort distribution that runners followed to achieve
their marathon time, by means of the time running at each one of the six related-intensity levels (sedentary, light,
moderate, vigorous, very vigorous and extremely vigorous activity) during the marathon. Values established for
delineating the six-relative intensity levels of physical activity are detailed in Table 2.
For all individuals, we estimated the energy cost of running a marathon, presenting the caloric consumption
for each one of the 9 marathon sections as well as for the full marathon distance (Tables 3 and 4). The calories
consumed by each runner were calculated based on both accelerometer data (Table 3), as previously described by
our research group25, and running speed (Table 4), following the methodology proposed by Ainsworth and cols12.
The aim of applying also the speed-based method12 in the estimation of energy consumption was to compare the
results obtained with accelerometer devices25. Note that a gold standard method for energy quantification in long
distance races has not been defined yet.
Except for the last race section, a higher number of calories was estimated to be consumed by a runner when
the accelerometry-based method was applied, as compared to the caloric consumption estimated by using the
Variable
Subjects
(N = 88)
Physiological characteristics*
age
38.68 ± 3.61
BMI
22.91 ± 1.62
Weight
69.96 ± 8.91
Heigh
174.44 ± 8.66
% body fat
14.74 ± 4.38
VO2 max (ml·kg−1·min−1)
54.41 ± 5.66
maximum METs
15.55 ± 1.62
Training indicators*
years of running
6.43 ± 2.78
sessions per week
4.90 ± 0.84
kilometers per week
63.45 ± 13.06
hours per week
7.44 ± 2.70
History as marathoner*
marathons finished
3.36 ± 3.02
marathon per year
1.10 ± 0.63
Work intensity#
high intensity
7.95%
medium intensity
30.68%
low intensity
61.36%
Levels of study#
school graduate
4.60%
high school graduate
6.90%
professional certificate
17.24%
undergraduate degree
71.26%
Table 1. Population description. Abbreviations: N, number of samples; BMI, body mass index; SD, standard
deviation. *Values are presented as mean ± SD. #Values are presented as percentage.
3
Scientific RepoRtS | (2020) 10:1523 | https://doi.org/10.1038/s41598-020-58492-8
www.nature.com/scientificreports
www.nature.com/scientificreports/
speed method (Table 4). It is worth highlighting that a greater variation of calories consumed per each individual
was observed after using accelerometry for energy cost estimation, rather than running speed (shown by higher
standard deviation values). The reason of this difference is due to the fact that the accelerometer-based method
takes into account the variability across individuals in terms of energy consumption, while speed-based method
tends to standardize values for all subjects26.
Although no significant differences between energy consumption and marathon time were observed (Fig. 1),
correlation analysis showed that the accelerometry-based method tended to increase the number of calories con-
sumed by the runner with marathon time (ρ = 0.179, p = 0.094). However, the Ainsworth’s method seemed to
present a negative correlation between the caloric consumption and marathon time (Fig. 1). This correlation was
also no significant (ρ = −0.137, p = 0.202).
For a better comparison between methods, the energy consumed by runners was expressed as a relative rate in kil-
ocalories per kilogram of body mass either per minute12,26 or per kilometer17,27, and as the number of times consum-
ing his/her Basal Metabolic Rate (BMR)26,28 (Table 4). The results of this comparison denoted statistically significant
differences in the energy estimated to be consumed by runners after applying the accelerometry- and speed-based
method. That was observed in each one of the 9 race sections as well as in the full marathon distance (Table 4).
Accelerometer output data allowed us to know the physical effort distribution of runners during the mara-
thon, in terms of physical activity intensity. That is, we were able to identify and quantify when a runner is racing
at each one of the six relative-intensity activity levels (sedentary, light, moderate, vigorous, very vigorous and
extremely vigorous)25. Therefore, following the values established in Table 2, the percentage of VO2 max
produced
per each runner was estimated, and this allowed then to calculate the energy of cost running above standing
(Crnet)28 (Table 5).
A negative correlation between the relative energy consumed and the marathon time was observed when
energy consumption was expressed as kilocalories per kilogram of body mass per minute. This negative correla-
tion was enlarged when the speed-based method was applied (ρ = −0.976, p = 1.12 × 10−58), in comparison with
the accelerometry-based method (ρ = −0.307, p = 0.004) (Fig. 2). When the relative rate of energy consumption
was expressed per distance (kcal·kg−1·km−1), the energy expended by runners was positively correlated with the
marathon time after using accelerometry (ρ = 0.402, p = 1.01 × 10−4). No significant correlation was observed
between energy consumption (expressed as a relative rate per kilogram of body weight per kilometre) and time
when speed-based method was applied (ρ = −0.200, p = 0.062).
Discussion
In this study, we aimed to estimate the energy consumed by middle-aged recreational marathoners during a
marathon race (a free-living condition) using accelerometry-based devices25. In our opinion, the application
of accelerometers should be useful to minimize the interpersonal differences in energy consumption caused by
physiological and biomechanical parameters and, therefore, to perform an individualized estimation of energy
consumption.
Up to now, the viability of accelerometers to measure VO2 in combination with other devices, such as pulsom-
eters or global positioning system (GPS) devices, has been analysed under laboratory conditions29–31.
Accelerometers have also been used to monitor athletes and infer their physical activity level24,32,33. However,
accelerometry-based devices had not been applied so far for estimating the energy consumed by a runner in a
marathon race, under normal race conditions, yet. By applying the GENEActiv cut-points for discriminating the
six relative-intensity activity levels in recreational marathoners (previously established in a lab-based study by our
research group25), we were able to know the amount of time that a runner was running at a specific
relative-intensity level (sedentary, light, moderate, vigorous, very vigorous and extremely vigorous activity)
Relative-intensity levels
of physical activity#
Reference values established for each intensity
level by Hernando et al.25
Values used for energy consumption
estimation
VO2
(ml·kg−1·min−1)
METs*
%VO2max
VO2
(ml·kg−1·min−1)
METs*
Sedentary
X < 10%
VO
5 45
2
<
.
METs < 1.56
8.26%
4.5
1.29
Ligth
10% ≤ X < 25%
.
≤
<
.
V
5 45
O
13 63
2
1.56 ≤ METs < 3.90
17.5%
9.54
2.73
Moderate
25% ≤ X < 45%
V
13 63
O
24 54
2
.
≤
<
.
3.9 ≤ METs < 7.01
35.0%
19.10
5.45
Vigorous
45% ≤ X < 65%
.
≤
<
.
V
24 54
O
35 44
2
7.01 ≤ METs < 10.13
55.0%
29.99
8.57
Very Vigorous
65% ≤ X < 85%
V
35 44
O
46 35
2
.
≤
<
.
10.13 ≤ METs < 13.24
75.0%
40.90
11.69
Extremely Vigorous
X ≥ 85%
VO
46 35
2
≥
.
METs ≥ 13.24
92.5%
50.44
14.41
Table 2. Values established for delineating the six-relative intensity levels of physical activity. Abbreviations: N,
number of individuals; VO2 max, maximum oxygen consumption; VO2
, oxygen consumption; MET, metabolic
equivalent task. Each minute of the cardiopulmonary test was classified into one of the six intensity categories of
physical activity relative to an individual’s level of cardiorespiratory V
( O
)
2max . *1 MET = 3.5 ml·kg−1·min−1. 1
MET = 1 kcal·h−1. #X denotes the percentage of a person’s aerobic capacity V
( O
2max)
used to classify each one of
the six relative-intensity categories.
4
Scientific RepoRtS | (2020) 10:1523 | https://doi.org/10.1038/s41598-020-58492-8
www.nature.com/scientificreports
www.nature.com/scientificreports/
during the marathon. Accordingly, the energy consumed by the runner along the race sections and the full mar-
athon distance was estimated.
Differences in the estimation of runners’ energy consumption were observed between the speed- and
accelerometry-based methods. These differences lie in the ability of the accelerometer output data to determine
the physical effort distribution of each runner during the marathon, in terms of physical activity intensity34–36.
Therefore, accelerometers are able to perform an individualized estimation of energy consumption. Note that
several physiological and biomechanical factors that are unique to the individual have been shown to affect the
running efficiency among runners at the same steady-state speed16,27,37. This fact pointed up that estimating the
energy consumption of a runner based uniquely on his/her running speed might be insufficient and that it might
be advisable to apply a correction factor for adjusting for individual differences when estimating the energy
cost of, at least, moderate/vigorous physical activities26. The speed-based approach, proposed by Ainsworth and
cols12, analyse the marathon pace of a runner without taking into account the runner’s effort to race at this speed.
Fewer interpersonal differences in the number of calories consumed by a runner were then observed with the
speed-based method as compared to the accelerometry-based approach. For example, two individuals racing
at identical speed and having equal body mass are estimated to present the same energy cost after applying the
Race
section
Time spend at each relative-intensity level (minutes)
Energy consumed according to the time spend at each relative-intensity level (kcal)
S
L
M
V
VV
EV
Total
S
L
M
V
VV
EV
Total
0–5 km
0.01 ± 0.11
0.00 ± 0.00
1.17 ± 4.87
1.30 ± 4.03
9.82 ± 10.65
14.81 ± 11.53
27.10 ± 3.35
0.02 ± 0.15
0.00 ± 0.00
6.76 ± 26.60
13.94 ± 45.35
136.71
± 148.23
244.83
± 191.55
402.26
± 76.44
5–10 km
0.00 ± 0.00
0.00 ± 0.00
1.42 ± 4.31
1.63 ± 4.00
8.67 ± 8.94
12.86 ± 10.32
24.58 ± 2.23
0.00 ± 0.00
0.00 ± 0.00
8.28 ± 23.95
17.16 ± 44.77
119.77
± 123.75
214.47
± 173.67
359.68
± 73.47
10–15 km
0.00 ± 0.00
0.00 ± 0.00
1.25 ± 3.66
1.84 ± 4.30
8.56 ± 8.94
13.09 ± 10.24
24.74 ± 2.32
0.00 ± 0.00
0.00 ± 0.00
7.84 ± 20.59
19.07 ± 45.53
118.40
± 126.80
216.93
± 171.78
362.25
± 69.49
15-HM
0.01 ± 0.11
0.00 ± 0.00
1.88 ± 4.90
2.23 ± 4.44
9.74 ± 10.07
16.16 ± 12.47
30.01 ± 2.87
0.01 ± 0.12
0.00 ± 0.00
11.62 ± 28.85
23.10 ± 48.98
135.17
± 141.08
267.77
± 208.66
437.67
± 87.00
HM-25km
0.00 ± 0.00
0.01 ± 0.11
0.51 ± 2.09
1.23 ± 3.48
6.06 ± 7.41
11.72 ± 8.45
19.52 ± 1.77
0.00 ± 0.00
0.03 ± 0.29
3.02 ± 12.38
12.57 ± 37.79
84.05
± 102.70
195.15
± 143.59
294.83
± 57.25
25–30 km
0.00 ± 0.00
0.01 ± 0.11
1.13 ± 2.94
1.91 ± 3.84
8.33 ± 8.41
14.11 ± 10.23
25.49 ± 2.51
0.00 ± 0.00
0.04 ± 0.33
6.85 ± 17.57
19.15 ± 38.97
115.57
± 118.10
235.14
± 172.56
376.75
± 72.58
30–35 km
0.00 ± 0.00
0.06 ± 0.38
1.53 ± 4.75
1.81 ± 3.95
8.06 ± 8.74
15.06 ± 11.00
26.51 ± 3.45
0.00 ± 0.00
0.20 ± 1.40
10.00 ± 31.54
18.34 ± 40.36
110.92
± 121.80
250.91
± 186.13
390.38
± 77.84
35–40 km
0.00 ± 0.00
0.09 ± 0.58
2.08 ± 5.38
1.64 ± 3.28
8.22 ± 8.66
15.11 ± 10.51
27.14 ± 3.89
0.00 ± 0.00
0.33 ± 2.21
13.50 ± 36.03
16.04 ± 31.78
114.47
± 120.83
251.14
± 175.75
395.48
± 72.99
40-M
0.02 ± 0.21
0.02 ± 0.15
0.67 ± 2.22
0.39 ± 0.84
2.55 ± 3.30
6.24 ± 4.10
9.89 ± 1.76
0.03 ± 0.31
0.07 ± 0.47
4.23 ± 13.93
3.79 ± 8.19
35.79
± 8.19
104.43
± 70.04
148.35
± 37.76
Marathon
0.05 ± 0.34
0.19 ± 0.92
11.6 ± 25.32
13.95 ± 27.75
69.99 ± 66.19
119.16 ± 82.86
214.98 ± 20.78
0.06 ± 0.47
0.67 ± 3.48
72.10 ± 160.10
143.17 ± 301.99
970.84
± 938.15
1980.78
± 1386.54
3167.63
± 584.12
Table 3. Evaluation of effort distribution and estimation of calories consumed by runners based on
accelerometry data. Abbreviations: S, Sedentary; L, Light; M, Moderate; V, Vigorous; VV, Very Vigorous;
EV, Extremely Vigorous; HM, Half marathon; M, marathon; SD, standard deviation. Values are presented as
mean ± SD.
Race
section
Running
speed
(m·min−1)
Absolute energy
(kcal)
Energy relative to body mass per time
(kcal·kg−1·min−1)
Energy relative to body mass per distance
(kcal·kg−1·km−1)
Number of BMR
Accelerometry
Running
speed*
Accelerometry
Running
speed*
Adjusted
p-value¥
Accelerometry
Running
speed*
Adjusted
p-value¥
Accelerometry
Running
speed*
Adjusted
p-value¥
0–5 km
187.27 ± 23.06
402.26 ± 76.44
352.30 ± 44.85
0.214 ± 0.031
0.189 ± 0.023
6.27 × 10-12
1.154 ± 0.195
1.008 ± 0.026
1.09 × 10-11
12.82 ± 1.84
11.30 ± 1.40
6.27 × 10-12
5–10 km
205.06 ± 18.43
359.68 ± 73.47
354.24 ± 47.26
0.210 ± 0.034
0.208 ± 0.019
0.149
1.030 ± 0.176
1.012 ± 0.023
0.495
12.59 ± 2.03
12.43 ± 1.11
0.149
10–15 km
203.85 ± 18.88
362.25 ± 69.49
355.00 ± 46.26
0.211 ± 0.032
0.207 ± 0.018
0.062
1.040 ± 0.171
1.015 ± 0.025
0.169
12.63 ± 1.93
12.38 ± 1.10
0.062
15-HM
204.94 ± 18.82
437.67 ± 87.00
427.90 ± 54.89
0.210 ± 0.033
0.206 ± 0.020
0.093
1.030 ± 0.177
1.003 ± 0.020
0.358
12.57 ± 2.00
12.31 ± 1.17
0.088
HM-25km
201.49 ± 18.00
294.83 ± 57.25
273.44 ± 35.86
0.217 ± 0.030
0.202 ± 0.022
1.05 × 10-5
1.055 ± 0.164
1.001 ± 0.026
2.46 × 10-3
12.99 ± 1.78
12.10 ± 1.33
1.05 × 10-5
25–30 km
198.01 ± 19.10
376.75 ± 72.58
353.60 ± 47.10
0.213 ± 0.030
0.200 ± 0.020
7.85 × 10-5
1.080 ± 0.170
1.010 ± 0.024
3.16 × 10-4
12.73 ± 1.82
11.98 ± 1.21
7.85 × 10-5
30–35 km
191.43 ± 22.58
390.38 ± 77.84
351.74 ± 46.67
0.213 ± 0.032
0.193 ± 0.025
2.06 × 10-7
1.119 ± 0.186
1.006 ± 0.040
8.18 × 10-7
12.73 ± 1.89
11.55 ± 1.49
2.06 × 10-7
35–40 km
187.65 ± 24.50
395.48 ± 72.99
353.24 ± 46.98
0.211 ± 0.032
0.190 ± 0.026
7.21 × 10-7
1.134 ± 0.174
1.010 ± 0.039
2.37 × 10-10
12.65 ± 1.91
11.36 ± 1.56
7.21 × 10-7
40-M
229.14 ± 42.02
148.35 ± 37.76
153.73 ± 20.93
0.215 ± 0.034
0.229 ± 0.038
0.202
0.964 ± 0.210
1.000 ± 0.039
1.000
12.90 ± 2.03
13.69 ± 2.26
0.209
Marathon
198.06 ± 18.78
3167.63 ± 584.12
2951.45 ± 394.20
0.212 ± 0.030
0.198 ± 0.021
3.48 × 10-5
1.076 ± 0.163
0.999 ± 0.023
8.75 × 10-5
12.70 ± 1.77
11.86 ± 1.23
3.48 × 10-5
Table 4. Comparison between accelerometry- and speed-based approaches in the estimation of energy
consumption. Abbreviations: BMR, Basal metabolic rate; HM, Half marathon; M, Marathon; SD, standard
deviation; p, p-value. Values are presented as mean ± SD. Bold indicates significant results (p-value < 0.05).
*The values are estimated based on running speed, and following the methodology proposed by Ainsworth
et al. (2000)12. ¥P-values were corrected for multiple comparisons by applying the Benjamini-Hochberg
procedure for decreasing the False Discovery Rate.
5
Scientific RepoRtS | (2020) 10:1523 | https://doi.org/10.1038/s41598-020-58492-8
www.nature.com/scientificreports
www.nature.com/scientificreports/
speed-based method, although their physical efforts are completely different according to accelerometry data.
Nevertheless, note that, as in the speed-based methods, accelerometry is not able to perform an absolute quantifi-
cation of the energy consumed by a runner and it is necessary, therefore, to combine different approaches, as well
as to explore other technologies, in future work.
In this regard, accelerometer data collected for each runner was thoroughly analyzed in order to compare
effort distribution between the fastest and the slowest runner of our dataset (Table 6). Note that the fastest runner
was almost running at very vigorous intensity level, showing a good control of physical effort along the full mar-
athon distance. In contrast, the effort distribution of the slowest runner was far from being well-balanced2,38,39.
In fact, the accelerometer data revealed a considerable decay of the intensity level at which the slowest runner
performed after completing 30 km (running at a moderate intensity from an extremely vigorous level). This was a
consequence of the high physical effort sustained by the runner from the beginning of the marathon line, which
reveals the importance of controlling effort distribution in a marathon race. In short, our results suggest that
future pacing analyses should include information of effort intensity distribution in order to adjust race pacing
appropriately to achieve the marathon goal time.
Thanks to accelerometer output data, we were also able to estimate the percentage of VO2 max produced per
each runner, and afterwards the energy of cost running above standing (Crnet)28, at each of the 9 marathon sec-
tions as well as at the full marathon distance. These physiological parameters seem to explain up to 87% of the
long distance race performance27. In addition, the accelerometry-based approach also allowed us to extrapolate
the running economy of each runner, which is considered an important physiological measure for long distance
runners37,40. It is thought that a variety of biomechanical characteristics are likely to contribute to having interper-
sonal differences in the running efficiency, such as the running technique, the elastic power of the muscle-tendon
unit, or the amount of ground contact and vertical oscillation when running41.
As results shown, the fastest runner seemed to present a better efficiency of movement than that presented
by the slowest runner. That is, the energy demanded for a given running velocity was lower by the fastest runner
as compared to the slowest runner. In fact, the average energy cost of marathon running was 3.31 J·kg−1·m−1 for
the fastest runner (whose average speed was 237.05 m·min−1), while it was 4.59 J·kg−1·m−1 for the slowest runner
Figure 1. Plot showing the linear correlation between the calories estimated to be consumed by each runner
and the marathon time. Energy consumption was estimated by using both accelerometry (solid line) and
running speed (dashed line). Each individual is represented by a specific point: filled circles are used when
accelerometry was applied for energy consumption estimation, and filled triangles when speed-based method
was used. Abbreviations: ρ, Spearman’s rank correlation coefficient; p, p-value.
Race
section
Percentage of maximum
oxygen consumption
V
(% O
2 max)
Oxygen uptake relative
to body mass per minute
(ml·kg−1·min−1)
Energy cost of running
above standing*
(J·kg−1·m−1)
0–5 km
82% ± 11.78
44.87 ± 6.43
4.54 ± 0.83
5–10 km
81% ± 13.05
44.07 ± 7.12
4.05 ± 0.76
10–15 km
81% ± 12.41
44.19 ± 6.77
4.09 ± 0.73
15-HM
81% ± 12.83
44.00 ± 7.00
4.04 ± 0.76
HM-25km
83% ± 11.46
45.45 ± 6.25
4.26 ± 0.72
25–30 km
82% ± 11.69
44.54 ± 6.38
4.25 ± 0.73
30–35 km
82% ± 12.11
44.55 ± 6.60
4.40 ± 0.79
35–40 km
81% ± 12.27
44.28 ± 6.70
4.44 ± 0.74
40-M
83% ± 13.06
45.15 ± 7.12
3.79 ± 0.86
Marathon
81% ± 11.38
44.43 ± 6.21
4.23 ± 0.70
Table 5. Estimation of the percentage of VO2 max
, the oxygen uptake relative to body mass per minute and the
energy cost of running above standing based on accelerometry data. Abbreviations: HM, Half marathon; M,
Marathon; VO2 max
, maximum oxygen consumption. *Energy cost of running above
standing = (
−
V
V
( O
O
)
2
2standing (running speed)−1) · 20.9.
6
Scientific RepoRtS | (2020) 10:1523 | https://doi.org/10.1038/s41598-020-58492-8
www.nature.com/scientificreports
www.nature.com/scientificreports/
(whose average speed was 152.88 m·min−1). Apart from physiological parameters, these differences may be also
resulted from biomechanical efficiency, which is influenced by anthropometric parameters, kinematic character-
istics and running style37.
This suggests that the design of training sessions for the slowest runner by his coach should focus on improv-
ing his running style and muscle strength, and subsequently his performance. The useful information offered by
accelerometers (distribution of physical effort in free-living conditions and inference of physiological parameters
as Crnet or % VO2 max) should become more and more important as race distance increase42. Application of accel-
erometers to monitor ultratrail runners may be useful not only for adjusting race strategy, which is crucial for
achieving performance goals2,27,43,44, but also to monitor training sessions and recovery time. Indeed, both
long-term data collection and wrist watch-like format are valuable characteristics of accelerometers since data can
be continuously collected for a long period of time (more than a week) without causing any physical discomfort
to ultraendurance runners45.
However, values of all physiological parameters analyzed in this study were merely estimations based on accel-
erometer data, and were not directly measured46. It is quite difficult, if not impossible, to perform a direct meas-
urement of VO2
on a marathon race, an extremely demanding free-living condition. This makes difficult to find a
gold standard method for quantifying calories consumed by an individual when she/he is performing a physical
activity. That is the reason why indirect measurement methods (such as heart-rate recording devices14,47, pedom-
eters48,49 and accelerometers14,34,36, or their combination29,30,50) are normally applied. Another limitation of our
study is related to the protocol followed to estimate energy consumption according to the range of % VO2 max
delimiting each relative-intensity activity level. Estimations can present a maximum error of 10%, since the
median value of the % VO2 max range was used for energy calculations (as shown in Table 2). Having said that, our
results indicate that accelerometry-based method allows to both identify the individual’s levels of physical activity
intensity during the marathon race and estimate an individualized energy consumption.
In summary, overall the results in this study lead us to believe that GENEActiv. accelerometer is an accu-
rate tool for estimating the energy consumption of middle-recreational marathoners running a marathon, an
extremely demanding free-living physical activity. Accelerometer-derived data was useful to evaluate the effort
intensity distribution along the race, by means of the time running at each six related-intensity levels (sedentary,
light, moderate, vigorous, very vigorous and extremely vigorous activity), and subsequently to estimate the energy
consumption. Therefore, accelerometers may be extremely useful for both athletes and coaches who need to
evaluate the race strategy to achieve marathon final time, but also to monitor training sessions and assess perfor-
mance level progression needed to reach a goal. Several physiological and biomechanical parameters that can be
inferred from accelerometer output data may also support coaches to design specific training sessions according
to runner’s characteristics. Furthermore, the ability to perform an objective assessment of a runner’s fitness level,
as well as energy consumption, in the context of free-living movement indicates that accelerometry-based devices
may be of great value to sport medical professionals.
Since accelerometry-based data is thought to be valuable for monitoring runners along ultra-trail races, future
studies determining cut-off points for quantifying energy consumption would help in the race strategy in terms
of food and fluid intake on race day (a key factor for performance success). Note that these future studies must
take into account that biomechanics and physiology of downhill and uphill running, as well as the energy cost of
running, may differ.
Methods
Sample set.
A total of 95 recreational marathon runners (80 males and 15 females) aged between 30 and
45 years lined up at the start of the Valencia Fundación Trinidad Alfonso EDP 2016 Marathon (20th November,
2016). From all of them, eighty-eight participants crossed the finish line (74 males and 14 females). Non-finishers
Figure 2. Plot showing the linear correlation between the energy estimated to be consumed by each runner
relative to his/her body mass per minute and the marathon time. Energy consumption was estimated by using
both accelerometry (solid line) and running speed (dashed line). Each individual is represented by a specific point:
filled circles are used when accelerometry was applied for energy consumption estimation, and filled triangles
when speed-based method was used. Abbreviations: ρ, Spearman’s rank correlation coefficient; p, p-value.
7
Scientific RepoRtS | (2020) 10:1523 | https://doi.org/10.1038/s41598-020-58492-8
www.nature.com/scientificreports
www.nature.com/scientificreports/
were discarded from further analyses. The entire process of sampling (contact approach and criteria for inclusion
and exclusion of volunteers) has been previously described25.
ethics statement.
All individuals included in the current study were fully informed and gave their writ-
ten consent to participate. The research was conducted according to the Declaration of Helsinki, and it was
approved by the Research Ethics Committee of the University Jaume I of Castellon. This study is enrolled in the
ClinicalTrails.gov database, with the code number NCT03155633 (www.clinicaltrials.gov).
Data collection and analysis.
Four weeks before the marathon, we made an appointment with all partic-
ipants in order to collect anthropometric data, demographics, medical information, training program and com-
petition history. Indeed, all individuals completed a cardiopulmonary test. Details of data collection, processing
and analysis have been previously described25. Population description according to data collected is also available
in our previous work25.
All participants were weighed one hour before the start of the marathon, wearing racing clothes and flats, by
using a Seca 770 scale (Seca Hamburg, Germany). BMI was then calculated (height·mass−2).
For this research, all the participants underwent the same testing under the same experimental conditions.
Participants completed the Valencia Fundación Trinidad Alfonso EDP 2016 Marathon, which was held in
November with a mean dry temperature of 15.6 °C and a mean relative humidity of 50%. The race course altitude
varied from 1 to 27 m above sea level.
During the race, participants wore a GENEActiv accelerometer (Activinsights Ltd., Kimbolton,
Cambridgeshire, United Kingdom). The accelerometer was worn on the non-dominant wrist as a watch.
Accelerometers were adjusted to record acceleration data at a rate of 85.7 Hz. Devices were calibrated by the man-
ufacturer prior to use. Processing of acceleration data has been previously explained in detail25.
Data analysis.
The marathon race was divided into 9 sections as follow: 6 sections of 5 km (0–5 km, 5–10 km,
10–15 km, 25–30 km, 30–35 km and 35–40 km), 1 section of 6.0975 km (15–21.0975 km), 1 section of 3.9025 km
(21.0975–25 km) and 1 section of 2.195 km (40–42.195 km). All data analyses were performed for each one of the
nine marathon sections and for the whole marathon distance. Statistical analyses were done using the IBM SPSS
Statistics v.23 software, and p-values lower than 0.05 were considered as statistically significant. Supplementary
information includes raw data used in this study.
Fastest runner: Marathon time of 178 min, body mass of 69.2 kg, and BMI of 21.36 kg·m−2
Race
section
Time running at each relative-
intensity level (min)
Energy consumption
Running
speed
(m·min−1)
V
% O2max
Crnet
(J·kg−1·m−1)
S
L
M
V
VV
EV
Total
Absolute
(kcal)
Relative to
time
(kcal·kg−1
·min−1)
Relative to
distance
(kcal·kg−1
·km−1)
0–5 km
0
0
0
0
21
0
21.00
283.70
0.20
0.82
238.10
75.00%
3.29
5–10 km
0
0
0
0
21
0
21.00
283.70
0.20
0.82
238.10
75.00%
3.29
10–15 km
0
0
0
1
20
0
21.00
280.09
0.19
0.81
238.10
74.05%
3.25
15-HM
0
0
0
3
22
0
25.00
326.92
0.19
0.77
243.90
72.60%
3.11
HM-25km
0
0
0
0
16
0
16.00
216.15
0.20
0.80
243.91
75.00%
3.22
25–30 km
0
0
0
0
22
0
22.00
297.21
0.20
0.86
227.27
75.00%
3.45
30–35 km
0
0
0
1
21
0
22.00
293.60
0.19
0.85
227.27
74.09%
3.41
35–40 km
0
0
0
0
18
3
21.00
293.13
0.20
0.85
238.10
77.50%
3.40
40-M
0
0
0
0
7
2
9.00
127.87
0.21
0.84
243.89
78.89%
3.38
Marathon
0
0
0
5
168
5
178.00
2402.37
0.20
0.82
237.05
74.93%
3.31
Slowest runner: Marathon time of 276 min, body mass of 74.9 kg, and BMI of 23.38 kg·m−2
0–5
0
0
0
0
24
5
29.00
441.06
0.20
1.18
172.41
78.02%
4.73
5–10
0
0
0
0
17
11
28.00
446.85
0.21
1.19
178.57
81.88%
4.79
10–15
0
0
1
0
7
20
28.00
469.66
0.22
1.25
178.57
86.07%
5.04
15-HM
0
0
0
0
4
31
35.00
617.25
0.24
1.35
174.21
90.50%
5.43
HM-25
0
0
0
0
0
22
22.00
396.54
0.24
1.36
177.39
92.50%
5.45
25–30
0
0
2
4
13
12
31.00
462.89
0.20
1.24
161.29
76.61%
4.97
30–35
0
0
41
1
1
0
43.00
304.84
0.09
0.81
116.28
36.40%
3.27
35–40
0
0
43
0
1
0
44.00
307.75
0.09
0.82
113.64
35.91%
3.30
40-M
0
0
11
0
0
5
16.00
165.11
0.14
1.00
137.19
52.97%
4.04
Marathon
0
0
98
5
67
106
276.00
3611.95
0.17
1.14
152.88
67.16%
4.59
Table 6. Comparison of effort distribution according to accelerometer output data between the fastest and the
slowest runner of our dataset. Abbreviations: S, Sedentary; L, Light; M, Moderate; V, Vigorous; VV, Very
Vigorous; EV, Extremely Vigorous; HM, Half marathon; M, marathon; VO2 max
, maximum oxygen
consumption; Crnet, energy cost of running above standing.
8
Scientific RepoRtS | (2020) 10:1523 | https://doi.org/10.1038/s41598-020-58492-8
www.nature.com/scientificreports
www.nature.com/scientificreports/
Firstly, accelerometer-derived data was used to determine the distribution of exercise intensity of runners
along the marathon with the aim to estimate the calories consumed per each runner. The intensity levels of phys-
ical activity were established following the cut-off points delineated by Hernando and cols25. For calculating the
energy cost, we used the median value of the range of % VO2 max delimiting each intensity category (Table 2),
except for the sedentary category where the standing oxygen cost (4.5 mlO2·kg−1·min−1) was applied as reference
value28. As unit of measurement, we considered that one MET is equal to 3.5ml O2·kg−1·min−1, and one MET is
equal to one kcal·kg−1·h−1. These equivalencies were applied in accordance with the determinations proposed by
Ainsworth and cols12, and taking into account that all volunteers included in the study reported similar BMI
(between 22.17 and 23.44 kg·m−2) and, therefore, differences in the percentage of fatty component among partic-
ipants were absence26,46,51.
Accelerometers were also used to estimate the percentage of VO2 max produced per each runner. Briefly, the
time racing at a specific intensity level was multiplied by its corresponding % VO2 max (Table 2). A weighted aver-
age relative to the total time spent at each section, as well as at the full marathon distance, was then performed.
Then, the VO2net of each runner was calculated by subtracting the VO2standing to the percentage of VO2 max esti-
mated17,28. Together with the running speed measured, the VO2net was finally used to calculate the energy of cost
running above standing (Crnet), following the methodology proposed by di Prampero and cols17.
Next, the average running speed was used to calculate the caloric consumption of runners, following the
methodology proposed by Ainsworth and cols12. The split-times in minutes were recorded for each one of the
marathon sections electronically, and the average running speed of all sections and the whole marathon distance
was calculated. Then, the running speed was associated with a specific MET value, which can be directly used to
calculate the number of calories consumed by a runner12,19.
Finally, the relative values of energy consumption estimated by the two models were compared. As the energy
consumption depends on the person’s body mass, the energy cost of each runner is presented as: (i) the calories
consumed per kilogram of body weight per minute (kcal·kg−1·min−1), in order to obtain the effort intensity;12,19,26
(ii) the calories consumed per kilogram of body weight per kilometer (kcal·kg−1·km−1), to infer the running effi-
ciency of runners;18,27 and (iii) as the number of Basal Metabolic Rate (BMR) consumed, used as an indicator of
the effort intensity degree above the basal metabolism26,28.
The Kolgomorov-Smirnov test was used for testing data normality. Since variables were not normally dis-
tributed, all statistical analyses were performed by applying non-parametric statistical tests. The Mann-Whitney
U test was used to compare the energy consumption values estimated by using the accelerometer-derived
data and the relative running speed. Then, P-values were corrected for multiple comparisons by applying the
Benjamini-Hochberg procedure for decreasing the False Discovery Rate.The Sperman’s correlation test was
applied to analyze linear association between two continuous variables.
Data availability
All data generated or analysed during this study are included in this published article (and its Supplementary
Information File). Any other relevant data can be obtained from the corresponding author upon reasonable
request.
Received: 13 February 2019; Accepted: 15 January 2020;
Published: xx xx xxxx
References
1. Ahmadyar, B., Rüst, C. A., Rosemann, T. & Knechtle, B. Participation and performance trends in elderly marathoners in four of the
world’s largest marathons during 2004–2011. SpringerPlus 4, 465 (2015).
2. Aschmann, A., Knechtle, B., Onywera, V. O. & Nikolaidis, P. T. Pacing Strategies in the New York City Marathon - Does Nationality
of Finishers Matter? | Request PDF. Asian J. Sports Med. june, (2018).
3. Esteve-Lanao, J. et al. Is Marathon Training Harder than the Ironman Training? An ECO-method Comparison. Front. Physiol. 8, 298
(2017).
4. Mansour, S. G. et al. Kidney Injury and Repair Biomarkers in Marathon Runners. Am. J. Kidney Dis. Off. J. Natl. Kidney Found.,
https://doi.org/10.1053/j.ajkd.2017.01.045 (2017).
5. Vickers, A. J. & Vertosick, E. A. An empirical study of race times in recreational endurance runners. BMC Sports Sci. Med. Rehabil.
8, 26 (2016).
6. Hoffman, M. D., Ong, J. C. & Wang, G. Historical analysis of participation in 161 km ultramarathons in North America. Int. J. Hist.
Sport 27, 1877–1891 (2010).
7. Hoffman, M. D. & Fogard, K. Factors related to successful completion of a 161-km ultramarathon. Int. J. Sports Physiol. Perform. 6,
25–37 (2011).
8. Maratón de Valencia Fundación Trinidad Alfonso EDP, https://www.valenciaciudaddelrunning.com/maraton/ediciones-anteriores-
maraton/ (2019).
9. Gabbett, T. J. et al. The athlete monitoring cycle: a practical guide to interpreting and applying training monitoring data. Br. J. Sports
Med. 51, 1451–1452 (2017).
10. Szabo, A., Vega, R. D. L., Ruiz-BarquÍn, R. & Rivera, O. Exercise addiction in Spanish athletes: Investigation of the roles of gender,
social context and level of involvement. J. Behav. Addict. 2, 249–252 (2013).
11. Nikolaidis, P. T., Onywera, V. O. & Knechtle, B. Running Performance, Nationality, Sex, and Age in the 10-km, Half-Marathon,
Marathon, and the 100-km Ultramarathon IAAF 1999–2015. J. Strength Cond. Res. 31, 2189–2207 (2017).
12. Ainsworth, B. E. et al. Compendium of physical activities: an update of activity codes and MET intensities. Med. Sci. Sports Exerc.
32, S498–504 (2000).
13. Ainsworth, B. E. et al. Compendium of Physical Activities: a second update of codes and MET values. Med. Sci. Sports Exerc. 43,
1575–1581 (2011).
14. Strath, S. J. et al. Guide to the Assessment of Physical Activity: Clinical and Research Applications A Scientific Statement From the
American Heart Association. Circulation 01.cir.0000435708.67487.da, https://doi.org/10.1161/01.cir.0000435708.67487.da (2013).
15. Lazzer, S. et al. Factors affecting metabolic cost of transport during a multi-stage running race. J. Exp. Biol. 217, 787–795 (2014).
9
Scientific RepoRtS | (2020) 10:1523 | https://doi.org/10.1038/s41598-020-58492-8
www.nature.com/scientificreports
www.nature.com/scientificreports/
16. Helgerud, J., Støren, Ø. & Hoff, J. Are there differences in running economy at different velocities for well-trained distance runners?
Eur. J. Appl. Physiol. 108, 1099–1105 (2010).
17. di Prampero, P. E., Atchou, G., Brückner, J. C. & Moia, C. The energetics of endurance running. Eur. J. Appl. Physiol. 55, 259–266
(1986).
18. di Prampero, P. E. Factors limiting maximal performance in humans. Eur. J. Appl. Physiol. 90, 420–429 (2003).
19. Ainsworth, B. E. et al. Compendium of physical activities: classification of energy costs of human physical activities. Med. Sci. Sports
Exerc. 25, 71–80 (1993).
20. Montoye, H. J. et al. Estimation of energy expenditure by a portable accelerometer. Med. Sci. Sports Exerc. 15, 403–407 (1983).
21. Smith, M. P., Horsch, A., Standl, M., Heinrich, J. & Schulz, H. Uni- and triaxial accelerometric signals agree during daily routine, but
show differences between sports. Sci. Rep. 8, 15055 (2018).
22. Esliger, D. W. et al. Validation of the GENEA Accelerometer. Med. Sci. Sports Exerc. 43, 1085–1093 (2011).
23. Welch, W. A. et al. Classification accuracy of the wrist-worn gravity estimator of normal everyday activity accelerometer. Med. Sci.
Sports Exerc. 45, 2012–2019 (2013).
24. Menai, M. et al. Accelerometer assessed moderate-to-vigorous physical activity and successful ageing: results from the Whitehall II
study. Sci. Rep. 8, 45772 (2017).
25. Hernando, C. et al. Establishing cut-points for physical activity classification using triaxial accelerometer in middle-aged recreational
marathoners. PLOS One 13, e0202815 (2018).
26. Byrne, N. M., Hills, A. P., Hunter, G. R., Weinsier, R. L. & Schutz, Y. Metabolic equivalent: one size does not fit all. J. Appl. Physiol.
Bethesda Md 1985 99, 1112–1119 (2005).
27. Lazzer, S. et al. The energetics of ultra-endurance running. Eur. J. Appl. Physiol. 112, 1709–1715 (2012).
28. Abe, D., Fukuoka, Y. & Horiuchi, M. Economical Speed and Energetically Optimal Transition Speed Evaluated by Gross and Net
Oxygen Cost of Transport at Different Gradients. PloS One 10, e0138154 (2015).
29. Fudge, B. W. et al. Estimation of oxygen uptake during fast running using accelerometry and heart rate. Med. Sci. Sports Exerc. 39,
192–198 (2007).
30. McGregor, S. J., Busa, M. A., Yaggie, J. A. & Bollt, E. M. High resolution MEMS accelerometers to estimate VO2 and compare
running mechanics between highly trained inter-collegiate and untrained runners. PLOS ONE 4, e7355 (2009).
31. Strath, S. J., Bassett, D. R., Thompson, D. L. & Swartz, A. M. Validity of the simultaneous heart rate-motion sensor technique for
measuring energy expenditure. Med. Sci. Sports Exerc. 34, 888–894 (2002).
32. Kobsar, D., Osis, S. T., Hettinga, B. A. & Ferber, R. Classification accuracy of a single tri-axial accelerometer for training background
and experience level in runners. J. Biomech. 47, 2508–2511 (2014).
33. Boyd, L. J., Ball, K. & Aughey, R. J. Quantifying external load in Australian football matches and training using accelerometers. Int.
J. Sports Physiol. Perform. 8, 44–51 (2013).
34. Troiano, R. P., McClain, J. J., Brychta, R. J. & Chen, K. Y. Evolution of accelerometer methods for physical activity research. Br. J.
Sports Med. 48, 1019–1023 (2014).
35. de Almeida Mendes, M. et al. Calibration of raw accelerometer data to measure physical activity: A systematic review. Gait Posture
61, 98–110 (2018).
36. Cordero, M. J. A. et al. Accelerometer description as a method to assess physical activity in diferent periods of life; review. Nutr. Hosp.
29, 1250–1261 (2014).
37. Barnes, K. R. & Kilding, A. E. Running economy: measurement, norms, and determining factors. Sports Med. - Open 1, 8 (2015).
38. Nikolaidis, P. T., Rosemann, T., Cuk, I & Knechtle, B. Performance and Pacing of Age Groups in Half-Marathon and Marathon. J.
Enviromental Res. Public Health, https://doi.org/10.3390/ijerph16101777 (2019).
39. Nikolaidis, P. T., Ćuk, I. & Knechtle, B. Pacing of Women and Men in Half-Marathon and Marathon Races. Med. Kaunas Lith. 55
(2019).
40. Saunders, P. U., Pyne, D. B., Telford, R. D. & Hawley, J. A. Factors Affecting Running Economy in Trained Distance Runners. Sports
Med. 34, 465–485 (2004).
41. Barnes, K. R. & Kilding, A. E. Strategies to Improve Running Economy. Sports Med. 45, 37–56 (2015).
42. Thompson, M. A. Physiological and Biomechanical Mechanisms of Distance Specific Human Running Performance. Integr. Comp.
Biol. 57, 293–300 (2017).
43. Bossi, A. H. et al. Pacing Strategy During 24-Hour Ultramarathon-Distance Running. Int. J. Sports Physiol. Perform. 12, 590–596
(2017).
44. Knechtle, B., Rosemann, T., Zingg, M. A., Stiefel, M. & Rüst, C. A. Pacing strategy in male elite and age group 100 km ultra-
marathoners. Open Access J. Sports Med. 6, 71–80 (2015).
45. Stiles, V. H., Pearce, M., Moore, I. S., Langford, J. & Rowlands, A. V. Wrist-worn Accelerometry for Runners: Objective Quantification
of Training Load. Med. Sci. Sports Exerc. 50, 2277 (2018).
46. Lavie, C. J. & Milani, R. V. Metabolic equivalent (MET) inflation–not the MET we used to know. J. Cardiopulm. Rehabil. Prev. 27,
149–150 (2007).
47. Bellenger, C. R. et al. Optimization of Maximal Rate of Heart Rate Increase Assessment in Runners. Res. Q. Exerc. Sport 89, 322–331
(2018).
48. Suchert, V. et al. Prospective effects of pedometer use and class competitions on physical activity in youth: A cluster-randomized
controlled trial. Prev. Med. 81, 399–404 (2015).
49. Tudor-Locke, C. et al. Walking cadence (steps/min) and intensity in 21–40 year olds: CADENCE-adults. Int. J. Behav. Nutr. Phys. Act.
16, 8 (2019).
50. Kinnunen, H. et al. Training-induced changes in daily energy expenditure: Methodological evaluation using wrist-worn
accelerometer, heart rate monitor, and doubly labeled water technique. PloS One 14, e0219563 (2019).
51. Franklin, B. A. et al. Using Metabolic Equivalents in Clinical Practice. Am. J. Cardiol. 121, 382–387 (2018).
Acknowledgements
Current research could be carried out thanks to the collaboration of Fundación Trinidad Alfonso, Vithas-Nisa
Hospitals group and Sociedad Deportiva Correcaminos. Authors are also grateful to all the stuff involved in
the organization of the Valencia Fundación Trinidad Alfonso EDP 2016 Marathon, and all marathoners and
volunteers participating in this study.
Author contributions
C.H. and B.H. contributed to conception and design of the study, article drafting, and critical revision of the
article. C.H. and C.H. contributed to data curation, analysis and interpretation. C.H., I.M.-N., E.C.-B. and N.P.
contributed to data collection and critical revision of the article. C.H., I.M.-N. and E.C.-B. contributed to funding
acquisition.
10
Scientific RepoRtS | (2020) 10:1523 | https://doi.org/10.1038/s41598-020-58492-8
www.nature.com/scientificreports
www.nature.com/scientificreports/
competing interests
The authors declare no competing interests.
Additional information
Supplementary information is available for this paper at https://doi.org/10.1038/s41598-020-58492-8.
Correspondence and requests for materials should be addressed to C.H.
Reprints and permissions information is available at www.nature.com/reprints.
Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and
institutional affiliations.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International
License, which permits use, sharing, adaptation, distribution and reproduction in any medium or
format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Cre-
ative Commons license, and indicate if changes were made. The images or other third party material in this
article are included in the article’s Creative Commons license, unless indicated otherwise in a credit line to the
material. If material is not included in the article’s Creative Commons license and your intended use is not per-
mitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the
copyright holder. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/.
© The Author(s) 2020
| Estimation of energy consumed by middle-aged recreational marathoners during a marathon using accelerometry-based devices. | 01-30-2020 | Hernando, Carlos,Hernando, Carla,Martinez-Navarro, Ignacio,Collado-Boira, Eladio,Panizo, Nayara,Hernando, Barbara | eng |
PMC9192635 | 1
Vol.:(0123456789)
Scientific Reports | (2022) 12:9749
| https://doi.org/10.1038/s41598-022-13965-w
www.nature.com/scientificreports
Differences in stress response
between two altitudes assessed
by salivary cortisol levels
within circadian rhythms
in long‑distance runners
Katsuhiko Tsunekawa*, Kazumi Ushiki, Larasati Martha, Asuka Nakazawa, Rika Hasegawa,
Risa Shimizu, Nozomi Shimoda, Akihiro Yoshida, Kiyomi Nakajima, Takao Kimura &
Masami Murakami
There are conflicting reports regarding the efficacy of cortisol as a stress marker in altitude training
due to the influence of the circadian rhythm. This study aimed to verify whether the automated
measurement of salivary cortisol concentration via sequential sampling could detect the differences
in exercise stress between two altitudes. We enrolled 12 elite female long‑distance runners living
near sea level. For the first higher‑altitude camp, the runners lived at 1800 m and trained at 1700 m
for 7 days. For the second lower‑altitude camp, they lived at 1550 m and trained at 1300 m for
7 days. Their saliva was sequentially collected on the last 2 days during each camp which involved
different intensity exercises in the morning and afternoon. The salivary cortisol concentrations were
measured using electrochemiluminescence immunoassay. Before dinner, the basal salivary cortisol
concentrations were significantly higher in the higher‑altitude camp. The rate of change in the salivary
cortisol concentration during the morning exercise was significantly higher in the higher‑altitude camp
than in lower‑altitude camp (p = 0.028) despite the same exercise programs and intensities. Salivary
cortisol level measurements during the athletes’ circadian rhythms could detect the differences in
acclimatization and exercise stress between two altitudes.
Abbreviations
ACTH
Adrenocorticotropic hormone
ECLIA
Electrochemiluminescence immunoassay
ELISA
Enzyme-linked immunosorbent assay
RPE
Rating of perceived exertion
SpO2
Oxygen saturation
VO2 max
Maximal oxygen consumption
Elite athletes in various sports often train at high altitudes to improve their performance when they return to
lower altitudes. At high altitudes with low atmospheric pressure and low oxygen concentrations, the amount
of red blood cells and their oxygen-carrying capacity are enhanced by an increase in circulating erythropoietin
concentration due to hypoxia-inducible factors1. However, a hypoxia during high-altitude training causes exces-
sive stress and decreases an athletes’ performance, resulting in poor acclimatization2. Therefore, monitoring of
physical and psychological stress during exercise training at high-altitude camps may help assess maladaptation.
Cortisol is an important biomarker that is secreted into the circulating plasma from the adrenal cortex via
the hypothalamus–pituitary axis as an acute response to stress including exercise3. Serum cortisol concentra-
tions are increased by moderate- to high-intensity exercise, but not by low-intensity exercise at less than 40% of
athletes’ maximal oxygen consumptions (VO2 max)4,5. However, there have been conflicting reports regarding the
efficacy of cortisol as a stress marker during high-altitude training6–11, which may be due to the influence of the
OPEN
Department of Clinical Laboratory Medicine, Gunma University Graduate School of Medicine, 3-39-22 Showa-machi
Maebashi, Gunma 371-8511, Japan. *email: [email protected]
2
Vol:.(1234567890)
Scientific Reports | (2022) 12:9749 |
https://doi.org/10.1038/s41598-022-13965-w
www.nature.com/scientificreports/
circadian rhythm. Changes in the serum cortisol concentrations resulting from exercise are greater in the evening
than in the morning because of cortisol circadian rhythm12. Thus, to accurately evaluate exercise-induced stress
using cortisol, it should be measured throughout day. To achieve this, continuous sampling and more efficient
cortisol measurement are needed. Serum cortisol concentrations are measured as total hormone conjugated to
corticosteroid-binding globulin, whereas salivary cortisol concentrations are measured as free hormones inde-
pendent of salivary flow rates13. Saliva is also advantageous to its ease of collection in the absence of medical
professional staff and without the stress of venipuncture14. Conventionally, the salivary cortisol concentration has
been manually measured using an enzyme-linked immunosorbent assay (ELISA), which this makes it difficult to
measure a large number of samples compared with automated methods. Recently, the automated electrochemi-
luminescence immunoassay (ECLIA) used for serum cortisol concentration measurement was applied to saliva,
and the salivary cortisol concentrations measured by ECLIA showed a significantly positive correlation with those
measured by liquid chromatography-tandem mass spectrometry15 and with conventional ELISA16. Moreover,
we reported that sequential saliva collection and automated ECLIA-based salivary cortisol measurements could
detect the exercise-induced stress within the circadian rhythm in female long-distance runners16. In this study,
the differences in the rate of change in salivary cortisol concentrations resulting from various exercise intensities
could be compared at the same time on different days, even in the early morning.
Training at altitudes of 500–2000 m, defined as low altitude, causes less stress on athletes than training at
moderate (2000–3000 m) and high (3000–5500 m) altitudes17. In contrast, VO2 max of endurance-trained athletes
decreased significantly beginning at 300 m above sea level and continued to decrease linearly by approximately
7% for every 1000 m ascended18,19. While these reports suggest that there may be differences in athletes’ stress
responses at different altitudes, even at low altitudes, few biomarkers have ever detected this difference. We
hypothesized that the automated measurement of salivary cortisol concentrations throughout the athletes’ cir-
cadian rhythms via sequential saliva sampling would enable the assessment of the differences in stress induced
by training camps at different altitudes, including those at low altitudes. If proven, this method can help in the
prevention of excessive stress, and foster the development of exercise programs in several altitude environments
for athletes. The present study expands on our previous study by verifying whether cortisol concentration meas-
urement via continuous saliva collection during the circadian rhythms could adequately detect the differences in
acclimatization and stress responses resulting from exercise at different altitudes in female long-distance runners.
Methods
Participants.
This study was conducted in accordance to the Declaration of Helsinki, and the protocol was
approved by the ethics committee of the Gunma University Graduate School of Medicine (Approval number
HS2018-140). All the study participants provided written informed consent before being included in the study.
We enrolled 12 Japanese elite female long-distance runners. All of them lived in the same dormitory before
the first training camp and stayed in the same hotels during camps. Their living conditions, such as wakeup time,
meal time, bedtime, and meal content, were standardized before and during the training camps16. Figure 1A
presents the schedules of the pre-camp and the two training camps. This altitude training was a study without
invasive interventions, because it was usually conducted to improve the condition of the runners. The runners
lived and trained near sea level (150 m) before the first training camp. Then, for the first training camp simulat-
ing the higher-altitude camp at low altitudes, they lived at 1800 m and trained at 1700 m twice a day, morning
and afternoon, for 7 days. Afterwards, for the second training camp simulating the lower-altitude camp at low
altitudes, they lived at 1550 m and trained at 1300 m twice a day, morning and afternoon, for 7 days. They then
returned to near sea level. We sequentially collected saliva from these runners on the last 2 days during each camp
which involved different exercise intensities in the morning and afternoon, modified as previously described16.
Figure 1B details the relation between runner’s altitudes and saliva collection times during 2 consecutive days
during each training camp. On both days, saliva samples were collected at eight time points: upon waking
(05:00), before morning exercise (06:00), after morning exercise (07:00), before breakfast (08:00), before lunch
(12:00), before afternoon exercise (15:00), after afternoon exercise (16:00), and before dinner (18:00), as described
previously16. On each training day at both camps, no differences were observed in the meteorological conditions;
the temperature was around 20 °C and the relative humidity was 50–60% with fine weather. The runners were
subjected to the following exercise program in the higher-altitude camp: 40-min fixed running in the morn-
ing and 50-min fixed running in the afternoon on day 1 (Higher-day 1); 8000-m fixed-distance running in the
morning and uphill interval training with 8 sets of 200-m fast uphill running and light jogging in the evening
on day 2 (Higher-day 2). The runners were subjected to the following exercise program in the lower-altitude
camp: 50-min fixed running in the morning and 60-min fixed running in the afternoon on day 1 (Lower-day
1); 8000-m fixed-distance running in the morning and uphill interval training with 5 sets of 200-m fast uphill
running and light jogging in the evening on day 2 (Lower-day 2). The runners drank enough water to prevent
dehydration during these trainings.
Physical examinations.
The participants were weighed, and their body mass indexes were calculated as
the weight divided by the squared height (kg/m2). After conducting interviews, no runners were found to use
any medications or supplements. The runners used the Apple Watch Series 3 (Apple Japan Inc., Tokyo, Japan)
during the camps. This allowed for the measurement of maximum pulse rate during each exercise and resting
pulse rate at awakening and before dinner. The distance and duration of running during each exercise session
was measured, and the running velocity was calculated as the distance divided by the duration (m/min)15. The
Borg Rating of Perceived Exertion (RPE) scale20 was utilized to measure the runner’s subjective exertion, breath-
lessness and fatigue after exercise. The runner’s RPE was scored using a scale ranging from 6 to 20 and used in
the analysis as a Borg scale score.
3
Vol.:(0123456789)
Scientific Reports | (2022) 12:9749 |
https://doi.org/10.1038/s41598-022-13965-w
www.nature.com/scientificreports/
Saliva collections and measurements of salivary cortisol concentrations.
Sample collections and
salivary cortisol concentration measurements were performed according to a previous study16. The runners
were not allowed to brush their teeth, chew gum, or consume any food or drink except water, 15 min before the
sample collection. Saliva samples were collected using Salivette® cotton swabs (Sarstedt, Nümbrecht, Germany),
centrifuged (1,500 × g) at 4 °C for 10 min, then immediately stored at − 80 °C until analysis. ECLIA measure-
ments of salivary cortisol concentrations were performed using the Elecsys Cortisol II on the Cobas 8000 sys-
tem (Roche Diagnostics K.K, Tokyo, Japan)15,16. The intra- and inter-assay coefficients of variation for salivary
cortisol were 4.1% and 4.6%, respectively. The rate of change in the salivary cortisol concentration by exercise
was calculated as the salivary cortisol concentration after exercise divided by the salivary cortisol concentration
before exercise (%)16.
Statistical analysis.
The results of each measurement are expressed as the median values and correspond-
ing 25th–75th percentile ranges. The Wilcoxon signed-rank test was utilized to identify statistically significant
differences in variables between two different time points. A p value of < 0.05 was considered statistically signifi-
cant. All statistical analyses were performed using SPSS Statistics, version 26.0 (IBM Corp., Armonk, NY, USA).
Ethics approval and consent to participate.
Written informed consent was obtained from all par-
ticipants. This study was approved by the ethics committee of Gunma University Graduate School of Medicine
(Approval number HS2018-140). All measurements were carried out by trained athletes and in accordance with
the Declaration of Helsinki.
Results
Running intensity of each exercise program.
Table 1 presents the characteristics of the participating
female long-distance runners, whereas Table 2 presents the different running intensities of each exercise pro-
gram during the two training camps. Because the exercise programs on days 1 and 2 were similar between the
two camps, the exercise intensities of each program were compared. During the morning exercise on day 1, at
the higher-altitude camp, the running velocity was significantly higher (p = 0.015) and the running distance and
Figure 1. The study design of the first and second training camps of 12 female long-distance runners. The
altitudes at which the runners lived and trained during the two training camps and pre- and post-camp (A). The
schema of altitudes and saliva sampling time in runners on the last 2 days during each training camps (B). The
downwards arrows denote the saliva sampling from the runners, and the tips of the arrows denote the altitudes
at the time of sampling at both camps. Higher-day 1, day 1 at the higher-altitude camp; Higher-day 2, day 2 at
the higher-altitude camp; Lower-day 1, day 1 at the lower-altitude camp; Lower-day 2, day 2 at the lower-altitude
camp.
4
Vol:.(1234567890)
Scientific Reports | (2022) 12:9749 |
https://doi.org/10.1038/s41598-022-13965-w
www.nature.com/scientificreports/
Borg scale scores were significantly lower (running distance, p = 0.029; Borg scale score, p = 0.047) than those at
the lower-altitude camp. During the afternoon exercise on day 1, the running velocity was significantly higher
at the higher-altitude camp (p = 0.003), but no differences were observed in other parameters between the two
camps. During the morning exercise on day 2, the maximum pulse rate was significantly lower at the higher-alti-
tude camp (p = 0.029), but no differences were observed in other parameters between the two camps. However,
during the afternoon exercise on day 2, the running distance, Borg scale score, and maximum pulse rate were
significantly higher at the higher-altitude camp (running distance, p = 0.002; Borg scale score, p = 0.005; maxi-
mum pulse rate, p = 0.036). When comparing the exercise intensities between the morning and afternoon on day
1, the running distance, running velocity, and Borg scale score were significantly higher during the afternoon
exercise at the higher-altitude camp (running distance, p = 0.002; running velocity, p = 0.002; Borg scale score,
p = 0.010), whereas the running distance was significantly higher during the afternoon exercise at the lower-
altitude camp (p = 0.004). On day 2, the running distance and Borg scale score were significantly higher and the
running velocity was significant lower during the afternoon exercise compared with those during the morning
exercise at the higher-altitude camp (running distance, p = 0.002; Borg scale score, p = 0.004; running velocity,
p = 0.002), whereas the running distance and velocity were significantly lower during the afternoon exercise at
the lower-altitude camp (running distance, p = 0.002; running velocity, p = 0.003).
Changes in the salivary cortisol concentrations in response to exercise within the circadian
rhythms in each camp.
Figure 2 presents the changes in the salivary cortisol concentrations in response
to exercise during the last 2 days at both camps. The salivary cortisol concentrations peaked after waking and
promptly decreased on both days at both camps. Within these circadian rhythms, the salivary cortisol concentra-
tions significantly decreased after the morning exercise on both days at both camps but significantly increased
after the afternoon exercise on 2 days at only the higher-altitude camp. These concentrations reached their low-
est levels before dinner on both days at both camps. Table 3 presents that the differences in the resting pulse
rates and salivary cortisol concentrations between the higher- and lower-altitude camps. The resting pulse rate
before dinner was significantly higher on day 2 at the higher-altitude camp than on day 2 at the lower-altitude
Table 1. Characteristics of female long-distance runners. Data are expressed as median (25th–75th
percentile).
Characteristics
Value
Number
12
Age (year)
23.5 (19.5–26.0)
Height (cm)
160.0 (155.5–164.5)
Weight (kg)
45.5 (41.0–47.5)
Body mass index (kg/m2)
17.4 (17.0–17.9)
Table 2. Running intensities of the exercise programs performed by female long-distance runners during the
two camps. Data are expressed as median (25th–75th percentile). *p < 0.05 and **p < 0.01 comparing variables
between the higher- and lower-altitude camps using the Wilcoxon signed-rank test. pday1 morning exercise vs.
afternoon exercise on day 1 using the Wilcoxon signed-rank test. pday2 morning exercise vs. afternoon exercise
on day 2 using the Wilcoxon signed-rank test.
Altitude camp
Day 1
pday1
Day 2
pday2
Morning exercise
Afternoon exercise
Morning exercise
Afternoon
exercise
Exercise program
Higher
40-min fixed
running
50-min fixed
running
8000-m fixed
running
Uphill interval
training
Lower
50-min fixed
running
60-min fixed
running
8000-m fixed
running
Uphill interval
training
Running distance
(m)
Higher
9700 (9250–
10,150)*
12,790 (12,300–
13,450)
0.002
8000 (8000–8000)
12,000 (9800–
14,000)**
0.002
Lower
11,585 (10,700–
12,000)
13,145 (12,445–
14,000)
0.004
8000 (8000–8000)
4700 (4200–5250)
0.002
Running velocity
(m/min)
Higher
242.5 (224.8–
247.7)*
252.0 (246.0–
263.0)*
0.002
235.3 (228.6–
238.9)
178.0 (153.9–
215.4)
0.002
Lower
224.2 (205.0–
236.8)
219.1 (207.5–
233.3)
0.477
235.3 (228.6–
236.5)
168.0 (125.0–
213.8)
0.003
Borg scale score
Higher
13.0 (12.0–15.0)*
15.0 (13.0–15.5)
0.010
13.0 (12.0–15.0)
16.0 (14.0–17.5)**
0.004
Lower
14.0 (13.0–15.5)
13.5 (12.0–15.0)
0.119
13.0 (12.0–15.0)
13.0 (12.5–15.0)
0.524
Maximum pulse
rate (beat/min)
Higher
175 (159–189)
182 (159–201)
0.182
170 (152–184)*
179 (165–192)*
0.167
Lower
168 (153–198)
168 (158–181)
0.937
196 (164–210)
163 (154–187)
0.050
5
Vol.:(0123456789)
Scientific Reports | (2022) 12:9749 |
https://doi.org/10.1038/s41598-022-13965-w
www.nature.com/scientificreports/
Figure 2. Changes in salivary cortisol concentrations in response to each exercise within the circadian rhythm
on 2 consecutive days during the higher-altitude camp (A) and lower-altitude camp (B). The white box plots
denote the cortisol concentration at the higher-altitude camp, whereas the gray box plots represent those at the
lower-altitude camp. The gray dot squares denote the time of exercise at each camp. The significant differences
between two time points of each exercise for runners were analyzed using the Wilcoxon signed-rank test.
Higher-day 1, day 1 at the higher-altitude camp; Higher-day 2, day 2 at the higher-altitude camp; Lower-day 1,
day 1 at the lower-altitude camp; Lower-day 2, day 2 at the lower-altitude camp.
Table 3. Comparison of the variables between the higher- and lower-altitude camps in runners. Data are
expressed as median (25th–75th percentile). *p < 0.05 and **p < 0.01 comparing variables with day 1 at the
lower-altitude camp using the Wilcoxon signed-rank test. † p < 0.05 and ††p < 0.01 comparing variables with day
2 at the lower-altitude camp using the Wilcoxon signed-rank test.
Higher-altitude camp
Lower-altitude camp
Day 1
Day 2
Day 1
Day 2
Resting pulse rate at awakening (beat/min)
50 (46–55)
49 (44–52)
48 (44–52)
50 (42–57)
Resting pulse rate before dinner (beat/min)
65 (57–70)
65 (57–71)†
56 (52–68)
54 (48–63)
Salivary cortisol at awakening (μg/dL)
0.41 (0.37–0.45)†
0.36 (0.31–0.42)†
0.40 (0.33–0.43)
0.43 (0.39–0.50)
Salivary cortisol at peak(μg/dL)
0.57 (0.44–0.61)
0.51 (0.45–0.61)
0.49 (0.43–0.59)
0.55 (0.42–0.64)
Salivary cortisol before dinner (μg/dL)
0.15 (0.09–0.18)* ††
0.11 (0.09–0.25)* ††
0.08 (0.06–0.11)
0.05 (0.05–0.07)
6
Vol:.(1234567890)
Scientific Reports | (2022) 12:9749 |
https://doi.org/10.1038/s41598-022-13965-w
www.nature.com/scientificreports/
camp (p = 0.021). The salivary cortisol concentrations upon waking were significantly lower on both days at the
higher-altitude camp than on day 2 at the lower-altitude camp (Higher-day 1 vs. Lower-day 2, p = 0.038; Higher-
day 2 vs. Lower-day 2, p = 0.025). The concentrations before dinner were also significantly higher on both days
at the higher-altitude camp than those on both days at the lower-altitude camp (Higher-day 1 vs. Lower-day 1,
p = 0.026; Higher-day 1 vs. Lower-day 2, p = 0.005; Higher-day 2 vs. Lower-day 1, p = 0.012; Higher-day 2 vs.
Lower-day 2, p = 0.003).
Rate of change in the salivary cortisol concentrations resulting from exercise in each
camp.
Figure 3 presents the comparison of the rate of change in the salivary cortisol concentrations after
each exercise at both camps. The rates of change in salivary cortisol concentrations were significantly lower dur-
ing the morning exercise than during the afternoon exercise on both days at each camp (Higher-day 1, p = 0.002;
Higher-day 2, p = 0.002; Lower-day 1, p = 0.005; Lower-day 2, p = 0.008; Fig. 3A,B). After the morning exercise,
the rate of change in the salivary cortisol concentrations was significantly higher on day 2 at the higher-altitude
Figure 3. Comparison of the rate of change in the salivary cortisol concentration resulting from exercise
between the morning and afternoon time points on days 1 and 2 at the higher-altitude camp (A), lower-altitude
camp (B), between the morning time points on days 1 and 2 at the higher- and lower-altitude camps (C), and
between the afternoon time points on days 1 and 2 at the higher- and lower-altitude camps (D). The white box
plots denote the rates of change in the cortisol concentration at the higher-altitude camp, whereas the gray box
plots represent those at the lower-altitude camp. The significant differences between two time points of exercise
for runners were analyzed using the Wilcoxon signed-rank test. Higher-day 1, day 1 at the higher-altitude camp;
Higher-day 2, day 2 at the higher -altitude camp; Lower-day 1, day 1 at the lower-altitude camp; Lower-day 2,
day 2 at the lower-altitude camp.
7
Vol.:(0123456789)
Scientific Reports | (2022) 12:9749 |
https://doi.org/10.1038/s41598-022-13965-w
www.nature.com/scientificreports/
camp than on day 2 at the lower-altitude camp (p = 0.028; Fig. 3C). After the afternoon exercise, the rate of
change in the salivary cortisol concentrations was significantly higher on both days at the higher-altitude camp
than on both days at the lower-altitude camp (Higher-day 1 vs. Lower-day 1, p = 0.012; Higher-day 2 vs. Lower-
day 2, p = 0.003; Fig. 3D).
Discussion
In this study, we demonstrated whether the stress responses of runners in training camps at different altitudes
could be evaluated via sequential saliva collection and automated salivary cortisol measurement. These methods
were able to detect the basal levels and exercise-induced changes in the salivary cortisol within the runners’ cir-
cadian rhythms at each altitude camp. The basal salivary cortisol concentrations before dinner were significantly
higher at the higher-altitude camp than at lower-altitude camp. The rate of change in the salivary cortisol concen-
trations during the afternoon exercise on days 1 and 2 and the indicators of exercise intensity were significantly
higher at the higher-altitude camp than at the lower-altitude camp. Moreover, the rate of change in the salivary
cortisol concentrations during the morning exercise on day 2 was significantly higher at the higher-altitude camp
than at lower-altitude camp; no differences were observed in the exercise programs and intensities, such as the
running distances, velocities, and Borg scale scores.
There have been contradictory reports with regard to the effects of altitude training on cortisol secretions. In
male elite climbers, the resting serum cortisol and plasma adrenocorticotropic hormone (ACTH) levels taken
at 07:00–07:30 did not change at 5200 m extreme altitude camp compared with those at sea level6. In male and
female elite skiers, no significant differences were observed in resting salivary cortisol concentrations taken
at 07:00–08:00 between a control group training and living at an altitude of 1200 m and another group train-
ing at 1200 m but living at a simulated altitudes of 2500 m, 3000 m, and 3500 m for 6 days in hypoxic rooms7.
In contrast, the basal concentrations of serum cortisol in the morning after 3–4 days at 4350 m were elevated
compared with those at sea level in healthy men, but this was not statistically significant8. In the present study, a
significant difference was observed in the basal salivary cortisol concentrations before dinner between the two
camps with an altitude difference of approximately 300 m, even at the relatively low altitudes of 1300–1800 m.
These differences may be due to the fact that the lowest cortisol levels were compared in the evening, whereas
the previous studies compared the concentrations in the morning6–8. Another reason may be that the athletes
were acclimatized to these altitudes when transitioning from the first higher-altitude camp to the second lower-
altitude camp. Overtraining syndrome causes a reduced cortisol response to exercise and changes in the circa-
dian rhythms of cortisol, including low resting levels and peak loss after waking21,22. In the present study, the
runners’ salivary cortisol concentrations were lower in the evening, but peaked after waking and increased after
high-intensity exercise. The low levels of salivary cortisol in the evenings indicated that the runners were not
suffering from overtraining syndrome but rather were able to adapt to the higher altitudes. Although further
analysis under higher-altitude conditions is required, it may be more useful to evaluate the cortisol levels in the
evening rather than in the morning using serum or saliva for assessing the acclimatization to the several altitude
environments among athletes.
Regarding the acute response to exercise, the increase in cortisol levels after interval training at 09:00 exhibited
an insignificant trend toward higher values at an altitude of 1800 m when compared with that observed at near
sea levels in highly trained endurance athletes9. Another study found that the serum cortisol levels significantly
increased after resistance training at 70% of the maximum strength under 13% hypoxic conditions from 08:00
to 11:30 but not under normoxic conditions in healthy male subjects10. Conversely, the serum cortisol levels did
not change after resistance training at 50% of the maximum strength under hypoxic and normoxic conditions11.
In the present study, the rates of change in the salivary cortisol concentration after the morning exercise were
significantly lower than those after the afternoon exercise on both days at each altitude camp due to the influence
of the circadian rhythm, which was validated by previous reports12,16. In contrast , the comparison of the rates of
change in cortisol concentration during exercise at the same time on different days, whether in the morning or
in the afternoon, was effective in evaluating the stress responses, as previously described16. In the present study,
we were able to detect the difference in the rates of change in the salivary cortisol concentration during exercise
with the same program at different altitudes, even in the morning. Moreover, we could detect the increase in
stress due to the differences in altitude and exercise intensity, more clearly in the afternoon, by assessing the
rate of change in salivary cortisol concentrations. It was revealed that an increase in the training altitude of
approximately 400 m at the low altitude of 1300–1700 m with high-intensity exercise resulted in an increase in
cortisol secretion in both morning and afternoon. This result validates the previous studies, where the serum
cortisol concentrations were acutely elevated by high-intensity exercise with higher VO2 max
4,5. However, we did
not measure the oxygen saturation (SpO2) or VO2 max as the oxygen tolerability of runners for different exercise
intensities. Future studies should evaluate the relationships between the changes in salivary cortisol concentra-
tions and these oxygen tolerability markers in response to endurance exercises at different altitudes. Addition-
ally, the range of the runners’ salivary cortisol concentrations was broad, especially after afternoon exercise on
day 2 at the higher-altitude camp, shown in Fig. 2. This broad range suggested individual differences in stress
responses induced by the altitude training in the runners. Therefore, the cortisol concentration measurement
technique described in this study which uses continuous saliva collection during the circadian rhythms would
be more useful as a personalized conditioning tool to detect the differences in stress responses under various
environments for an individual athlete, rather than as a statistical analysis tool for a large number of athletes.
This study has several limitations. First, the sample size was relatively small. We focused on enrolling well-
trained female runners with standardized meal and sleep times during the two consecutive camps. Second, the
sequential saliva collections and measurements of salivary cortisol concentration were not performed at sea level.
The temperatures were around 20 °C during both altitude camps, but the temperature during the same period
8
Vol:.(1234567890)
Scientific Reports | (2022) 12:9749 |
https://doi.org/10.1038/s41598-022-13965-w
www.nature.com/scientificreports/
was as high as 30–40 °C at near sea level where the runners lived and trained. Previous research found that the
salivary cortisol levels detected in a maximal progressive test using a cycle ergometer were significantly higher
under the hot conditions (40 °C) than under normal conditions (22 °C) in nine young healthy men23. Therefore, it
was impossible in this study to collect saliva at near sea level without the stress resulting from high temperatures.
Future study should utilize hypoxic rooms, in which the environmental conditions, including temperature, are
standardized. Additionally, more runners should be enrolled, specifically males.
Conclusions
Measurement of the salivary cortisol levels within the circadian rhythm led to the detection of the changes in the
stress response due to the same intensity exercise at different altitudes, even in the morning. Additionally, evening
resting salivary cortisol levels can be used to assess athletes’ acclimatization to high altitudes. The combination
of sequential saliva collection and automated cortisol measurements may be useful for assessing adaptation dis-
orders and excessive exercise stress, and also may help develop adequate altitude training programs for athletes.
Data availability
The datasets used and/or analyzed during the current study are available from the corresponding author on
reasonable request.
Received: 3 February 2022; Accepted: 31 May 2022
References
1. Wang, G. L. et al. Hypoxia-inducible factor 1 is a basic-helix-loop-helix-PAS heterodimer regulated by cellular O2 tension. Proc.
Natl. Acad. Sci. U.S.A. 92, 5510–5514 (1995).
2. Flaherty, G., O’Connor, R. & Johnston, N. Altitude training for elite endurance athletes: A review for the travel medicine practi-
tioner. Travel Med. Infect. Dis. 14, 200–211 (2016).
3. Hackney, A. C. Stress and the neuroendocrine system: The role of exercise as a stressor and modifier of stress. Expert Rev. Endo-
crinol. Metab. 1, 783–792 (2006).
4. Davies, C. T. & Few, J. D. Effects of exercise on adrenocortical function. J. Appl. Physiol. 35, 887–891 (1973).
5. Hill, E. E. et al. Exercise and circulating cortisol levels: The intensity threshold effect. J. Endocrinol. Invest. 31, 587–591 (2008).
6. Benso, A. et al. Endocrine and metabolic responses to extreme altitude and physical exercise in climbers. Eur. J. Endocrinol. 157,
733–740 (2007).
7. Tiollier, E. et al. Living high-training low altitude training: Effects on mucosal immunity. Eur. J. Appl. Physiol. 94, 298–304 (2005).
8. Richalet, J. P., Leteournel, M. & Souberbielle, J. C. Effects of high-altitude hypoxia on the hormonal response to hypothalamic
factors. Am. J. Physiol. Regul. Integr. Comp. Physiol. 299, R1685-1692 (2010).
9. Niess, A. M. et al. Evaluation of stress responses to interval training at low and moderate altitudes. Med. Sci. Sports Exerc. 35,
263–269 (2003).
10. Kon, M. et al. Effects of acute hypoxia on metabolic and hormonal responses to resistance exercise. Med. Sci. Sports Exerc. 42,
1279–1285 (2010).
11. Kon, M. et al. Effects of low-intensity resistance exercise under acute systemic hypoxia on hormonal responses. J. Strength Cond.
Res. 26, 611–617 (2012).
12. Chtourou, H. et al. The effect of time of day on hormonal responses to resistance exercise. Biol. Rhythm. Res. 45, 247–256 (2014).
13. Peters, J. R. et al. Salivary cortisol assays for assessing pituitary-adrenal reserve. Clin. Endocrinol. 17, 583–592 (1982).
14. Hofman, L. F. Human saliva as a diagnostic specimen. J. Nutr. 131, 1621S-1625S (2001).
15. Gagnon, N. et al. Establishment of reference intervals for the salivary cortisol circadian cycle, by electrochemiluminescence
(ECLIA), in healthy adults. Clin. Biochem. 54, 56–60 (2018).
16. Ushiki, K. et al. Assessment of exercise-induced stress by automated measurement of salivary cortisol concentrations within the
circadian rhythm in Japanese female long-distance runners. Sports Med. Open. 6, 38 (2020).
17. Stellingwerff, T. et al. Nutrition and altitude: Strategies to enhance adaptation, improve performance and maintain health: A nar-
rative review. Sport Med. 49, 169–184 (2019).
18. Wehrlin, J. P. & Hallén, J. Linear decrease in.VO2max and performance with increasing altitude in endurance athletes. Eur. J. Appl.
Physiol. 96, 404–412 (2006).
19. Chapman, R. F. et al. Altitude training considerations for the winter sport athlete. Exp. Physiol. 95, 411–421 (2010).
20. Borg, G. A. V. Psychophysical bases of perceived exertion. Med. Sci. Sports Exerc. 14, 377–381 (1982).
21. Urhausen, A. & Kindermann, W. Diagnosis of overtraining: What tools do we have?. Sports Med. 32, 95–102 (2002).
22. Cadegiani, F. A. & Kater, C. E. Novel causes and consequences of overtraining syndrome: The EROS-DISRUPTORS study. BMC
Sports Sci. Med. Rehabil. 11, 21 (2019).
23. Silva, R. P. M. et al. The influence of a hot environment on physiological stress responses in exercise until exhaustion. PLoS ONE
14, e0209510 (2019).
Acknowledgements
We are grateful to Kenichi Morikawa, Mai Murata, Mayumi Nishiyama, and Tomoyuki Aoki for providing tech-
nical assistance and helpful discussions.
Author contributions
K.T. participated in the collection and analysis of data and manuscript writing, reviewing and editing. K.U., L.M.,
A.N., R.H., R.S., N.S., A.Y., N.K., and T.K. participated in data collection and analysis. M.M. participated in con-
ception of the study, supervision, and manuscript editing. All authors read and approved the final manuscript.
Funding
This work was supported by the Ministry of Education, Culture, Sports, Science, and Technology of Japan [Grant
number 18K07406].
9
Vol.:(0123456789)
Scientific Reports | (2022) 12:9749 |
https://doi.org/10.1038/s41598-022-13965-w
www.nature.com/scientificreports/
Competing interests
The authors declare no competing interests.
Additional information
Correspondence and requests for materials should be addressed to K.T.
Reprints and permissions information is available at www.nature.com/reprints.
Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and
institutional affiliations.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International
License, which permits use, sharing, adaptation, distribution and reproduction in any medium or
format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the
Creative Commons licence, and indicate if changes were made. The images or other third party material in this
article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the
material. If material is not included in the article’s Creative Commons licence and your intended use is not
permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from
the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/.
© The Author(s) 2022
| ||||
PMC6436800 | RESEARCH ARTICLE
Anthropometry-driven block setting improves
starting block performance in sprinters
Valentina Cavedon1☯, Marco SandriID1, Mariola Pirlo2, Nicola Petrone3, Carlo Zancanaro1,
Chiara MilaneseID1☯*
1 Department of Neurosciences, Biomedicine and Movement Sciences, University of Verona, Verona, Italy,
2 OMR Automotive, Brescia, Italy, 3 Department of Industrial Engineering, University of Padua, Padua, Italy
☯ These authors contributed equally to this work.
* [email protected]
Abstract
This study tested the effect of two block setting conditions i.e., the usual block setting [US]
and an anthropometry-driven block setting [AS] on the kinematic and kinetic parameters of
the sprint start. Furthermore, we verified whether this effect is influenced by the relative
lengths of the sprinter’s trunk and lower limbs i.e., the Cormic Index by subdividing sprinters
into brachycormic, metricormic and macrocormic groups. Forty-two sprinters performed 6
maximal-effort 10 m sprints using the US and AS conditions. Dynamometric starting blocks
measured forces generated by the sprinters. The times at 5 m and 10 m in the sprint trials
were measured with photocells. Results showed that the anteroposterior block distances
were significantly different between the two conditions (P<0.001). Across the sample, the
horizontal block velocity, the rear peak force, the rear force impulse, the total force impulse,
the horizontal block power, the ratio of horizontal to resultant impulse in the rear block, the
first and second step lengths and the times at 5 m and 10 m improved in AS vs. US (P values
from 0.05 to 0.001). Considering the interaction between the block setting condition and the
Cormic Index, the rear peak force and the rear force impulse were significantly increased in
the metricormic and brachycormic groups (P0.001) and the metricormic group (P<0.001),
respectively. Kinetic variables in the rear block and the difference (Delta) in the front block/
starting line distance between US and AS were related with each other (Adjusted R2 values
from 0.07 to 0.36). In conclusion, AS was associated with improvement in the kinematic
and kinetic parameters of the sprint start performance vs. US; however, AS is apparently
best suited for metricormic sprinters. Further work is needed to verify how the sprint start
kinetic and kinematic parameters are related to the front block/starting line distance and
whether a block setting driven by the sprinter’s Cormic Index is able to improve sprint start
performance.
Introduction
In track sprints, the success of the sprint start performance depends on the ability of the
sprinter to generate a large impulse over the shortest time and reach the highest running speed
as soon as possible [1–3]. This phase is especially important in the 100 m sprint [4–7].
PLOS ONE | https://doi.org/10.1371/journal.pone.0213979
March 27, 2019
1 / 20
a1111111111
a1111111111
a1111111111
a1111111111
a1111111111
OPEN ACCESS
Citation: Cavedon V, Sandri M, Pirlo M, Petrone N,
Zancanaro C, Milanese C (2019) Anthropometry-
driven block setting improves starting block
performance in sprinters. PLoS ONE 14(3):
e0213979. https://doi.org/10.1371/journal.
pone.0213979
Editor: Alena Grabowski, University of Colorado
Boulder, UNITED STATES
Received: June 25, 2018
Accepted: March 5, 2019
Published: March 27, 2019
Copyright: © 2019 Cavedon et al. This is an open
access article distributed under the terms of the
Creative Commons Attribution License, which
permits unrestricted use, distribution, and
reproduction in any medium, provided the original
author and source are credited.
Data Availability Statement: All relevant data are
within the paper and its Supporting Information
files.
Funding: This work was supported in part by
Departmental intramural funds (Joint Projects
2009) to CM. The funders had no role in study
design, data collection and analysis, decision to
publish, or preparation of the manuscript. There
was no additional external funding received for this
study. The funding organization (OMR Automotive)
did not play any role in the study design, data
Acceleration at the start of a race is influenced by the way sprinters positions themselves in
the blocks at the set command and the mechanics of leaving the blocks at the sound of the gun
[8,9]. The kinematic and kinetic patterns of elite athletes during the starting block phase and
acceleration phase have received considerable attention and many variables have been studied
to account for the effects of starting technique [2,3,5, 9–18].
According to the literature an effective sprint start mainly depends on the start block posi-
tioning and the joint angles of the lower limbs in the set position [8,9,11]. Furthermore, the
pushing time on the blocks and the forces generated by the front and rear legs in the pushing
phase [3,9–18] are also important. Recent studies found that the most predictive factor of
sprint start performance was the magnitude of force generated by the rear leg [16,19]. What is
more, the average horizontal external power (i.e., the ability to translate the centre of mass in
the running direction in a short period of time) was identified as an excellent descriptor of
start performance in sprinters [2].
The starting technique is greatly influenced by the setting of the block positions with
regards to spacing and obliquities [8,20,21]. Bezodis, Salo, & Trewartha [3] demonstrated that
“a single optimal set position” for everybody is not recommended due to varying physical fac-
tors, and therefore sprinters generally find their own preferred distance between the blocks
according to sensations or outcomes. For example, one of the most popular adjustments fre-
quently modified by the sprinters is the anteroposterior inter-block spacing and finding the
optimum setting for each athlete may take a long period of training. Furthermore, athletes
may not actually be selecting their ideal block setting for best performance when only basing
their preference on sensation.
The importance of anteroposterior inter-block spacing on the sprint movement during the
block start phase has been extensively investigated in several studies [3,6,9,12, 21–24]. The three
main types of block spacing investigated in the literature [20] are the bunched start (spacing gen-
erally <30 cm), the medium start (30 to 50 cm) and the elongated start (>50 cm). A number of
studies found that the velocity of the centre of mass at block clearing is higher when the inter-
block spacing increases due to an increase of force impulse [2,11,22]. This is due to an increased
duration of force generated against the blocks and a greater contribution of total force impulse
from the rear leg [21,25–27]. What is more, an increase in force generation from the rear leg has
been associated with higher block clearing velocities in elite sprinters [7,14,18]. However, a
recent study [12] demonstrated that in the elongated start, despite a greater velocity of the center
of mass at block clearing, the performance at 5 m and 10 m is significantly worse compared to
the bunched start. It is known that in the elongated start, the duration of force application is
increased during the block phase [9]. Spending longer on the blocks increases the total run time
which then conflicts with the objective of a sprint. Instead, at 10 m the best performance results
were obtained from the medium start. Further, a number of studies [3,9,12,22–24] suggested
that the medium start creates the best balance between total force generated and the increased
time of force generation to obtain the best performance in the early acceleration phase.
When looking for the best front block/starting line and inter-block distances for an individ-
ual athlete, it would seem obvious that the block distances should be relative to the individual’s
body dimensions. However, very few studies considered the anteroposterior block distances in
relation to the individual body dimensions of the athlete [11,20,22]. A study conducted by
Dickinson [20] stated that the distance of the front block from the starting line should depend
on the height of the individual, and the distance of the rear block from the starting line on the
leg and the thigh lengths, irrespective of the type of the start used (bunched, medium, elon-
gated). Henry [22] investigated if the optimum block spacing is related to the individual leg
length by analyzing the interaction between four experimental inter-block distances (27.9,
40.6, 53.3 and 66.0 cm) and the individual’s leg length. Contrary to Dickinson, Henry
Starting block performance in sprinters
PLOS ONE | https://doi.org/10.1371/journal.pone.0213979
March 27, 2019
2 / 20
collection and analysis, decision to publish, or
preparation of the manuscript. The funder provided
support in the form of salaries for an author [PM],
but did not have any additional role in the study
design, data collection and analysis, decision to
publish, or preparation of the manuscript. The
specific roles of this author are articulated in the
‘author contributions’ section.
Competing interests: We confirm that the
commercial affiliation (OMR Automotive) does not
alter our adherence to all PLOS ONE policies on
sharing data and materials.
concluded that leg length is not important in determining the best block spacing. A later study
conducted by Schot & Knutzen [11] went further in associating the athlete’s physical charac-
teristics with the anteroposterior block spacing. In this study, the authors stated that a medium
start could be achieved according to a calculation based on the individual’s leg length from the
greater trochanter to the lateral malleolus (60% of this length was used as the front block/start-
ing line distance and 45% for the inter-block spacing).
Anthropometry has been shown to play an important role in sports where body proportion-
ality differences may affect the biomechanics of movement and the resulting performance (e.g.
in running and in gymnastics) [28,29]. However, there is a lack of research addressing anthro-
pometric characteristics in the sprint start. Accordingly, research is needed to elucidate any
connection between the physical characteristics of the athlete, the block settings and the kine-
matic and kinetic parameters during the sprint start. In starting block performance, it is rea-
sonable to assume that body proportionality would influence optimal anteroposterior block
distances. The starting block performance involves a closed kinematic chain of movements
where the body extremities are in a fixed position and the only modifiable parameters are the
anteroposterior block distances. Thus, it can be argued that, in addition to body dimensions,
the proportion between the leg and trunk length may also affect the optimal anteroposterior
block distances. It is well known that individuals present different proportionality characteris-
tics between the leg and the trunk lengths and a way to assess such a proportionality is the Cor-
mic Index [30]. The Cormic Index expresses sitting height as a proportion of the total height,
representing a measure of the relative lengths of the trunk and lower limbs. Individuals are
classified as brachycormic, metricormic and macrocormic according to a Cormic Index
51%, 51–53%, or 53%, respectively [31].
Using instrumented starting blocks and high speed video cameras, the first aim of this
study was to test the effect of two different block settings in terms of anteroposterior block dis-
tances on the kinematic and kinetic parameters of well-trained sprinters. The two setting con-
ditions were the usual personal block setting used by the athlete and a block setting based on a
proportion of the individual’s leg length [11]. We hypothesized that an anthropometry-driven
intervention may improve the sprint start performance outcome. The second aim was to verify
whether an effect of the two block settings persists when the Cormic Index of the participants
is considered. The body proportionality characteristics of the sprinters may be of relevance in
a sprint start, a skill where the entire body is involved in a closed chain of movements. We
hypothesized that the Cormic Index of participants could influence the effect of a block setting
based on the leg length. It is expected that results of this work would help coaches and athletes
to improve sprint start performance using a quick anthropometry-driven procedure.
Materials and methods
Participants
The required sample was estimated “a priori” and calculated using GPower ver.3.1.9.2 [32].
Setting the type I error [SS3] at α = 0.05, the effect size at f = 0.25 and the correlation among
repeated measures at 0.6, the minimum sample size required for a within-between interaction
in a mixed-design ANOVA for having an 80% power (i.e., β = 0.20) was 36 subjects. In order
to comply with a possible ~15% dropout, forty-two participants were initially recruited. Partic-
ipants were well-trained skilled sprinters (20 women and 22 men) with a competitive athletic
career of at least 2 years in sprint running. Female and male participants’ age, height and body
mass (±SD) were 19.70 ± 2.23 and 19.36 ± 2.11 y, 165.4 ± 5.2 and 176.7 ± 5.9 cm, 55.6 ± 6.8
and 67.1 ± 9.8 kg, respectively. All sprinters were involved in regional and national level com-
petitions and trained at least 6 times a week for 2/3 hours per day. Their best time over 100 m
Starting block performance in sprinters
PLOS ONE | https://doi.org/10.1371/journal.pone.0213979
March 27, 2019
3 / 20
ranged between 10.45 s and 11.30 s for men and between 11.45 s and 12.68 s for women.
According to the Cormic Index, the sprinters were brachycormic (n = 12), metricormic
(n = 19) and macrocormic (n = 11). The mean Cormic Index in the three groups was 50.62% ±
1.14, 52.48% ± 0.98 and 53.78% ± 0.51, respectively. All participants gave their written
informed consent to participate in the study, and the protocol was performed in accordance
with the Declaration of Helsinki. Ethics approval was obtained from the University of Verona
Institutional Review Board.
Experimental procedure
The sprint testing took place on an outdoor track (Olimpic Plast SWD surface, Olimpia Cost-
ruzioni, Forlı`, Italy) during the early competition phase of the outdoor season. One operator
(VC) attached ten retro-reflective passive flat markers (14 mm diameter) bilaterally over spe-
cific anatomical landmarks on the participant’s body (i.e., right [R] and left [L] acromion, R
and L femur greater trochanter, R and L femur lateral epicondyle, R and L fibula apex of lateral
malleolus, R and L 5th metatarsal).
Following a warm-up consisting of jogging, dynamic stretching and sprints of submaximal
intensity, all participants performed a total of 6 maximal-effort 10 m sprints using two differ-
ent starting conditions: 1) the athlete’s usual personal block setting (US), and 2) an anthro-
pometry-driven setting (AS). The order of the two starting conditions was randomized for
each athlete. In the AS condition, the start block positions were set according to a calculation
based on the individual’s leg length from the greater trochanter to lateral malleolus [11]; 60%
of this length was used as the front block/starting line distance and 45% as the inter-block
spacing. The anteroposterior block distances in both the US and AS conditions were measured
to the nearest 0.1 cm in the outdoor track with a fiberglass tape. The obliquity of the blocks
was that usually used by the participants and was the same in both conditions. Participants
used their own spiked shoes for sprint running. For all trials, each sprint was initiated by the
same experimenter (MP), who provided standard ‘on your marks’ and ‘set’ commands. The
experimenter then pressed a custom-designed trigger button to provide the auditory start sig-
nal through a sounder device. The rest period between trials was 5–7 minutes.
After an adequate rest period, the standing long jump test was used to measure lower
extremity strength. Participants stood behind a line marked on the ground with their feet
slightly apart. A two-foot take-off and landing was used, with swinging of the arms and bend-
ing of the knees to provide forward drive, the subject attempting to jump as far as possible.
The horizontal distance between the starting line and the back of the heel closest to the starting
line at landing was recorded via tape measure to the nearest 0.1 cm. Three trials were per-
formed (with adequate rest between trials) and the maximum distance was recorded. Although
the standing long jump performance is usually expressed in absolute terms as the overall dis-
tance covered, it has been suggested that the subject’s leg length can play a significant role in
the performance [33]. Accordingly, performance in the standing long jump test was expressed
relative to the leg length (SLJ-relative).
Data collection
Anthropometric data. Anthropometric data were taken by one operator (VC) using con-
ventional criteria and measuring procedures [34]. Body mass was assessed to the nearest 0.1 kg
using a certified electronic scale (Tanita electronic scale BWB-800 MA, Wunder SA.BI. Srl,
Milano, Italy). Standing height and sitting height were measured to the nearest 0.1 cm using a
Harpenden portable stadiometer (Holtain Ltd., Crymych, Pembs. UK). For the sitting height,
the subject was asked to sit on a flat stool of a known height. Measurement was taken with the
Starting block performance in sprinters
PLOS ONE | https://doi.org/10.1371/journal.pone.0213979
March 27, 2019
4 / 20
subject sitting in a standard position. The sitting height was then obtained by subtracting the
height of the stool from the reading on the stadiometer. The lower limb length was measured
with a Harpenden anthropometer (Holtain Ltd., Crymych, Pembs. UK) as the distance
between the greater trochanter and the lateral malleolus (cm). Body circumferences were mea-
sured with a fiberglass tape at the mid-thigh and the calf. The Cormic Index was calculated for
each participant as sitting height (cm)/standing height (cm)100. The body mass index (BMI)
was also computed for each participant as body mass (kg)/standing height (m)2.
Kinetic and kinematic data.
Each trial was performed using a set of dynamometric start-
ing blocks equipped with a set of load cells (CU K5D and CU K1C models, GEFRAN SpA, Bre-
scia, Italy) enabling the measurement of the magnitude and direction of forces generated by the
sprinter during the starting block phase. The acquisition frequency was 1 kHz and the sensitiv-
ity was 0.01 N. These blocks respected all the features of those normally used in the track sprint
start, as well the output characteristics of similar blocks used in previous works [13,16]. There
were four monoaxial load cells installed on each block, which were used to determine the loads:
three cells were used to measure the vertical loads and one to measure the horizontal loads. The
force data were resolved into horizontal and vertical components for each foot (Fig 1). The
Fig 1. Example of resultant force curves of the front and rear blocks during a sprint start recorded by the instrumented starting blocks. RT, reaction
time; RBT, rear block time; FBT, front block time; TBT, total block time; RPF, rear peak force; FPF, front peak force.
https://doi.org/10.1371/journal.pone.0213979.g001
Starting block performance in sprinters
PLOS ONE | https://doi.org/10.1371/journal.pone.0213979
March 27, 2019
5 / 20
blocks were connected to a portable personal computer (SoftPLC, GEFRAN SpA, Brescia, Italy)
that stored data, and handled signal processing and parameter calculation. The dynamometric
starting blocks were developed at OMR Automotive (Officine Meccaniche Rezzatesi, Brescia,
Italy).
The reliability and validity tests were performed at the Department of Industrial Engineer-
ing, University of Padua, Italy. Both static and dynamic bench tests were carried out on a
mechanical bench equipped with two MTS 242 servo hydraulic actuators to calibrate the force
transducers. Several tests were performed in the horizontal and vertical directions on each part
of the apparatus in order to obtain the single axis calibration constants of the system. Static
and dynamic load tests with combined loading were also performed to validate the results of a
realistic usage, showing good results with low error: the difference between the maximum
force applied on the block and the measured force was lower than 1% [35]. After bench tests, a
series of in-vivo sprint starts were recorded during a training session with beginner, intermedi-
ate and expert athletes and comparison was made with a commercial force plate (Model 4060–
10, Bertec Columbus, OH, USA) output [35] included in a SMART DX-6000 motion capture
system (BTS Bioengineering, Quincy, MA, USA) after mounting the dynamometric blocks
over the force platform. The results showed that both vertical and horizontal forces measured
by the dynamometric starting blocks and the force plate were very similar to each other: hori-
zontal and vertical forces were able to follow the loads also in the region of rapid loading and
unloading with good accuracy. Regression coefficients were found to be on average R = 0.9946
for horizontal loads and R = 0.9978 for vertical loads. The dynamometric starting blocks appa-
ratus was portable and made resistant to splash water and could therefore be used outdoors.
These dynamometric starting blocks can easily be used on the track during training, giving
precise quantitative data which would usually only be available in a biomechanics laboratory
setting using sophisticated instruments in a controlled environment.
Two high-speed video cameras (Casio Exilim ex-zr 1000, Casio Europe Gmbh, Barcelona,
Spain) captured the movement of each athlete in two dimensions during the starting block
and acceleration phases (first and second stride lengths). One camera (Camera 1) was posi-
tioned for the front block side view of the participant and the other camera (Camera 2) for the
rear block side view of the participant. According to Bartlett [36], in order to limit the potential
technological errors in the filming set-up connected with the 2D videography (e.g. the parallax
errors due to visual distortion), the optical axis of each camera was lined directly perpendicular
to the sprinter’s movement plane. Furthermore, the optical axis of each camera was centred in
correspondence to an imaginary line perpendicular to the ground and passing through the hip
joint of the leg facing the camera at the “on your marks” position. Each camera was placed on
a tripod at a height of approximately 1 m and located at approximately 5 m from the partici-
pant. Each camera’s field of view provided a sagittal view of each sprinter for the first two full
strides. The position of the two cameras was standardized for all participants to ensure no
environmental changes during field testing (Fig 2). Each video was calibrated with a 50 cm
cube positioned to the rear of the blocks and defined by X the horizontal axis and Y the vertical
axis. Images of the starting block phase and acceleration phase were collected at a resolution of
1280 x 1024 pixels using a shutter speed of 1/1000 s and a sampling frequency of 200 Hz.
Three pairs of photocells (Polifemo Light Radio, Microgate SRL, Bolzano, Italy) based on a
radio impulse transmission system and a reflection system were used to measure the times at 5
m and 10 m in the sprint trials. The timing between the dynamometric starting blocks and the
photocells system was synchronized using the digital output available from the block control
system and connecting it to the available input for timing available in the Microgate unit.
Data analysis.
The kinematic and kinetic outputs were stored during the trials in a porta-
ble personal computer connected to the instrumented blocks and were then exported for
Starting block performance in sprinters
PLOS ONE | https://doi.org/10.1371/journal.pone.0213979
March 27, 2019
6 / 20
further analysis. Raw data were filtered using a low-pass Butterworth filter (fourth order) with
a cutoff frequency of 120 Hz and analyzed using a custom program written in Matlab R2008a
(MathWorks, Natick, MA, USA).
Force signals were resolved into horizontal and vertical components using a coordinate sys-
tem affixed to the runway. The x-axis pointed forward along the running surface (horizontal
plane), the y-axis pointed vertically upwards. Force data were used to define the force onset
threshold (i.e., when the first derivative of the resultant force-time curve was greater than 500
Ns-1) and the force offset thresholds (i.e., when the resultant force was lower than 50 N). The
pushing phase was defined between the first instant of block start (i.e., corresponding to the
force onset threshold) to block clearance (i.e., corresponding to the force offset threshold on
the front block). The pushing phases of the front and the rear blocks were defined as the period
between the instant of the block start and the end of the respective sub-phases for each block.
The following temporal parameters were extracted for analysis from the instrumented blocks
data: the reaction time (RT), defined as the time from the auditory signal to the first force
onset threshold; the front block time (FBT), defined as the pushing time on the front block
sub-phase; the rear block time (RBT), defined as the pushing time on the rear block sub-phase
and the total time block (TBT), defined as the pushing time on the total pushing phase. The
following kinetic variables obtained from the instrumented blocks data during the pushing
phase were also measured: the front peak force (FPF), defined as the maximum resultant front
force value; the rear peak force (RPF), defined as the maximum resultant rear force value; the
horizontal and the vertical front peak force (H_FPF and V_FPF); the horizontal and the verti-
cal rear peak force (H_RPF and V_RPF); the average total force (ATF). The front force impulse
(FFimpulse), the rear force impulse (RFimpulse) and the total force impulse (Total Fimpulse) were
computed according to Otsuka and colleagues [15]. All the kinetic variables were normalized
to the body mass of the sprinters expressed in kg. In addition, the following variables were
computed: the ratio of horizontal to resultant force impulse of both legs (Ratio_front and
Ratio_rear) [37]; the horizontal block velocity (H_BV) measured as the sum between the hori-
zontal impulse on the front block plus the horizontal impulse on the rear block (both expressed
in Ns) divided by the body mass of the sprinter expressed in kg; the normalized average hori-
zontal external block power (NAHEP) calculated according to the procedures by Bezodis and
colleagues [2].
Fig 2. Experimental set-up.
https://doi.org/10.1371/journal.pone.0213979.g002
Starting block performance in sprinters
PLOS ONE | https://doi.org/10.1371/journal.pone.0213979
March 27, 2019
7 / 20
For each participant the video material of the 12 video clips (6 trials and 2 cameras) was
uploaded onto a PC and digitalized at full resolution with a zoom factor of 2.5 using freeware
motion-analysis software (Kinovea; version 0.8.15, available for download at: http://www.
kinovea.org). One operator (VC) blinded to condition (AS, US) manually digitalized the mark-
ers and quantified the joint angles and stride lengths at specific video frames on each video
(see details below). The analysis of the video material was carried out in two stages. Firstly,
each video clip was rewound frame by frame and frozen at the set position to directly estimate
the angles in the sagittal plane with the digital goniometer built into the Kinovea software. The
operator visually digitalized the markers through the “cross marker” function and placed the
digital goniometers. The videos from Camera 1 (i.e., with the front block side view) were used
to estimate the hip, knee and ankle joint angles on the front leg while the videos from Camera
2 (i.e., with the rear block side view) were used to measure the hip, knee and ankle joint angles
on the rear leg (Fig 3). The joint angles were measured to the nearest 1 degree. Secondly, in
order to measure the length of the first stride, each video clip from Camera 1 was stopped at
the instant of the first foot strike. Stride was estimated at the first frame of the video clip where
the sprinter’s foot made contact with the track surface. The length of the inferior side of the
cube positioned to the rear of the blocks in the frame was used as a reference to calibrate all
other line lengths. The “line drawing tool” function was used to assess the horizontal distance
between the front block and the toe of the rear foot at the first foot strike (first stride length
[SL1]). Thirdly, in order to measure the length of the second stride, each video clip from Cam-
era 2 was stopped at the instant of the second foot strike. The “line drawing tool” function was
used to assess the horizontal distance to the nearest 0.1 cm between the rear foot at the first toe
off and toe of the front foot at the first foot strike (second stride length [SL2]). SL1 and SL2
were normalized to leg length to account for differences among subjects and labelled as
NorSL1 and NorSL2, respectively.
In order to limit any technical errors involved in 2D videography as much as possible, the
procedure adopted to quantify the joint angles and stride lengths was repeated in three sepa-
rate sessions, with a minimum interval of 7 days between sessions and the mean value was
recorded only when the coefficient of variation was <0.05. Furthermore, the operator was
familiar with the use of high-speed video as a tool to quantify joint angles in sprint running
and in sprint starts. It has been noted that the markers, despite being properly positioned dur-
ing data collection, can move in relation to the skin throughout the range of motion [38].
Accordingly, in line with Bradshaw and colleagues [39], the operator paid close attention dur-
ing data analysis to this fact and visually adjusted for skin movement by only using the markers
as a guide.
Fig 3. Two-dimensional functional representation of the joint angles on the front (Panel A) and rear blocks (Panel B).
https://doi.org/10.1371/journal.pone.0213979.g003
Starting block performance in sprinters
PLOS ONE | https://doi.org/10.1371/journal.pone.0213979
March 27, 2019
8 / 20
In this study, errors (i.e. parallax and lens distortion errors) in the measurement of the first
and second stride lengths has to be taken into account. A typical error (%Error) was estimated
starting from the calibration cube (50 cm) positioned behind the blocks and then every 50 cm
until the end of the field of view. The %Error was calculated according to the following formula:
%Error ¼ jE were found for the sitting height and the lower limb length (F = 13.238 and F = 14.495, respec-
tively; P<0.001 for both). Post-hoc analysis with Bonferroni’s correction revealed that the bra-
chycormic group had significantly lower sitting height and lower limb length values in
comparison with both the metricormic (P = 0.002 for both) and the macrocormic (P<0.001
for both) groups (Table 1). The mean values (±SD) of the anteroposterior block distances and
the kinematic and kinetic data in the US and AS conditions in the whole sample and in the
three Cormic Index groups are reported in Table 2.
One-way ANOVA showed that in the US condition, the front block/starting line distance
was significantly different within the three groups (F = 9.500, P<0.001), but the inter-block
distance was not. Post-hoc analysis with Bonferroni’s correction revealed a significantly lower
front block/starting line distance in the brachycormic group vs. the metricormic group (-5.9
cm, P = 0.001) and macrocormic group (-6.1 cm, P = 0.002). One-way ANOVA also indicated
that all of the measured sprint start kinematic and kinetic variables were similar in the US con-
dition in the three Cormic Index groups with the exclusion of the TBT (F = 3.369, P = 0.045).
However, post-hoc analysis with Bonferroni’s correction, revealed no significant group-group
difference in the TBT.
A mixed-design ANOVA with three groups (brachycormic, metricormic, macrocormic)
and two block setting conditions (US and AS) with repeated measures on the second factor
showed a significant main effect of condition for both anteroposterior block distances and for
several kinematic and kinetic measurements (Tables 2 and 3).
The front block/starting line distance was significantly lower (-3.2 cm) in the AS vs. the US
whereas the inter-block distance was significantly greater (+9.2 cm). Furthermore, at the set
position, the rear hip and the rear knee joint angles were significantly greater in the AS vs. the
US (+6 and +5 degrees, respectively) while the opposite was found at the front hip and knee
joint angles (-4 and -3 degrees, respectively). The rear ankle joint angle was significantly lower
in the AS vs. the US (-2 degrees); no significant difference was found for the front ankle joint
angle. Results also showed that, during the pushing phase, several kinematic and kinetic vari-
ables were significantly greater in the AS vs. the US namely, the FBT (+0.02 s), the RPF (+1.63
N/kg), the H_RPF (+0.43 N/kg), the RFimpulse (+0.10 Ns/kg), the Total Fimpulse (+0.17 Ns/kg),
the Ratio_rear (+0.01), H_BV (+0.14 m/s), and the NAHEP (+0.03). No significant differences
were found for the other variables during the pushing phase. In the acceleration phase, the times
at 5 m and 10 m were significantly lower in the AS vs. the US (-0.05 and -0.04 s, respectively).
Table 1. Characteristics of the participants in the whole sample and the three Cormic Index groups. Data are means ± SD.
Whole sample
(n = 42)
Brachycormic
(n = 12)
Metricormic
(n = 19)
Macrocormic
(n = 11)
Age (y)
19.52 ± 2.14
19.75 ± 2.38
19.63 ± 2.41
19.09 ± 1.38
Sprinting experience (y)
4.76 ± 2.67
5.83 ± 3.04
3.95 ± 2.27
5.00 ± 2.65
Body mass (kg)
61.6 ± 10.2
59.6 ± 9.9
63.7 ± 10.4
60.3 ± 10.4
Height (cm)
171.3 ±7.9
167.6 ± 8.1
172. 4 ± 6.8
173.6 ± 8.8
BMI (kg/m2)
20.89 ± 2.19
21.10 ± 1.98
21.33 ± 2.44
19.89 ± 1.76
Sitting height (cm)
89.6 ± 5
84.9 ± 5
90.4 ± 3
93.33 ± 4.6
Lower limb length (cm)
81.8 ± 5
77.1 ± 5
82.5 ± 2.8
85.8 ± 4.2
Thigh circumference (cm)
49.04 ± 3.27
49.57 ± 3.97
49.32 ± 3.38
47.98 ± 2.09
Calf circumference (cm)
34.72 ± 2.56
34.73 ± 3.08
35.01 ± 2.41
34.22 ± 2.35
SLJ-relative
2.8 ± 0.2
2.7 ± 0.2
2.9 ± 0.3
2.7 ± 0.2
BMI, body mass index; SLJ-relative, performance in the standing long jump test expressed relative to the leg length.
, significantly different (P<0.05; Bonferroni’s post-hoc) vs. metricormic and macrocormic.
https://doi.org/10.1371/journal.pone.0213979.t001
Starting block performance in sprinters
PLOS ONE | https://doi.org/10.1371/journal.pone.0213979
March 27, 2019
10 / 20
Table 2. Block distances, and kinematic and kinetic data in the sprint start for the whole sample as well as for the three Cormic Index groups in their usual block
setting condition (US) and an anthropometry-driven block setting (AS) condition. Data are means ± SD.
Variable
Whole sample
(n = 42)
Brachycormic
(n = 12)
Metricormic
(n = 19)
Macrocormic
(n = 11)
Block distances
US
AS
US
AS
US
AS
US
AS
FB/SL distance (cm)
52.3 ± 4.8
49.1 ± 3.0
48.0 ± 3.6
46.3 ± 3.1
53. 9 ± 4.3
49.5 ± 1.7
54.2 ± 3.9
51.5 ± 2.5
I-B distance (cm)^
27.6 ± 2.4
36.8 ± 2.3
27.3 ± 2.5
34.7 ± 2.3#
27.2 ± 2.2
37.1 ± 1.3#
28.6 ± 2.5
38.6 ± 1.9#
Set position
Front hip (˚)
47 ± 6
43 ± 6
44 ± 6
41 ± 7
49 ± 4
43 ± 4
47 ± 9
44 ± 8
Rear hip (˚)
77 ± 8
84 ± 8
75 ± 8
80 ± 10
79 ± 7
85 ± 5
77 ± 11
85 ± 10
Front knee (˚)
92 ± 9
90 ± 8
94 ± 7
93 ± 5
92 ± 9
88 ± 8
92 ± 9
90 ± 9
Rear knee (˚)
112 ± 11
117 ± 11
115 ± 7
119 ±8
111 ± 12
115±11
112 ± 12
118 ± 15
Front ankle (˚)
92 ± 6
93 ± 7
94 ± 4
95 ± 3
92 ± 7
93 ± 7
92 ± 6
92 ± 8
Rear ankle (˚)
87 ± 6
85 ± 7
88 ± 5
86 ± 6
88 ± 6
86 ± 8
85 ± 6
84 ± 7
Pushing phase
RT (s)
0.185 ± 0.035
0.189 ± 0.035
0.195 ± 0.026
0.192 ± 0.029
0.184 ± 0.040
0.189 ± 0.032
0.175 ± 0.035
0.186 ± 0.047
FBT (s)
0.402 ± 0.041
0.416 ± 0.064
0.402 ± 0.051
0.402 ± 0.044
0.394 ± 0.036
0.419 ± 0.086
0.419 ± 0.036
0.424 ± 0.027
RBT (s)
0.211 ± 0.041
0.212 ± 0.041
0.220 ± 0.052
0.225 ± 0.059
0.203 ± 0.038
0.210 ± 0.035
0.215 ± 0.031
0.202 ± 0.025
TBT (s)
0.421 ± 0.047
0.427 ± 0.038
0.411 ± 0.035
0.415 ± 0.028
0.409 ± 0.035
0.424 ± 0.036
0.451 ± 0.066
0.448 ± 0.045
FPF (N/kg)
16.50 ± 2.27
16.60 ± 2.09
15.18 ± 2.16
15.31 ± 1.80
17.13 ± 2.20
17.16 ± 2.29
16.85 ± 2.04
17.03 ± 1.44
RPF (N/kg)^
11.44 ± 2.48
13.07 ± 3.37
10.83 ± 3.10
12.74 ± 3.10#
11.76 ± 2.03
14.08 ± 2.76#
11.55 ± 2.58
11.68 ± 4.24
H_FPF (N/kg)
6.02 ± 0.71
5.91 ± 0.65
5.65 ± 0.74
5.62 ± 0.67
6.25 ± 0.70
6.02 ± 0.71
6.02 ± 0.56
6.06 ± 0.45
V_FPF (N/kg)^
6.13 ± 0.92
6.12 ± 0.90
6.01 ± 0.98
5.91 ± 1.20
6.25 ± 0.87
5.99 ± 0.74
6.06 ± 0.98
6.58 ± 0.68#
H_RPF (N/kg)^
4.52 ± 1.09
4.95 ± 1.34
4.13 ± 1.17
4.58 ± 1.21#
4.70 ± 0.97
5.41 ± 1.00#
4.63 ± 1.20
4.55 ± 1.78
V_RPF (N/kg)^
3.78 ± 1.12
3.96 ± 1.20
3.36 ± 1.04
3.55 ± 1.07
3.76 ± 1.08
4.23 ± 1.00#
4.28 ± 1.16
3.94 ± 1.58
ATF (N/kg)
11.37 ± 1.19
11.55 ± 1.12
11.12 ± 1.58
11.22 ± 1.51
11.63 ± 0.91
11.83 ± 0.87
11.21 ± 1.14
11.44 ± 0.96
FFimpulse (Ns/kg)
3.48 ± 0.50
3.56 ± 0.55
3.32 ± 0.61
3.32 ± 0.62
3.52 ± 0.48
3.57 ± 0.58
3.60 ± 0.38
3.79 ± 0.33
RFimpulse (Ns/kg)^
1.29 ± 0.40
1.39 ± 0.42
1.25 ± 0.47
1.34 ± 0.40
1.25 ± 0.30
1.48 ± 0.34#
1.40 ± 0.47
1.31 ± 0.56
Total Fimpulse (Ns/kg)
4.76 ± 0.55
4.93 ± 0.56
4.56 ± 0.69
4.64 ± 0.65
4.75 ± 0.45
5.01 ± 0.54
5.01 ± 0.51
5.10 ± 0.38
Ratio_front
0.69 ± 0.05
0.69 ± 0.05
0.68 ± 0.05
0.69 ± 0.06
0.70 ± 0.05
0.70 ± 0.05
0.70 ± 0.05
0.67 ± 0.04
Ratio_rear
0.75 ± 0.05
0.76 ± 0.05
0.75 ± 0.06
0.76 ± 0.06
0.76 ± 0.05
0.77 ± 0.04
0.71 ± 0.03
0.73 ± 0.05
Ratio_total
0.71 ± 0.04
0.72 ± 0.04
0.71 ± 0.05
0.71 ± 0.05
0.72 ± 0.05
0.73 ± 0.04
0.70 ± 0.03
0.69 ± 0.04
H_BV (m/s)
3.36 ± 0.35
3.50 ± 0.39
3.18 ± 0.41
3.27 ± 0.36
3.40 ± 0.28
3.62 ± 0.38
3.48 ± 0.35
3.52 ± 0.34
NAHEP
0.47 ± 0.90
0.50 ± 0.10
0.44 ± 0.10
0.46 ± 0.09
0.49 ± 0.08
0.54 ± 0.10
0.46 ± 0.08
0.47 ± 0.08
Acceleration Phase
5 m (s)
1.34 ± 0.10
1.29 ± 0.09
1.36 ± 0.10
1.31 ± 0.12
1.33 ± 0.10
1.28 ± 0.09
1.33 ± 0.09
1.29 ± 0.06
10 m (s)
2.07 ± 0.13
2.03 ± 0.13
2.12 ± 0.14
2.09 ± 0.16
2.05 ± 0.15
2.00 ± 0.12
2.06 ± 0.10
2.01 ± 0.07
NorSL1
1.09 ± 0.11
1.12 ± 0.12
1.06 ± 0.09
1.12 ± 0.12
1.11 ± 0.14
1.14 ± 0.15
1.07 ± 0.09
1.07 ± 0.07
NorSL2
1.15 ± 0.14
1.19 ± 0.12
1.12 ± 0.15
1.16 ± 0.11
1.18 ± 0.15
1.12 ± 0.13
1.14 ± 0.10
1.15 ± 0.10
FB/SL, front block/starting line; I-B, inter-block; RT, reaction time; FBT, front block time; RBT, rear block time; TBT, total block time; FPF, front peak force; RPF, rear
peak force; H_FPF, horizontal front peak force; V_FPF, vertical front peak force; H_RPF, horizontal rear peak force; V_RPF, vertical rear peak force; ATF, average total
force; FFimpulse, front force impulse; RFimpulse, rear force impulse; Total Fimpulse, total force impulse; Ratio_front, Ratio of horizontal to resultant force impulse of front
leg; Ratio_rear, Ratio of horizontal to resultant force impulse of rear leg; NAHEP, normalized average horizontal external power; H_BV, horizontal block velocity; 5 m,
time at 5 meters; 10 m, time at 10 meters; NorSL1, first stride length normalized to leg length; NorSL2, second stride length normalized to leg length.
, statistically significant (P<0.05) difference between the AS and US conditions
^, statistically significant (P<0.05) difference of the AS vs. the US effect across the three Cormic Index groups
#, statistically significant difference between the AS and US conditions (post-hoc analysis with Bonferroni’s correction).
https://doi.org/10.1371/journal.pone.0213979.t002
Starting block performance in sprinters
PLOS ONE | https://doi.org/10.1371/journal.pone.0213979
March 27, 2019
11 / 20
NorSL1, and NorSL2 were significantly larger in the AS vs. the US (+0.03 and +0.04, respectively).
The effect size was large (>0.14) for all the significantly different variables except for FBT,
NorSL1, and NorSL2 where the effect size was medium (>0.10 effect size <0.12).
Table 3. Results of the mixed-design ANOVA to quantify the effect of condition and effect of group by condition interaction on block distances, and kinematic and
kinetic data.
Condition
Group by condition
Variable
F value
P value
ηp
2
F value
P value
ηp
2
Block distances
FB/SL distance
29.232
<0.001
0.428
2.308
0.113
0.106
I-B distance
545.112
<0.001
0.933
4.494
0.018
0.187
Set position
Front hip
46.494
<0.001
0.544
2.392
0.105
0.109
Rear hip
60.857
<0.001
0.609
1.520
0.231
0.072
Front knee
11.569
0.002
0.229
1.224
0.305
0.059
Rear knee
10.482
0.002
0.212
1.643
0.531
0.032
Front ankle
3.830
0.058
0.089
0.213
0.809
0.011
Rear ankle
7.485
0.009
0.161
0.150
0.861
0.008
Pushing phase
RT
0.925
0.342
0.023
0.621
0.543
0.031
FBT
4.261
0.046
0.098
1.201
0.312
0.058
RBT
0.031
0.862
0.001
2.661
0.083
0.120
TBT
1.647
0.207
0.041
1.770
0.184
0.083
FPF
0.356
0.554
0.009
0.066
0.937
0.003
RPF
25.821
<0.001
0.398
5.402
0.008
0.217
H_FPF
1.969
0.168
0.048
2.247
0.119
0.103
V_FPF
0.274
0.603
0.007
4.935
0.012
0.202
H_RPF
10.706
0.002
0.215
4.436
0.018
0.185
V_RPF
1.299
0.261
0.032
6.770
0.003
0.258
ATF
2.237
0.143
0.054
0.165
0.848
0.008
FFimpulse
1.950
0.171
0.048
0.815
0.450
0.040
RFimpulse
6.821
0.013
0.149
10.191
<0.001
0.343
Total Fimpulse
11.838
0.001
0.233
2.255
0.118
0.104
Ratio_front
1.321
0.257
0.033
3.241
0.050
0.143
Ratio_rear
9.244
0.004
0.192
0.829
0.444
0.041
Ratio_total
0.008
0.930
0.000
1.803
0.178
0.085
H_BV
11.946
0.001
0.234
3.107
0.056
0.137
NAHEP
6.815
0.013
0.149
1.446
0.248
0.069
Acceleration phase
5 m
27.802
<0.001
0.416
0.273
0.762
0.014
10 m
19.531
<0.001
0.334
0.219
0.804
0.011
NorSL1
12.041
0.001
0.236
3.002
0.061
0.133
NorSL2
12.732
0.001
0.246
1.241
0.300
0.060
US, usual sprinter’s block setting; AS, anthropometry-driven block setting; ηp
2, partial eta squared; FB/SL, front block/starting line; I-B, inter-block; RT, reaction time;
FBT, front block time; RBT, rear block time; TBT, total block time; FPF, front peak force; RPF, rear peak force; H_FPF, horizontal front peak force; V_FPF, vertical front
peak force; H_RPF, horizontal rear peak force; V_RPF, vertical rear peak force; ATF, average total force; FFimpulse, front force impulse; RFimpulse, rear force impulse;
Total Fimpulse, total force impulse; Ratio_front, Ratio of horizontal to resultant force impulse of front leg; Ratio_rear, Ratio of horizontal to resultant force impulse of rear
leg; NAHEP, normalized average horizontal external power; H_BV, horizontal block velocity; 5 m, time at 5 meters; 10 m, time at 10 meters; NorSL1, first stride length
normalized to leg length; NorSL2, second stride length normalized to leg length.
https://doi.org/10.1371/journal.pone.0213979.t003
Starting block performance in sprinters
PLOS ONE | https://doi.org/10.1371/journal.pone.0213979
March 27, 2019
12 / 20
As shown in Table 3, the results of the mixed-design ANOVA (group by condition)
revealed a significant main effect for condition (US and AS) by group (brachycormic, metri-
cormic and macrocormic) interaction for a number of variables. The effect size for all the sig-
nificantly different variables was large (>0.14). Post hoc analysis showed significantly greater
mean inter-block distance in the AS vs. the US in the brachycormic, metricormic and macro-
cormic groups by respectively +7.4 cm, +9.9 cm and +10.0 cm (P<0.001 for all); in the AS con-
dition, RPF was significantly greater in the brachycormic and the metricormic group (+1.91
N/kg, P = 0.001 and +2.32 N/kg, P<0.001, respectively), but not in the macrocormic group.
H_RPF was significantly greater for brachicormic group (+0.45 N/kg, P = 0.031) and the
metricormic group (+0.71 N/kg, P<0.001). In the AS V_RPF was significantly greater in the
metricormic group vs. the US (+0.47 N/kg, P = 0.001) and the V_FPF was significantly higher
in the macrocormic group (+0.52 N/kg, P = 0.013). RFimpulse was also significantly greater in
the metricormic group (+0.23 Ns/kg, P<0.001) in the AS vs. the US.
The estimation of a regression model for each performance variable using the difference
(Delta) in the anteroposterior block distances between the two conditions (US and AS) as pre-
dictors and the Delta in the performance variable as dependent variable, evidenced some sta-
tistically significant relationships which were summarized in Table 4.
Discussion
The first aim of this study was to investigate the effect of two different block setting conditions
(US and AS) on kinematic and kinetic performance outcomes during the sprint start in well-
trained sprinters focusing on block phases (set position and pushing-phase), and the early
acceleration phase (times at 5 m and 10 m, first and second stride lengths). Results showed
that an anthropometry-driven block setting condition (AS) based on the sprinter’s leg length
[11] is associated with several statistically significant changes in postural parameters at the set
position, as well as in kinetic and kinematic variables at the pushing and acceleration phases in
comparison with the sprinter’s usual block setting, leading to improved performance.
When using the US condition, which was based on the individual sprinter’s preferences, all
participants in the sample adopted an inter-block distance (27.6 ± 2.4 cm) classifiable as a
bunched start, suggesting that sprinters prefer a low anteroposterior distance between the
front and rear foot. This is in agreement with recent studies [15,18] showing that sprinters
with different ability levels chose an anteroposterior inter-block distance from 25 cm to 30 cm
Table 4. Multiple linear regression model estimated using the difference (Delta) in the anteroposterior block dis-
tances between the two conditions (US and AS) as predictors and the delta in the performance variable as the
dependent variable.
Delta FB/SL distance
Delta I-B distance
Adj. R2
Delta RPF
r = -0.35; P = 0.024
r = -0.26; P = 0.102
0.11
Delta H_RPF
r = -0.31; P = 0.047
r = -0.21; P = 0.175
0.07
Delta V_RPF
r = -0.37; P = 0.017
r = -0.19; P = 0.246
0.10
Delta RFimpulse
r = -0.34; P = 0.027
r = -0.03; P = 0.836
0.08
Delta Total Fimpulse
r = -0.50; P < 0.001
r = -0.10; P = 0.545
0.25
Delta Rear hip
r = 0.33; P = 0.036
r = 0.62; P < 0.001
0.36
US, usual sprinter’s block setting; AS, anthropometry-driven block setting; FB/SL, front block/starting line; I-B, inter-
block; Adj. R2, adjusted coefficient of determination; r, partial correlation coefficient; RPF, rear peak force; H_RPF,
horizontal rear peak force; V_RPF, vertical rear peak force; RFimpulse, rear force impulse; Total Fimpulse, total force
impulse.
https://doi.org/10.1371/journal.pone.0213979.t004
Starting block performance in sprinters
PLOS ONE | https://doi.org/10.1371/journal.pone.0213979
March 27, 2019
13 / 20
on the basis of their sensations. Nevertheless, the bunched start has been demonstrated to be
the least efficient from a biomechanical perspective because less force is exerted on the starting
blocks with a reduction in block velocity [12,21,42].
Several studies (2,7,11,14,18,21,22,25–27) highlighted that the increase in inter-block dis-
tance is associated with improved performance in several kinetic and kinematic variables
linked to the sprint start (i.e., a greater contribution of total force generation and force impulse
from the rear leg and higher block clearing velocities). On the other hand, in the AS condition,
which was based on the individual sprinter’s leg length, sprinters decreased the front block/
starting line distance (-6.56%) and increased the inter-block distance (+25.02%). Such varia-
tions in the anteroposterior block distances in the AS condition lead to a postural adaptation
at the set position resulting in a decrease in the front hip, front knee and rear ankle joint angles
and an increase in the rear hip and rear knee joint angles (Table 3). Based on the literature
[3,4,17,43,44], it is reasonable to assume that these postural changes at the set position in the
AS condition compared to the US condition could be associated with an improvement in per-
formance in the subsequent pushing phase such as greater RPF, H_RPF, RFI, Ratio_rear, Total
Fimpulse and NAHEP. In fact, the resulting joint angles (front/rear) in the AS condition were
similar to those reported in sprinters with ability levels ranging from national-level to world-
class [3,4,17]. Consistently, scientific data also showed that the front hip and knee as well as
the ankle joint angles were found to be smaller in faster than slower sprinters, allowing for the
stretch-reflex of the hip extensor and the soleus muscles and the greatest velocity when leaving
the blocks [43,44].
A recent study identified an important role of the rear hip joint angle to assist the genera-
tion of NAHEP during the block phase [3]. This is supported by results of regression analysis
in the current study showing that the difference between the US and AS conditions in the ante-
roposterior front/starting line and inter-block distances predicted the difference between the
two conditions in the rear hip angle (R2 = 0.39 for both distances). This suggests that the action
of the rear hip extensor can be enhanced by adjusting the anteroposterior block distances. This
finding is supported by a study by Slawinski and colleagues [12] reporting changes in the rear
and front hip angular velocity among three different inter-block distances (bunched, medium
and elongated). What is more, in the AS condition, the more flexed front hip, front knee and
rear ankle joint angles as well as the shorter front block/starting line distance led sprinters to
assume a lower crouched set position, (i.e., the centre of mass is closer to the ground). In the
literature it is reported that a crouched set position is able to generate greater H_ BV [7,18].
From a biomechanical point view, the ability to leave the blocks at a high velocity depends
on the horizontal force impulse on the blocks during the pushing phase [7]. In accordance
with classic mechanical physics, impulse is equal to the product of force and time. Conse-
quently, a higher block velocity could either be due to an increase in the net propulsion force
generated or to the push duration. In our sample, no significant difference in the duration of
the applied force (TBT) was found between the two starting conditions, thus the increase asso-
ciated with the AS in H_BV (+4%) was due to an increased force production, not to an
increase in the duration of the push against the blocks. The H_BV is a commonly measured
variable when evaluating sprint start performance [6,10,11,22]. However, we agree with the
statement of Bezodis and colleagues [2] that the NAHEP best describes the sprint start perfor-
mance because it summarizes in a single parameter how much a sprinter is able to increase
their velocity and the amount of time duration to achieve this. In our study, results showed a
greater NAHEP (+6%) in the AS vs. the US indicating that anthropometry-driven anteropos-
terior block distances may assist sprinters in translating their centre of mass in the horizontal
direction.
Starting block performance in sprinters
PLOS ONE | https://doi.org/10.1371/journal.pone.0213979
March 27, 2019
14 / 20
In line with previous studies underlining the importance of the rear leg [1,3,7,16,18] in the
sprint start, another important result in the current study was the significant increase observed
in the AS condition for several kinetic parameters in the rear block such as the RPF, RFimpulse
and the Ratio_rear by respectively +12.47%, +7.19%, and +1.32%. It is likely that the greater
force generated by the rear leg allowed sprinters to achieve significantly greater H_BV.
According to Slawinski et al. [7], the ability of faster sprinters to leave the blocks at a higher
velocity depends on the rear block total force and the rate of force development.
When considering the peak force components on the front and rear blocks, we found sig-
nificantly higher horizontal and vertical peak forces at the rear block in the AS vs. the US
(+8.68% and +4.54%, respectively. However, the difference in the mean value on peak force
between the front and the rear blocks was -3.53 N/kg (-21.26%) and -5.06 N/kg (-30.66%) in
the AS and US conditions, respectively. A similar pattern was observed for the Total Fimpulse
which was -2.17 Ns/kg (-60.95%) and -2.19 Ns/kg (-62.93%) in the AS and US conditions,
respectively. The finding of a smaller force difference between the rear and front legs in the AS
condition suggests that sprinters are able to get a more balanced force generation between the
rear and front legs. Although results showed that sprinters generated higher FPF than RPF in
both conditions, in the AS condition the RPF was greater than in the US condition. Recent
studies have reported that the generation of greater forces against the rear block was the stron-
gest predictor of sprint start performance, suggesting that forces at the rear block need to be
maximised to increase performance [16,19].
In our study, the results of the regression analysis showed that the difference in the front
block/starting line distance between the AS and US conditions is able to predict the difference
between the two conditions in RPF, the horizontal and vertical components of RPF, as well as
RFI (Table 4). These findings suggest that the front block/starting line distance is an important
predictor of the ability to generate greater rear block force. On the other hand, results of the
present study showed that the difference between the AS and US conditions in the front block/
starting line distance is not able to predict the difference between the two conditions in the
Ratio_rear. This suggests that the reduction of the front block/starting line distance does not
increase the ability to direct the forces in a more horizontal direction. This is in line with a
recent study [16] highlighting that the ratio of horizontal to resultant impulse against the rear
block was less important in predicting sprint start performance because of its low correlation
with block phase performance (standardized regression coefficient = 0.010).
The changes of the lower limb joint angles at the set position showed in the AS condition
and subsequent improvement in several kinetic and kinematic parameters measured in the
pushing phase, might explain the significant differences between the two conditions in the
acceleration phase. In fact, several studies underpinned the importance of the block phase in
the subsequent sprinting times at 5 m and 10 m as well as in the first two step lengths
[6,7,9,12]. In our sample, the AS condition was associated with a decrease in the times at 5 m
and 10 m (-3.59% and -2.32%, respectively) and an increase of the first and second stride
lengths (+2.67% and +3,36%, respectively) vs. the US condition. This is in agreement with pre-
vious studies showing that the contributions of the hip and ankle joints to force production
play an important role in the acceleration phase of the sprint start [3,6,15,45–47].
The findings of our study expand on previous findings by showing that a number of pos-
tural, kinematic and kinetic parameters of the sprint start change when setting the starting
blocks according to anthropometry causing sprinters to adopt a medium start. In addition, the
findings highlight the importance of the front/starting line distance as predictor of certain
kinetic and kinematic variables. However, further studies are needed to better understand how
the kinetic and kinematic parameters are related to this block distance. Interestingly enough,
modifying the anteroposterior block distances, allowed for an immediate improvement of
Starting block performance in sprinters
PLOS ONE | https://doi.org/10.1371/journal.pone.0213979
March 27, 2019
15 / 20
performance in the sprinters with no period of familiarization. However, achieving an optimal
automatized movement is a long-term process. Accordingly, further studies should investigate
the retention of improvement of performance after at least 24/48 hours.
The second aim of the present study was to assess whether an interaction exists between the
two block setting conditions (US and AS) and the body proportionality of the sprinters (bra-
chycormic, metricormic and macrocormic) affecting the kinematic and kinetic parameters of
sprint start. In other words, the present study tested if the kinematic and kinetic parameters of
the sprint start were different between the US and AS conditions depending on the body pro-
portionality of the sprinters.
It is important to underline that the three Cormic Index groups were similar in age, sprint-
ing experience, several anthropometric parameters (e.g. BMI, lower limb circumferences),
SLJ-relative (Table 1). All of the measured kinematic and kinetic parameters in the US condi-
tion were similar as well (Table 2). This suggests that the three groups were largely comparable.
In the US condition the three groups adopted a similar inter-block distance whereas the front
block/starting line distance was lower in the metricormic and macrocormic groups in compar-
ison with the brachycormic group (Table 2). Actually, the front block/starting line distance
increased with the mean trunk length (sitting height) of the sprinters (Tables 1 and 2). Taken
together, these findings suggest that body proportionality has some effect on the front block/
starting line distance when the sprinter adopts a block setting on the basis of personal
sensation.
When considering the interaction effect between the three Cormic Index groups over the
two block setting conditions, a significant effect was found for the inter-block distance but not
for the front block/starting line distance. This may be explained by the following: first, the
finding of a significantly lower limb length in the brachycormic vs. metricormic and macro-
cormic groups in the presence of similar stature (Table 1); second, the lower limb length was
the parameter used to calculate the anteroposterior block distances in the AS condition; third,
in the US condition the three Cormic Index groups chose similar inter-block distances, while
the front block/starting line distance increased together with the trunk length (Tables 2 and 3).
Intriguingly, it was found that sprinters modified the joint angles at the set position in the US
and AS conditions irrespective of their body proportionality. In fact, the changes in the front
and rear lower limb joint angles were consistent across the brachycormic, metricormic and
macrocormic groups.
A statistically significant group by condition interaction was also found in a number of per-
formance outcomes (Tables 2 and 3). In particular, the metricormic group showed an increase
in the AS condition for the RPF, (+16.52%), the H_RPF and the V_RPF (+13.12% and 11.11%,
respectively), the RFimpulse, (+15.54%) (all significant at P<0.05); a significant increase was
also found in the brachycormic group for the RPF (+14.99%) and the H_RPF (+9.82%); a sig-
nificant increase in the V_FPF in the macrocormic group (+7.90%). These findings show that
the AS condition is able to drive important improvements in kinetic variables measured in the
pushing phase only for the metricormic and in the brachycormic groups. These findings are
not attributable to differences in lower limbs muscle mass or strength, because thigh and calf
girth as well as performance in the standing long jump expressed relative to the leg length were
similar in the three Cormic Index groups (Table 1). Taken together, the above findings suggest
that the anthropometry-driven condition used in this study is not best suited for macrocormic
sprinters, while the sprinters with an intermediate proportionality between the trunk and the
lower limbs (i.e. the metricormic sprinters) are those who are more likely to receive extensive
benefits in the kinetics of the rear leg from a block setting based on their lower limb length.
A limitation of this study is the relatively small size of the three Cormic Index groups, mak-
ing it difficult to generalize our findings to performance outcomes associated with trunk and
Starting block performance in sprinters
PLOS ONE | https://doi.org/10.1371/journal.pone.0213979
March 27, 2019
16 / 20
lower limb proportionality. Moreover, 2D kinematic measurement was used, which is well
suited to data acquisition in a number of settings [48]. However, accurate analysis of joint
angles and the two first stride lengths would have benefit from 3D acquisition [49]. Accord-
ingly, the values of the stride lengths and joint angles presented herein should be interpreted
with caution as per the presence of the parallax error and the lens distortion error which have
occurred in our set-up using a 2D kinematic measurement. Despite these limitations and uti-
lizing instrumented starting blocks, the current data provides important biomechanical evi-
dence to further understand the influence of the set position on sprint start performance and
on the importance of the rear leg action.
Taken together, the findings presented in this study show that a block setting position, cal-
culated on the basis of a proportion of the leg length allows sprinters to improve performance
in both the block phase and early acceleration, thereby confirming the hypothesis that consid-
ering the athlete’s body dimensions when calculating block setting is beneficial to the sprint
start. In view of these results, future research is needed to adjust the anthropometry-driven
condition investigated in this study in relation to the body proportionality of the sprinter to
find the optimal anteroposterior block distances.
Conclusion
In conclusion, the current study confirmed the role played by an anthropometry-driven block
setting on the starting block performance and underpins the relevance of body proportionality
in calculating personalized anteroposterior block distances. The results obtained in the present
study provide new relevant information that may represent the starting point for future studies
aimed to develop new guidelines for helping coaches and athletes to identify the ideal personal
anteroposterior block distances. From a practical standpoint, the results of this study should
encourage coaches to pay more attention to the anthropometric characteristics of their athletes
when searching for a more effective block start position. It would seem that a good starting
point for further exploration of this field of research would be the inclusion of the Cormic
Index in the identification of an individual block settings. Moreover, current findings showed
that a key kinetic and kinematic determinant of the rear block performance is the front block/
starting line distance. Therefore, further research is required to elucidate the effect of this dis-
tance with kinetic and kinematic parameters to more completely understand its influence on
sprint start performance.
Supporting information
S1 File. Database.
(XLSX)
Acknowledgments
The authors thank all the participants for kind cooperation. This study was supported by a
grant from the University of Verona (Joint Projects 2009) to CM.
Author Contributions
Conceptualization: Valentina Cavedon, Chiara Milanese.
Data curation: Valentina Cavedon, Marco Sandri, Mariola Pirlo, Nicola Petrone, Chiara
Milanese.
Formal analysis: Marco Sandri.
Starting block performance in sprinters
PLOS ONE | https://doi.org/10.1371/journal.pone.0213979
March 27, 2019
17 / 20
Investigation: Mariola Pirlo.
Supervision: Valentina Cavedon, Marco Sandri, Carlo Zancanaro, Chiara Milanese.
Validation: Valentina Cavedon, Nicola Petrone, Carlo Zancanaro, Chiara Milanese.
Writing – original draft: Valentina Cavedon, Chiara Milanese.
References
1.
Fortier S, Basset FA, Mbourou GA, Fave´rial J, Teasdale N. Starting block performance in sprinters: a
statistical method for identifying discriminative parameters of effect of providing feedback over a 6-week
period. J Sports Sci Med. 2005; 4, 134–143. PMID: 24431969
2.
Bezodis NE, Salo AIT, Trewartha G. Choice of sprint start performance measure affects the perfor-
mance-based ranking within a group of sprinters: which is the most appropriate measure? Sports Bio-
mech. 2010; 9, 258–269. https://doi.org/10.1080/14763141.2010.538713 PMID: 21309300
3.
Bezodis NE, Salo AIT, Trewartha G. Relationships between lower-limb kinematics and block phase per-
formance in a cross section of sprinters. Eur J Sport Sci; 2015, 15, 118–124. https://doi.org/10.1080/
17461391.2014.928915 PMID: 24963548
4.
Čoh M, Josˇt B, Sˇ kof B, Tomazˇin K, Dolence A. Kinematic and kinetic parameters of sprint start and start
acceleration model of top sprinters. Gymnica. 1998; 28, 33–42.
5.
Čoh M, Tomazˇin K. Kinematic analysis of the sprint start and acceleration from the blocks. New Studies
in Athletics. 2006; 21, 23–33.
6.
Mero A, Kuitunen S, Harland M, Kyrolainen H, Komi PV. Effects of muscle-tendon length on joint
moment and power during sprint starts. J Sports Sci. 2006; 24, 165–173. https://doi.org/10.1080/
02640410500131753 PMID: 16368626
7.
Slawinski J, Bonnefoy A, Levêque JM, Ontanon G, Riquet A, Dumas R, Chèze L. Kinematic and kinetic
comparisons of elite and well-trained sprinters during sprint start. J Strength Cond Res. 2010; 24, 896–
905. https://doi.org/10.1519/JSC.0b013e3181ad3448 PMID: 19935105
8.
Guissard N, Duchateau J, Hainaux K. EMG and mechanical changes during sprint start at different front
blocks obliquities. Med Sci Sports Exerc. 1992; 11, 1257–1263.
9.
Slawinski J, Dumas R, Cheze L, Ontanon G, Miller C, Mazure-Bnnefoy A. 3D kinematic of bunched,
medium and elongated sprint start. Int J Sports Med. 2012; 33, 555–560. https://doi.org/10.1055/s-
0032-1304587 PMID: 22499565
10.
Mero A. Force-time characteristics and running velocity of male sprinters during a sprint start. Res. Q.
Exerc Sport. 1988; 59, 94–98.
11.
Schot PK, Knutzen KM. A biomechanical analysis of four sprint start positions. Res Q Exerc Sport.
1992; 63, 137–147. https://doi.org/10.1080/02701367.1992.10607573 PMID: 1585060
12.
Slawinski J, Dumas R, Cheze L, Ontanon G, Miller C, Mazure-Bnnefoy A. Effect of postural changes on
3D joint angular velocity during starting block phase. J Sports Sci. 2013, 31, 256–263. https://doi.org/
10.1080/02640414.2012.729076 PMID: 23062070
13.
Willwacher S, Feldker MK, Zohren S, Herrmann V, Bru¨ggemann GP. A novel method for the evaluation
and certification of false start apparatus in sprint running. Procedia Eng. 2013; 60, 124–129.
14.
Milanese C, Bertucco M, Zancanaro C. The effects of three different rear knee angles on kinematics in
the sprint start. Biol Sport. 2014; 31, 209–215. https://doi.org/10.5604/20831862.1111848 PMID:
25177099
15.
Otsuka M, Shim JK, Kurihara T, Yoshioka S, Nokata M, Isaka T. Effect of expertise on 3Dforce applica-
tion during the starting block phase and subsequent steps in sprint running. J Appl Biomech. 2014; 30,
390–400. https://doi.org/10.1123/jab.2013-0017 PMID: 24615252
16.
Willwacher S, Herrmann V, Heinrich K, Funken J, Strutzenberger G, Goldmann J-P, Braunstein B, Bra-
zil A, Irwin G, Potthast W, Bru¨ggemann G-P. Sprint start kinetics of amputee and non-amputee sprint-
ers. PLoS One. 2016; 1– 18, 2016.
17.
Ciacci S, Merni F, Bartolomei S, Di Michele R. Sprint start kinematics during competition in elite and
world-class male and female sprinters. J Sport Sci. 2017; 35, 1270–1278.
18.
Čoh M, Peharec S, Bačić P, Mackala K. Biomechanical differences in the sprint start between faster
and slower high-level sprinters. J Hum Kinet. 2017; 56, 29–38. https://doi.org/10.1515/hukin-2017-
0020 PMID: 28469741
Starting block performance in sprinters
PLOS ONE | https://doi.org/10.1371/journal.pone.0213979
March 27, 2019
18 / 20
19.
Bezodis NE, Walton SP, Nagahara R. Understanding the track and field sprint start through a functional
analysis of the external force features which contribute to higher levels of block phase performance. J
Sports Sci. 2018; 11, 1–8.
20.
Dickinson AD. The effect of foot spacing on the starting time and speed in sprinting and the relation of
physical measurements to foot spacing. Res Quart. 1934; 5, 12–19.
21.
Harland MJ, Steele JR. Biomechanics of the sprint start. Sports Med. 1997; 23, 11–20. https://doi.org/
10.2165/00007256-199723010-00002 PMID: 9017856
22.
Henry MF. Force time characteristics of the sprint start. Res Quart. 1952; 23, 301–318.
23.
Sigerseth P, Grinaker V. Effect of foot spacing on velocity in sprints. Res Quart. 1962; 33(4), 599–606.
24.
Stock M. Influence of various track starting positions on speed. Res Quart. 1962; 33, 607–614.
25.
Payne AH, Blader FB. The mechanics of the sprint start. In Vredenbregt J. & Wartenweiler J. (Eds.),
Biomechanics II. 1971; 225–231. Baltimore, MD: University Park Press.
26.
van Coppenolle H, Delecluse C, Goris M, Bohets W, Vanden Eynde E. Technology and development of
speed: Evaluation of the start, sprint and body composition of Pavoni, Cooman and Desruelles. Athlet-
ics Coach. 1989; 23, 82–90.
27.
Lemarie ED, Robertson DGE. Force-time data acquisition-system for sprint starting. Can J Sport Sci.
1990; 15, 149–152. PMID: 2383820
28.
Vucetić V, Matković BR, Sentija D. Morphological differences of elite Croatian track-and-field athletes.
Coll Antropol. 2008; 32, 863–868. PMID: 18982762
29.
Massidda M, Toselli S, Brasili P, Calò CN. Somatotype of elite Italian gymnasts. Coll Antropol. 2013;
37, 853–857. PMID: 24308228
30.
Ukwuma M. A study of the cormic index in a Southeastern Nigerian population. The Internet Journal of
Biological Anthropology. 2009; 4, 1–6.
31.
Cagnazzo F, and Cagnazzo R. Valutazione Antropometrica in clinica, riabilitazione e sport. Edi.Ermes
srl–Milano. 2009; p. 65.
32.
Faul F, Erdfelder E, Buchner A, Lang AG. Statistical power analyses using G*Power 3.1: Tests for cor-
relation and regression analyses. Behav Res Methods. 2009; 41, 1149–1160. https://doi.org/10.3758/
BRM.41.4.1149 PMID: 19897823
33.
Chamari K, Chaouachi A, Hambli M, Kaouech F, Wisløff U, Castagna C. The five-jump test for distance
as a field test to assess lower limb explosive power in soccer players. 2008; 22, 944–950.
34.
Lohman TG, Roche FA, Martorell R. Manuale di riferimento per la standardizzazione antropometrica.
Milano: EDRA, 1992.
35.
Milanese C, Astegno P, Pirlo M, Barbi G, Zancanaro C, Petrone N. The use of instrumented starting
blocks for sprint training. 19th Annual Congress of the European College of Sport Sciences, Amster-
dam–the Netherlands, 2–5 July 2014; p. 283.
36.
Bartlett RM. Introduction to sports biomechanics: Analysing human movement patterns. London, New
York, N.Y.: Routledge. 2007; p. 126–129.
37.
Morin JB, Edouard P, Samozino P. Technical ability of force application as a determinant factor of sprint
performance. Med Sci Sports Exerc. 2011; 43, 1680–1688. https://doi.org/10.1249/MSS.
0b013e318216ea37 PMID: 21364480
38.
Reinschmidt C, van den Bogert AJ, Nigg BM, Lundberg A, Murphy N. Effect of skin movement on the
analysis of skeletal knee joint motion during running. J Biomech.1997; 30, 729–732. PMID: 9239553
39.
Bradshw E, Maulder P, Keogh JL. Biological movement variability during the sprint start: Performance
enhancement or hindrance? Sport Biomech. 2007; 6, 246–260.
40.
Box GEP, Cox DR. An analysis of transformations. J. R. Stat. Soc. Series B Stat. Methodol. 1964; 26,
211–252.
41.
Cohen J. Statistical Power Analysis for the Behavioral Sciences. 2nd ed. Hillsdale (NJ): Lawrence Erl
baum Associates.1988.
42.
Kraan GA, van Veen J, Snijders CJ, Storm J. Starting from standing; why step backwards? J Biomech.
2001; 34, 211–215. PMID: 11165285
43.
Mero A, Luhtanen P, Komi PV. A biomechanical study of the sprint start. Scand. J Sports Sci. 1983; 5,
20–26.
44.
Schro¨dter E, Bru¨ggemann, Willwacher S. Is soleus muscle-tendon-unit behavior related to ground-
force application during the sprint start? Int J Sports Physiol Perform. 2017; 12, 448–454. https://doi.
org/10.1123/ijspp.2015-0512 PMID: 27448392
45.
Debaere S, Delecluse C, Aerenhouts D, Hagman F, Jonkers I. Form block clearance to sprint running:
Characteristics underlying an effective transition. J Sport Sci. 2013; 31, 137–149.
Starting block performance in sprinters
PLOS ONE | https://doi.org/10.1371/journal.pone.0213979
March 27, 2019
19 / 20
46.
Brazil A, Exell T, Wilson C, Willwacher S, Bezodis I, Irwin G. Lower limb joint kinetics in the starting
blocks and first stance in athletic sprinting. J Sports Sci. 2017; 35, 1629–1635. https://doi.org/10.1080/
02640414.2016.1227465 PMID: 27598715
47.
Brazil A, Exell T, Wilson C, Willwacher S, Bezodis IN, Irwin G. Joint kinetic determinants of starting
block performance in athletic sprinting. J Sports Sci. 2018; 36, 1656–1662. https://doi.org/10.1080/
02640414.2017.1409608 PMID: 29173043
48.
Frossard L, O’Riordan A, Goodman S. Applied biomechanics for evidence-based training of Australian
elite seated throwers. International Council of Sport Science and Physical Education Perspectives
Series, 2005.
49.
Fleisig G, Nicholls R, Elliot BC, Escamilla R. Kinematics used by world class tennis players to produce
high-velocity serves. Sports Biomech. 2003; 2, 51–71. https://doi.org/10.1080/14763140308522807
PMID: 14658245
Starting block performance in sprinters
PLOS ONE | https://doi.org/10.1371/journal.pone.0213979
March 27, 2019
20 / 20
| Anthropometry-driven block setting improves starting block performance in sprinters. | 03-27-2019 | Cavedon, Valentina,Sandri, Marco,Pirlo, Mariola,Petrone, Nicola,Zancanaro, Carlo,Milanese, Chiara | eng |
PMC7827622 | International Journal of
Environmental Research
and Public Health
Article
Analysis of Traffic Crashes Caused by Motorcyclists Running
Red Lights in Guangdong Province of China
Guangnan Zhang 1, Ying Tan 2, Qiaoting Zhong 1,*
and Ruwei Hu 3
Citation: Zhang, G.; Tan, Y.;
Zhong, Q.; Hu, R. Analysis of Traffic
Crashes Caused by Motorcyclists
Running Red Lights in Guangdong
Province of China. Int. J. Environ. Res.
Public Health 2021, 18, 553.
http://doi.org/10.3390/
ijerph18020553
Received: 1 December 2020
Accepted: 6 January 2021
Published: 11 January 2021
Publisher’s Note: MDPI stays neu-
tral with regard to jurisdictional clai-
ms in published maps and institutio-
nal affiliations.
Copyright: © 2021 by the authors. Li-
censee MDPI, Basel,
Switzerland.
This article is an open access article
distributed under the terms and con-
ditions of the Creative Commons At-
tribution (CC BY)
license (https://
creativecommons.org/licenses/by/
4.0/).
1
Center for Studies of Hong Kong, Macao and Pearl River Delta, Institute of Guangdong, Hong Kong and
Macao Development Studies, Sun Yat-Sen University, Guangzhou 510275, China; [email protected]
2
School of Economics and Trade, Guangdong University of Finance, Guangzhou 510521, China;
[email protected]
3
School of Public Health, Sun Yat-Sen University, Guangzhou 510080, China; [email protected]
*
Correspondence: [email protected]
Abstract: Motorcycles are among the primary means of transport in China, and the phenomenon
of motorcyclists running red lights is becoming increasingly prevalent. Based on the traffic crash
data for 2006–2010 in Guangdong Province, China, fixed- and random-parameter logit models are
used to study the characteristics of motorcyclists, vehicles, roads, and environments involved in red
light violations and injury severity resulting from motorcyclists’ running red lights in China. Certain
factors that affect the probability of motorcyclists running red lights are identified. For instance,
while the likelihood of violating red light signals during dark conditions is lower than during light
conditions for both car drivers and pedestrians, motorcyclists have significantly increased probability
of a red light violation during dark conditions. For the resulting severe casualties in red-light-running
crashes, poor visibility is a common risk factor for motorcyclists and car drivers experiencing severe
injury. Regarding the relationship between red light violations and the severity of injuries in crashes
caused by motorcyclists running red lights, this study indicated that driving direction and time
period have inconsistent effects on the probability of red light violations and the severity of injuries.
On the one hand, the likelihood of red light violations when a motorcycle rider is turning left/right
is higher than when going straight, but this turning factor has a nonsignificant impact on the severity
of injuries; on the other hand, reversing, making a U-turn and changing lanes have nonsignificant
effects on the probability of motorcyclists’ red light violations in contrast to going straight, but have a
very significant impact on the severity of injuries. Moreover, the likelihood of red light violations
during the early morning is higher than off-peak hours, but this time factor has a negative impact on
the severity of injuries. Measures including road safety educational programs for targeted groups
and focused enforcement of traffic policy and regulations are suggested to reduce the number of
crashes and the severity of injuries resulting from motorcyclists running red lights.
Keywords: traffic violation; injury severity; road safety; risk factor; motorcyclist
1. Introduction
Running red lights is a major cause of crashes at intersections, posing a higher
risk of injury than other kinds of traffic violations [1,2]. The AAA Foundation for Traf-
fic Safety has revealed that red-light running (RLR) deaths in the United States hit a
10-year high in 2017 and 28% of crash deaths that occurred at signalized intersections
were the result of a driver running through a red light. More than 184 drivers daily
were caught failing to stop at red lights in the United Kingdom in 2015 (more details
can be found at https://www.thisismoney.co.uk/money/cars/article-3589194/The-roads-
drivers-caught-running-red-lights-revealed.html.). Among all police-reported crashes in
2015 in Thailand, 1.96% were caused by drivers violating the red light signal [3]. According
to the statistics revealed by the Ministry of Public Security, 4227 severe-injury crashes and
789 fatalities between January and October 2012 in China were attributed to RLR [4].
Int. J. Environ. Res. Public Health 2021, 18, 553. https://doi.org/10.3390/ijerph18020553
https://www.mdpi.com/journal/ijerph
Int. J. Environ. Res. Public Health 2021, 18, 553
2 of 11
Previous studies have reported various rates of running red lights among road users.
For instance, a study by Yan et al. [5] showed that the RLR rate in Changsha, China was
0.14% for motorists, which was far lower than those for motorcyclists (18.64%), bicyclists
(18.74%) and pedestrians (18.54%). Kim et al. [6] compared the odds of running red lights
between drivers and pedestrians in Hawaii, United States and concluded that drivers tend
to commit proportionately more RLR than pedestrians.
It is critical to identify the influencing factors leading to RLR behavior for different
road users in different contexts [5]. However, research efforts have focused on car drivers,
e-bike riders, cyclists and pedestrians rather than motorcycle riders, who dominate traffic
in developing countries such as China. Not only the riders of motorcycles but also the
use of motorcycles in China are common, making the problem of motorcyclists running
red lights more prominent in China than in other countries. Most of the motorcyclists in
China have not undergone formal training on how to ride a motorcycle, and riding without
number plates is quite common [7]. In addition, motorcycles in China are mainly used
for delivering food and goods, services with a high demand for efficiency, resulting in
many motorcyclists choosing to run red lights or violating traffic safety laws to complete
their tasks on time. By assessing the impact of various risk factors on motorcyclists’ RLR
violations and related accident severity in Guangdong Province of China, results arising
from this study will shed lights on the development of similar (adjusted) measures to
reduce the number of crashes and the severity of injuries resulting from motorcyclists
running red lights, and to promote road safety in other regions.
Based on previous limited studies, there are few contributing factors affecting motor-
cyclists’ RLR violations, such as human characteristics, driving conditions, and driving
environment. For human characteristics, both Chen et al. [8] and Jensupakarn and Kanit-
pong [9] revealed that male and young motorcycle riders were more likely to commit RLR
violations. Motorcyclists whose occupation is a business person/trader and students are
more likely to run red lights than other occupations [9].
The approaching speed and the direction of travel of motorcycles, helmet use, the
presence of a pillion passenger, and the distance from the subject to the stop line all
significantly affect RLR violations. Evidence shows that motorcycle riders carrying pillion
passengers are less likely to execute RLR violations [8,9], which could be explained by
the ease and stability of riding without a passenger, or a sense of responsibility for the
passenger’s safety [8]. While the approaching speed had a negative effect on RLR rates in
Jensupakarn and Kanitpong’s study [9] on motorcyclists in Thailand, Chen et al. [8] found
that motorcyclists travelling at higher speeds were more likely to commit RLR violations
in Taiwan. Jensupakarn and Kanitpong’s research [9] also suggested that motorcyclists
riding straight through an intersection are less likely to run a red light compared to when
they have to make a turn, and motorcycle riders who do not wear a helmet tend to commit
RLR violations. Finally, motorcycle riders are less likely to run a red light if they are further
from a junction when the light turns red [8].
With regard to the driving environment, evidence shows that the likelihood of mo-
torcyclists running a red light is higher at night time [8,9], in periods with lower traffic
volume [8] and during off-peak hours [5,8], but is lower on a weekend or a holiday [5].
Despite the increasing prevalence of motorcyclists running red lights in China, the
relevant literature is limited. Only a study by Yan et al. [5] examined the relationship
between RLR violation rates and the day (weekend or work day) and time period for
motorcyclists in Changsha, China. Although such observational investigation is extremely
valuable, it is costly to undertake by its very nature, and thus is performed on selected
intersections. Even if the estimates obtained are reliable, they can only reflect some specific
circumstances at a particular time and place; therefore, sampling errors may arise, and
many important variables (such as individual characteristics of motorcycle riders) may be
lost. More importantly, the extent to which the results arising from observation surveys are
broadly generalizable is unclear [10], leading to a lack of comprehensiveness in the analyses
of risk factors in red light violations and the severity of injuries caused by motorcyclists.
Int. J. Environ. Res. Public Health 2021, 18, 553
3 of 11
With this in mind, the specific objectives of this study were to identify the risk factors related
to personal characteristics, vehicle characteristics, road conditions and environmental
conditions affecting (1) motorcycle riders’ RLR violations and (2) the severity of injuries
caused by motorcycle RLR crashes, using data from the road traffic crash database of
China’s Public Security Department from 2006 to 2010 in the Guangdong Province of China.
2. Materials and Methods
2.1. Data
The data used in this study, obtained from the Guangdong Provincial Security De-
partment, were extracted from the Traffic Management Sector-Specific Incident Case Data
Report. The data were recorded and reported by the traffic police who conducted on-
scene assessments and provided feedback within 24 h to the headquarters of the Traffic
Management Department. The information was recorded according to the Code of Traffic
Crash Information issued by the Computer and Information Processing Standardization
Commission under the Security Department of the country. Each sample included detailed
indexes about the characteristics of drivers/riders, injury severity, vehicle features, road
conditions, crash time, as well as environmental conditions, such as the level, form, and
cause of the crash [2].
Reports of multi-vehicle motorcycle-related crashes occurring at intersections between
2006 and 2010 were extracted for the current research study (see Figure 1 for data inclusion
and exclusion). Among 8054 cases relevant to red light violations of motorcycle riders
together with non-traffic violation accidents, 2317 (28.7%) involved no injury (property
damage only, PDO), 3968 (49.3%) involved minor injury, 931 (11.6%) resulted in serious
injury, and 838 (10.4%) resulted in death. A lower proportion of PDO crashes than that
of crashes involved in minor injury in the dataset is inconsistent with the fact that the
number of crashes decreases with the increase in injury severity [11,12]. Given the potential
under-reporting of PDO crashes [13], PDO crashes were excluded from our analysis.
Moreover, cases with an absence of motorcycle rider characteristics (e.g., rider age), cases
involving foreign riders and cases occurring on expressways were also removed. Thus,
5304 motorcycle-related traffic crashes were selected in the final sample, among which
409 involved red light violations by motorcyclists. The China Road Traffic Accidents
Statistics Report showed that the phenomenon of running red lights was very common
during the data period. Previous studies have analyzed the red light violations in China
that happened in the same period, but with different road users or in different cities
(e.g., [2,14,15]). These works provide a comparable foundation for this study to explore the
unique factors behind motorcyclists’ RLR behaviours. Therefore, the results arising from
the current research on motorcyclists running red lights in Guangdong province of China
during 2006 to 2010 have an appropriate level of generality.
In addition, Zhang et al. [2] identified factors influencing red light violations by car
drivers, cyclists, and pedestrians from Guangdong province of China. Their results provide
a comparative foundation to study the unique factors for motorcyclists’ RLR behaviors
from the same province. Therefore, the same data were also used here to study the factors
for the severity of car drivers’ injuries in RLR crashes so that the risk factors of severe
injuries between motorcyclists and car drivers in RLR crashes could be distinguished.
Int. J. Environ. Res. Public Health 2021, 18, 553
4 of 11
Int. J. Environ. Res. Public Health 2021, 18, x FOR PEER REVIEW
4 of 11
Figure 1. Data flow diagram for analysis of crashes caused by motorcyclists running red lights.
2.2. Risk Factors
The risk factors under consideration in the current study were described in a previ-
ous study derived from the same database [2]. Four dimensions, namely personal factors,
vehicle factors, road factors and environmental factors, were established as follows.
Personal factors: Rider gender, age and occupation are considered to be potential risk
factors. Rider age is divided into four categories following the WHO’s age classification
criteria: ≤24, 25–44, 45–59, and ≥60 years. Occupation and residential registration are used
to capture the education level, income and social status of riders. Rider occupation is di-
vided into six categories: self-employed, worker, migrant worker, farmer, no occupation
and other. Residential registration is divided into rural and urban. Additionally, the im-
pact of head injury on the severity of injuries is examined.
Vehicle factors: These mainly include whether motorcycles have number plates,
whether motorcycles carry passenger(s), vehicle safety conditions and vehicle driving sta-
tus, where vehicle driving status is divided into four types: going straight, turning left,
turning right, and other.
Road factors: These mainly include road types, junction types and whether there are
physical barriers on the roads. Roads are considered as two types, i.e., general highways
(including the first-class and second-class or below highways) and urban roads (including
general urban roads and other urban roads). Junctions are divided into three types, i.e.,
fork, crossroads and other.
Environmental factors: Environmental factors include street-light conditions, weather
conditions, visibility, weekends, holidays, time periods and years. Street-light conditions
include daylight, nighttime with lighting and nighttime without lighting. While bad
weather conditions include cloudy, snowy, rainy, foggy, and very windy, poor visibility
is defined as visibility below 50 meters. Following the previous work of red light viola-
tions [2,8,9], the time period is divided into early morning hours (midnight to dawn),
morning peak hours (7:00–8:59 am), after work peak hours (5:00–7:59 pm) and other time
periods.
2.3. Statistical Data Analysis
Binary logit models are widely used in the related literature on motorcycle riders’
RLR behavior at signalized intersections (e.g., [8,9]). To facilitate comparison with the
literature using the same method, it was appropriate to adopt binary logit models in this
study to estimate the effect of different risk factors on the likelihood of the occurrence of
motorcyclists running a red light in China. Specifically, multivariate logistic regression
was conducted and the adjusted odds ratios (ORs) of significant factors and their 95%
confidence intervals (CIs) were computed using Stata 14 (StataCorp, College Station, TX,
USA).
Data obtained from the Guangdong Provincial Security Department are extracted from the
Traffic Management Sector-Specific Incident Case Data Report, covering 21,132 multi-
vehicle motorcycle-related crashes occurring at intersections with 25,013 motorcycle
riders for the period of 2006–2010.
Together with non-traffic violation accidents, 8054 samples relevant to red light violations
of motorcycle riders were selected.
Of the 8054 samples reported, 3968 (49.3%) involved minor injury, 931 (11.6%) resulted in
serious injury, and 838 (10.4%) resulted in death.
Data available for analysis in the current research (5304 in total)
Red light violations of motorcyclists (n=409)
Given the potential under-reporting of no injury crashes,
2317 (28.7%) property damage only crashes were
excluded from our samples.
EXCLUDED (Total=433)
Foreign riders (n=9)
Crashes occurred on the expressways (n=311)
Absence of motorcycle rider characteristics
(n=113)
Figure 1. Data flow diagram for analysis of crashes caused by motorcyclists running red lights.
2.2. Risk Factors
The risk factors under consideration in the current study were described in a previous
study derived from the same database [2]. Four dimensions, namely personal factors,
vehicle factors, road factors and environmental factors, were established as follows.
Personal factors: Rider gender, age and occupation are considered to be potential risk
factors. Rider age is divided into four categories following the WHO’s age classification
criteria: ≤24, 25–44, 45–59, and ≥60 years. Occupation and residential registration are
used to capture the education level, income and social status of riders. Rider occupation is
divided into six categories: self-employed, worker, migrant worker, farmer, no occupation
and other. Residential registration is divided into rural and urban. Additionally, the impact
of head injury on the severity of injuries is examined.
Vehicle factors: These mainly include whether motorcycles have number plates, whether
motorcycles carry passenger(s), vehicle safety conditions and vehicle driving status, where
vehicle driving status is divided into four types: going straight, turning left, turning right,
and other.
Road factors: These mainly include road types, junction types and whether there are
physical barriers on the roads. Roads are considered as two types, i.e., general highways
(including the first-class and second-class or below highways) and urban roads (including
general urban roads and other urban roads). Junctions are divided into three types, i.e.,
fork, crossroads and other.
Environmental factors: Environmental factors include street-light conditions, weather
conditions, visibility, weekends, holidays, time periods and years. Street-light conditions in-
clude daylight, nighttime with lighting and nighttime without lighting. While bad weather
conditions include cloudy, snowy, rainy, foggy, and very windy, poor visibility is defined
as visibility below 50 meters. Following the previous work of red light violations [2,8,9],
the time period is divided into early morning hours (midnight to dawn), morning peak
hours (7:00–8:59 a.m.), after work peak hours (5:00–7:59 p.m.) and other time periods.
2.3. Statistical Data Analysis
Binary logit models are widely used in the related literature on motorcycle riders’
RLR behavior at signalized intersections (e.g., [8,9]). To facilitate comparison with the
literature using the same method, it was appropriate to adopt binary logit models in this
study to estimate the effect of different risk factors on the likelihood of the occurrence of
motorcyclists running a red light in China. Specifically, multivariate logistic regression was
conducted and the adjusted odds ratios (ORs) of significant factors and their 95% confidence
intervals (CIs) were computed using Stata 14 (StataCorp, College Station, TX, USA).
Int. J. Environ. Res. Public Health 2021, 18, 553
5 of 11
Although motorcycle riders’ red light violations are treated as a serious problem
in terms of related injuries and fatalities in developing countries such as China, limited
research has considered the analysis of influencing factors for the severity of injuries
in motorcycle RLR crashes. Previous injury severity studies have reported a significant
correlation among unobserved effects crossing discrete injury outcome categories (e.g., [16]);
to further estimate the effect of different risk factors on the likelihood of the occurrence of
severe casualties for motorcyclists in RLR crashes in comparison with car drivers, random-
parameter logit models were applied in the current research. Random parameters were
assumed to be normally distributed and 200 Halton draws were used in this study, which
have been widely used assumptions in previous research (e.g., [13,17]).
3. Results
3.1. Sample Description
Among 5304 motorcycle-related traffic crashes, even though the motorcycle-related
crashes caused by RLR are less common (approximately 7.7%) when compared to other
causes, they are considered a serious problem. In fact, the ratio of severe injuries among all
injuries caused by red light violations of motorcyclists is as high as 44.0%, whereas this
ratio for other causes is much smaller, i.e., 38.3% for all motorcycle-related traffic crashes
(see Table 1).
Table 1. Descriptive statistics of variables.
Variables
Motorcycle-Related Crashes
(n = 5304)
Crashes Caused by Red Light
Violations of Motorcyclists
(n = 409)
Crashes Caused by Red Light
Violations of Car Drivers
(n = 435)
Frequency
Proportion
(%)
Frequency
Proportion
(%)
Frequency
Proportion
(%)
Signal violation
409
7.7
409
1
435
1
Killed or seriously injured
2031
38.3
180
44.0
126
29.0
(1)
Gender
Male
4776
90.0
374
91.4
409
94.0
(2)
Age
≤24
1133
21.4
81
19.8
47
10.8
25–44
2988
56.3
241
58.9
322
74.0
45–59
982
18.5
76
18.6
65
15.0
≥60
201
3.8
11
2.7
1
0.2
(3)
Residential registration
Rural
1818
34.3
129
31.5
60
13.8
(4)
Occupation
Farmer
1296
24.4
76
18.6
22
5.0
The self-employed
427
8.1
39
9.5
79
18.2
Worker
1053
19.9
85
20.8
74
17.0
Migrant worker
690
13.0
74
18.1
50
11.5
Unemployed
236
4.4
21
5.1
10
2.3
Other occupations
1602
30.2
114
27.9
143
32.9
(5)
Whether motorcycles
carry a passenger
No passenger
3576
67.4
273
66.7
279
64.1
(6)
Whether motorcycles
have number plates
No number plates
1539
29.0
120
29.3
7
1.6
(7)
Vehicle safety condition
Unfit
394
7.4
17
4.2
16
3.7
(8)
Vehicle driving status
Straight
4524
85.3
326
79.7
319
73.3
Turning left
308
5.8
47
11.5
74
17.0
Turning right
83
1.6
8
2.0
14
3.2
Others
389
7.3
28
6.8
28
6.4
Int. J. Environ. Res. Public Health 2021, 18, 553
6 of 11
Table 1. Cont.
Variables
Motorcycle-Related Crashes
(n = 5304)
Crashes Caused by Red Light
Violations of Motorcyclists
(n = 409)
Crashes Caused by Red Light
Violations of Car Drivers
(n = 435)
Frequency
Proportion
(%)
Frequency
Proportion
(%)
Frequency
Proportion
(%)
(9)
Type of road
First-class highways
828
15.6
53
13.0
45
10.3
Second-class or below
highways
2223
41.9
167
40.8
118
27.1
General urban roads
1663
31.4
164
40.1
208
47.8
Other urban roads
590
11.1
25
6.1
64
14.7
(10)
Type of junctions
Fork
441
8.3
31
7.6
27
6.2
Crossroads
464
8.7
88
21.5
81
18.6
Others
4399
83.0
290
70.9
327
75.2
(11)
Whether there are
physical barriers in roads
No physical barriers
3182
60.0
192
46.9
193
44.4
(12)
Visibility
Bad visibility
525
9.9
37
9.0
39
9.0
(13)
Street-light condition
Daylight
2938
55.4
218
53.3
212
48.7
Dark but lighted
1527
28.8
167
40.8
204
46.9
Dark
839
15.8
24
5.9
19
4.4
(14)
Weather condition
Bad weather condition
1089
20.5
71
17.4
98
22.5
(15)
Holiday
Holiday
381
7.2
30
7.3
31
7.1
(16)
Day of the week
Weekends
1401
26.4
109
26.7
135
31.0
(17)
Time of day
Early morning
804
15.2
64
15.6
98
22.5
Morning peak hours
712
13.4
51
12.5
48
11.0
After work peak hours
911
17.2
63
15.4
58
13.3
Others
2877
54.2
231
56.5
231
53.1
(18)
Year
2006
922
17.4
108
26.4
107
24.6
2007
960
18.1
64
15.6
74
17.1
2008
1075
20.3
71
17.4
75
17.2
2009
1127
21.2
85
20.8
78
17.9
2010
1220
23.0
81
19.8
101
23.2
(19)
Injured parts
Head
1390
26.2
101
24.7
15
3.4
3.2. Risk Factors Affecting Motorcyclists Running Red Lights
The first model in Table 2 shows the risk factors associated with motorcyclists running
red lights. Concerning personal characteristics, the probability of male riders running red
lights is 1.48 times greater than that of female riders, which is consistent with conclusions in
most of the literature. Compared with motorcyclists over the age of 60, young motorcyclists
under the age of 24 (OR = 1.61) shows a significant increase in the probability of running
red lights. Chen et al. [8] argued that the young-rider effect could be explained by the
fact that young riders, in general, tend to demonstrate risk-taking road behaviors. The
risk of running red lights among migrant worker motorcyclists is significantly higher
than that among farmers (OR = 1.23), but the impact of residential registration of rider is
insignificant.
Int. J. Environ. Res. Public Health 2021, 18, 553
7 of 11
Among vehicle factors, whether motorcycles carry passengers, whether motorcycles
have number plates, vehicle safety conditions and vehicle driving status are significant
factors for the probability of motorcyclists running red lights. The probability of running
a red light by a motorcyclist riding alone is 1.18 times the probability with passengers.
Such results with regard to the presence of a pillion passenger are similar to the study
conducted by Chen et al. [8] and Jensupakarn and Kanitpong [9]. The probability of a
motorcyclist running a red light is higher when riding a motorcycle that does not satisfy
safety requirements (OR = 1.25). However, the probability of a motorcyclist running a red
light is lower when riding a motorcycle that does have a number plate (OR = 0.88) As
reported in Jensupakarn and Kanitpong’s study [9], the driving direction of a motorcycle
also affects whether a motorcyclist runs a red light: compared to driving straight ahead,
the probability of a motorcyclist running a red light is significantly higher when they are
turning (left: OR = 1.52, right: OR = 1.71, respectively).
Table 2. Factors influencing motorcyclists running red lights and resulting severe casualties.
Factors
Red Light Violations for
Motorcyclists
Severe Casualties for Motorcyclists
in Red-Light-Running Crashes
Severe Casualties for Car Drivers
in Red-Light-Running Crashes
ORs
(95% CI)
ORs (s.d.)
(95% CI)
ORs (s.d.)
(95% CI)
n
5304
409
435
(1) Personal Factors
Gender of rider (base: female)
Male
1.48 ***
0.17 **
[1.20, 1.82]
[0.04, 0.70]
Age of rider (base: ≥60)
≤24
1.61***
1.38 (35.53 *)
7.11 **
[1.15, 2.26]
[0.16, 12.11]
[1.53, 33.05]
25–44
2.79 *
[0.92, 8.45]
Residential registration of rider (base: urban)
Rural
3.21 *
[0.87, 11.80]
Occupation of rider (base: farmer)
The self-employed
0.19 *
[0.03, 1.36]
Migrant worker
1.23 *
[1, 1.52]
Worker
0.14 **
[0.02, 0.97]
Injured parts (base: others)
Head
NA
11.80 ***
[3.15, 44.15]
(2) Vehicle factors
Carry passenger (base: yes)
No passenger
1.18 **
0.33 ***
[1.04, 1.34]
[0.16, 0.71]
Whether motorcycles have number plates (base: yes)
No number plates
0.88 *
[0.77, 1.01]
Vehicle safety condition (base: fit)
Unfit
1.25 **
[1.01, 1.56]
Vehicle driving status (base: straight)
Turning left
1.52 ***
[1.19, 1.93]
Turning right
1.71 **
[1.09, 2.68]
Int. J. Environ. Res. Public Health 2021, 18, 553
8 of 11
Table 2. Cont.
Factors
Red Light Violations for
Motorcyclists
Severe Casualties for Motorcyclists
in Red-Light-Running Crashes
Severe Casualties for Car Drivers
in Red-Light-Running Crashes
ORs
(95% CI)
ORs (s.d.)
(95% CI)
ORs (s.d.)
(95% CI)
n
5304
409
435
Others
10.24 ***
[2.33, 44.95]
(3) Road factors
Type of road (base: first-class highways)
Second-class or
below highways
1.29 ***
1.35 (11.60*)
6.06 **
[1.07, 1.54]
[0.40, 4.55]
[1.41, 26.12]
General urban roads
1.22
[0.33, 4.59]
Other urban roads
1.23 *
1.54
[0.97, 1.56]
[0.33, 7.26]
Type of junction (base: others)
Fork
0.62 ***
[0.50, 0.78]
Crossroads
0.62 ***
[0.50, 0.78]
(4) Environmental factors
Visibility (base: others)
Poor visibility
3.28 *
3.15 **
[0.91, 11.87]
[1.14, 8.66]
Street-light condition (base: dark)
Daylight
0.57 ***
0.22 *
0.85 (73.51 **)
[0.48, 0.69]
[0.04, 1.15]
[0.09, 8.31]
Dark but lighted
0.75 ***
2.45
[0.62, 0.91]
[0.54, 11.15]
Weather condition (base: others)
Bad weather
condition
0.29 **
[0.09, 0.92]
Holiday (base: others)
Holiday
0.13 *
[0.02, 1.04]
Time of day (base: others)
Early morning
1.27 ***
0.35 *
[1.07, 1.51]
[0.11, 1.12]
Morning peak hours
After work peak
hours
0.16 ***
[0.04, 0.62]
log likelihood
−3371.41
−219.13
−214.66
AIC
6814.83
516.27
511.31
For brevity, insignificant results are omitted, and standard deviations are presented for significant random variables only. * p < 0.1,
** p < 0.05, *** p < 0.01.
Compared with the probability of motorcyclists running red lights on first-class
highways, the probabilities of RLR violations by motorcyclists occurring on second-class or
below highways, and other urban roads were 1.29 times and 1.23 times higher, respectively.
For motorcyclists, the risk of running red lights at forks or crossroads is significantly
lower than at other types of intersections. This result differs from the findings of previous
studies (e.g., [4]), likely because the surrounding conditions of motorcyclists also influence
their likelihood of running red lights. Street lighting can significantly reduce the risk of
a motorcyclist running a red light: the risk of running a red light is lower in the daytime
(OR = 0.57) and at night with lighting (OR = 0.75) than that at night without lighting. The
risk of red light violations by motorcyclists is higher during early morning (OR = 1.27)
Int. J. Environ. Res. Public Health 2021, 18, 553
9 of 11
than during off-peak hours. However, the impact of peak hours on motorcyclists’ red light
violations is insignificant, which is inconsistent with the literature [5,8]. Differing from
Yan et al. [5], neither weekends nor holidays have significant effects on motorcyclists’ red
light violations.
3.3. Factors for the Severity of Injuries in Red-Light-Running Crashes
Although motorcyclists are less likely to be involved in RLR traffic accidents than car
drivers (7.7% vs. 9.6%), crashes involving motorcyclists running a red light have higher
odds of resulting in severe casualties than car drivers (44% vs. 29%) in Guangdong, China;
therefore, further comparisons between motorcyclists and car drivers were carried out
to examine risk factors related to the severity of outcome in different types of vehicle
crashes involving a red light violation. Specifically, random-parameter logit models were
conducted for two sub-samples separately, i.e., motorcyclists and car drivers in RLR crashes.
In cases of motorcyclists in RLR crashes, as shown in the second model in Table 2, the
estimated parameters for young motorcycle riders were insignificant and random, with a
mean of 1.38 and a standard deviation of 35.53. Motorcycle riders registered as residing
in rural areas were more likely to get severely injured than those with urban residential
registration (OR = 3.21). Furthermore, a motorcyclist’s occupation impacts the level of
injury. The probability of serious casualties sustained by the self-employed (OR = 0.19)
and workers (OR = 0.14) was significantly lower than that of farmers, but there were no
significant differences between the probabilities of migrant workers, the unemployed and
those employed in other professions. When a head injury occurs for the motorcycle rider in
a RLR crash, the risk of serious casualties increases (OR = 11.80). With regard to the vehicle
driving state, a motorcycle rider was more strongly associated with serious casualties when
reversing, making a U-turn or changing lanes rather than going straight (OR = 10.24). The
estimated parameters for second-class or below highways were insignificant and random,
with a mean of 1.35 and a standard deviation of 11.60. Poor visibility can significantly
increase the risk of serious casualties in motorcyclists (OR = 3.28), but daylight (OR = 0.22)
and bad weather (OR = 0.29) conditions significantly decrease the risk of serious casualties.
The risk of serious casualties in motorcyclists is lower during early morning (OR = 0.35)
and after work peak hours (OR = 0.16) than during off-peak hours.
By comparing empirical results of the injury models for motorcyclists and car drivers in
RLR crashes, this study found that the common risk factor for motorcyclists and car drivers
experiencing severe injury in RLR crashes is poor visibility. Moreover, riders/drivers’
residential registration and occupation, vehicle driving status, weather condition, time of a
day and whether the rider/driver suffers a head injury are significantly associated with
severe casualties in motorcycle crashes related to RLR but have no significant effects on car
drivers in RLR crashes.
4. Discussion
4.1. Red Light Violations and Injury Severity for Motorcyclists
By comparing the influential factors that affect the probability of motorcyclists running
red lights and the severity of injuries in crashes caused by red light violations, we found
that the common risk factor is daylight condition. Rider gender, rider age, passengers,
vehicle number plates, vehicle safety conditions, road type and junction type affect the
probability of motorcyclists running red lights but do not affect the severity of injuries.
Poor visibility does not affect the likelihood of motorcyclists’ RLR violations, but in the
event of RLR crashes, this factor often leads to serious casualties. Notably, the following
factors have inconsistent effects on the probability of red light violations and the severity of
injuries for motorcycle riders: on the one hand, the likelihood of red light violations when
a motorcycle rider is turning left/right is higher than when going straight, but this turning
factor has a nonsignificant impact on the severity of injuries; on the other hand, reversing,
making a U-turn and changing lanes have nonsignificant effects on the probability of
motorcyclists’ RLR violations in contrast to going straight, but have a very significant
Int. J. Environ. Res. Public Health 2021, 18, 553
10 of 11
impact on the severity of injuries. The likelihood of red light violations during the early
morning is higher than off-peak hours, but this time factor has a negative impact on the
severity of injuries.
4.2. Comparison Between Motorcyclists and Other Road Users
Previous studies have reported on the factors contributing to traffic signal violations
for car drivers, cyclists and pedestrians in Guangdong, China [2], where the effects of
gender have been found to be insignificant for all three groups. In contrast, in this study,
male motorcycle riders are confirmed to be more likely to violate traffic signals than female
riders. While the likelihood of violating traffic signals during dark conditions is lower
than during light conditions for both car drivers and pedestrians, motorcyclists have a
significantly increased probability of red light violations during dark conditions. Since
travelling on second-class or below highways is a factor contributing to red light violations
for car drivers, cyclists, pedestrians and motorcyclists, this factor shall not be treated as a
unique factor for motorcyclists running red lights.
For the resulting severe casualties in RLR crashes, as reported in a previous section,
poor visibility is a common risk factor for motorcyclists and car drivers experiencing severe
injury in RLR crashes; therefore, we can infer that the visibility factor is not unique to
motorcyclists with regard to severe injuries related to red light violations.
4.3. Policy Implications and Further Remarks
The empirical evidence presented in this article suggests the need for an increase
in inspections and punishment for riding a motorcycle with a poor vehicle safety status.
In addition, supervision of red light violations should be strengthened during the early
morning hours. The lack of regard for and awareness of traffic laws and safety is a major
factor affecting behavior related to running red lights. According to the empirical results in
this study, males under the age of 24 are the main target group that should be engaged in
traffic safety promotion, campaigns, and educational activities.
Due to data availability, the present study analyzed the traffic crash data for 2006–
2010 in Guangdong Province, China, which limits its implications, because road safety
trends may be different for other provinces in China and may have changed in the same
province over the years. Using data from other provinces and cities in China would be
of merit in future research. Moreover, a meta-analysis of old and new data from the
same province could help us to understand the significant differences in risky behaviors
of motorcycle riders in China. Potential factors, such as traffic volume, speed limits,
traffic light characteristics and other risky driving behaviors among motorcyclists were
not analyzed in this study. It would be worth exploring the effects of these factors on
motorcyclists’ RLR violations and injuries in the future.
5. Conclusions
Using data collected from Guangdong Province of China, this study offers insights into
risk factors related to personal characteristics, vehicle characteristics, road conditions and
environmental conditions affecting red light violations and injury severity resulting from
motorcyclists’ running red lights. Measures including road safety educational programs
for targeted groups and focused enforcement of traffic policy and regulations are suggested
to reduce the number of crashes and the severity of injuries resulting from motorcyclists
running red lights.
Author Contributions: Conceptualization, G.Z. and Y.T.; formal analysis, G.Z., Y.T. and Q.Z.; method-
ology, G.Z., Y.T. and Q.Z.; software, Q.Z.; writing—original draft, Y.T. and Q.Z.; writing—review and
editing, G.Z., Y.T., Q.Z. and R.H. All authors have read and agreed to the published version of the
manuscript.
Int. J. Environ. Res. Public Health 2021, 18, 553
11 of 11
Funding: This research was supported in part by the National Natural Science Foundation of China
grant 71573286 and Ministry of Education Project for Humanities and Social Sciences Research
(16JJDGAT006).
Institutional Review Board Statement: Not applicable.
Informed Consent Statement: Not applicable.
Data Availability Statement: No new data were created or analyzed in this study. Data sharing is
not applicable to this article.
Conflicts of Interest: The authors declare no conflict of interest. The funders had no role in the design
of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, or
in the decision to publish the results.
References
1.
Retting, R.A.; Williams, A.F.; Preusser, D.F.; Weinstein, H.B. Classifying urban crashes for countermeasure development. Accid.
Anal. Prev. 1995, 27, 281–294. [CrossRef]
2.
Zhang, G.; Tan, Y.; Jou, R. Factors influencing traffic signal violations by car drivers, cyclists, and pedestrians: A case study from
Guangdong, China. In Transportation Research Part F: Traffic Psychology and Behaviour; Charlton, S., Ed.; Elsevier: Amsterdam,
The Netherlands, 2016; Volume 42, pp. 205–216.
3.
Kanitpong, K.; Jensupakarn, A.; Jensupakarn, P.; Jiwattanakulpaisarn, P. National Statistics of Traffic Accident in Thailand 2015;
ThaiRoads Foundation: Bangkok, Thailand, 2015.
4.
Wang, X.; Yu, R.; Zhong, C. A field investigation of red-light-running in Shanghai, China. In Transportation Research Part F: Traffic
Psychology Behaviour; Charlton, S., Ed.; Elsevier: Amsterdam, The Netherlands, 2016; Volume 37, pp. 144–153.
5.
Yan, F.; Li, B.; Zhang, W.; Hu, G. Red-light running rates at five intersections by road user in Changsha, China: An observational
study. Accid. Anal. Prev. 2016, 95, 381–386. [CrossRef] [PubMed]
6.
Kim, K.; Brunner, I.M.; Yamashita, E. Modeling violation of Hawaii’s crosswalk law. Accid. Anal. Prev. 2008, 40, 894–904.
[CrossRef] [PubMed]
7.
Wu, C.Y.H.; Loo, B.P.Y. Motorcycle safety among motorcycle taxi drivers and nonoccupational motorcyclists in developing
countries: A case study of Maoming, South China. Traffic Inj. Prev. 2016, 17, 170–175. [CrossRef] [PubMed]
8.
Chen, P.L.; Pai, C.W.; Jou, R.C.; Saleh, W.; Kuo, M.S. Exploring motorcycle red-light violation in response to pedestrian green
signal countdown device. Accid. Anal. Prev. 2015, 75, 128–136. [CrossRef] [PubMed]
9.
Jensupakarn, A.; Kanitpong, K. Influences of motorcycle rider and driver characteristics and road environment on red light
running behavior at signalized intersections. Accid. Anal. Prev. 2018, 113, 317–324. [CrossRef] [PubMed]
10.
Levitt, S.; Porter, J. How dangerous are drinking drivers? J. Political Econ. 2001, 109, 1198–1237. [CrossRef]
11.
Elvik, R.; Mysen, A. Incomplete accident reporting: Meta-analysis of studies made in 13 countries. Transp. Res. Rec. 1999, 1665,
133–140. [CrossRef]
12.
Ahmed, A.; Sadullah, A.F.M.; Yahya, A.S. Errors in accident data, its types, causes and methods of rectification-analysis of the
literature. Accid. Anal. Prev. 2019, 130, 3–21. [CrossRef] [PubMed]
13.
Chang, F.; Xu, P.; Zhou, H.; Chan, A.H.; Huang, H. Investigating injury severities of motorcycle riders: A two-step method
integrating latent class cluster analysis and random parameters logit model. Accid. Anal. Prev. 2019, 131, 316–326. [CrossRef]
[PubMed]
14.
Wu, C.; Yao, L.; Zhang, K. The red-light running behavior of electric bike riders and cyclists at urban intersections in China: An
observational study. Accid. Anal. Prev. 2012, 49, 186–192. [CrossRef] [PubMed]
15.
Long, K.; Liu, Y.; Han, L.D. Impact of countdown timer on driving maneuvers after the yellow onset at signalized intersections:
An empirical study in Changsha, China. Saf. Sci. 2013, 54, 8–16. [CrossRef]
16.
Scheneider, W.; Savolainen, P.; Zimmerman, K. Driver injury severity resulting from single-vehicle crashes along horizontal
curves and rural two-lane highways. Transp. Res. Rec. 2012, 1, 85–92. [CrossRef]
17.
Wang, W.; Yuan, Z.; Liu, Y.; Yang, X.; Yang, Y. A random parameter logit model of immediate red-light running behavior of
pedestrians and cyclists at major-major intersections. J. Adv. Transp. 2019, 2019, 1–13. [CrossRef]
| Analysis of Traffic Crashes Caused by Motorcyclists Running Red Lights in Guangdong Province of China. | 01-11-2021 | Zhang, Guangnan,Tan, Ying,Zhong, Qiaoting,Hu, Ruwei | eng |
PMC6939913 | royalsocietypublishing.org/journal/rspb
Research
Cite this article: Bohm S, Mersmann F,
Santuz A, Arampatzis A. 2019 The force–
length–velocity potential of the human soleus
muscle is related to the energetic cost of
running. Proc. R. Soc. B 286: 20192560.
http://dx.doi.org/10.1098/rspb.2019.2560
Received: 5 November 2019
Accepted: 26 November 2019
Subject Category:
Morphology and biomechanics
Subject Areas:
biomechanics, physiology
Keywords:
biomechanics, muscle-tendon unit,
force–length–velocity relationships, gear ratio,
running economy
Author for correspondence:
Sebastian Bohm
e-mail: [email protected]
Electronic supplementary material is available
online at https://doi.org/10.6084/m9.figshare.
c.4767113.
The force–length–velocity potential of
the human soleus muscle is related to the
energetic cost of running
Sebastian Bohm1,2, Falk Mersmann1,2, Alessandro Santuz1,2
and Adamantios Arampatzis1,2
1Department of Training and Movement Sciences, and 2Berlin School of Movement Sciences, Humboldt-
Universität zu Berlin, Berlin, Germany
SB, 0000-0002-5720-3672; FM, 0000-0001-7180-7109; AS, 0000-0002-6577-5101;
AA, 0000-0002-4985-0335
According to the force–length–velocity relationships, the muscle force
potential is determined by the operating length and velocity, which affects
the energetic cost of contraction. During running, the human soleus
muscle produces mechanical work through active shortening and provides
the majority of propulsion. The trade-off between work production and
alterations of the force–length and force–velocity potentials (i.e. fraction
of maximum force according to the force–length–velocity curves) might
mediate the energetic cost of running. By mapping the operating length
and velocity of the soleus fascicles onto the experimentally assessed force–
length and force–velocity curves, we investigated the association between
the energetic cost and the force–length–velocity potentials during running.
The fascicles operated close to optimal length (0.90 ± 0.10 L0) with moderate
velocity (0.118 ± 0.039 Vmax [maximum shortening velocity]) and, thus,
with a force–length potential of 0.92 ± 0.07 and a force–velocity potential
of 0.63 ± 0.09. The overall force–length–velocity potential was inversely
related (r = −0.52, p = 0.02) to the energetic cost, mainly determined by a
reduced
shortening
velocity.
Lower
shortening
velocity
was
largely
explained ( p < 0.001, R2 = 0.928) by greater tendon gearing, shorter Achilles
tendon lever arm, greater muscle belly gearing and smaller ankle angle vel-
ocity. Here, we provide the first experimental evidence that lower shortening
velocities of the soleus muscle improve running economy.
1. Background
Humans are capable runners compared with most other mammals and it has
been suggested that the endurance performance has been a crucial aspect for
human evolution [1]. Running economy is an important physiological factor
for endurance performance [2] and is defined as the mass-specific rate of
oxygen uptake or metabolic energy consumption at a given speed [3,4]. The
main determinant of the metabolic energy consumption during running is
the muscular force needed to support and accelerate the body mass [5]. The
level of muscle activation necessary to generate the required force is dictated
by the force–length and force–velocity potential of the muscle. The force–
length and force–velocity potential express the operating length and velocity
of the muscle fibres with respect to the force–length [6] and force–velocity
relationships [7] (i.e. fraction of maximum force according to the force–
length–velocity curves) [8,9]. When fibres operate at lower shortening velocities
and close to the optimal length, the required active muscle volume for a given
force diminishes, together with the metabolic energy expenditure [4,10]. Besides
the operating length and velocity as the main determinants, the history depen-
dence of force generation (i.e. increased force after active muscle lengthening
© 2019 The Authors. Published by the Royal Society under the terms of the Creative Commons Attribution
License http://creativecommons.org/licenses/by/4.0/, which permits unrestricted use, provided the original
author and source are credited.
[11] and decreased force after active shortening [12]) may
additionally influence the force potential. Thus, it is reason-
able to argue that the fibre dynamics of the large lower
limb muscles during running are explanatory factors of the
energetic cost and thereby endurance performance.
During human running, the soleus actively shortens [13]
and is the most important muscle for propulsion [14,15]. How-
ever, during active shortening, increased length excursion and
shortening velocity reduce the force–length–velocity potential
of muscle fibres. Due to the steep slope of the hyperbolic
force–velocity curve at low to moderate shortening velocities,
the force–velocity potential might be particularly sensitive to
changes in shortening velocity. Yet the association between
energetic cost and operating fibre dynamics as the force–
length and force–velocity potential during human running
has not been experimentally investigated thus far.
From a mechanical point of view, the soleus fibre operating
length and velocity are mainly mediated by the decoupling of
the fibre length trajectories from those of the muscle–tendon
unit (MTU), the Achilles tendon lever arm and the excursions
of the ankle joint. The decoupling of the fibre length trajectories
from the MTU is a result of tendon compliance and the variable
pennation of muscle fibres within the series muscle belly and
can be quantified by the so-called MTU gearing (i.e. ratio of
MTU and fibre velocity) [16]. Tendons, due to their compliance,
take over important portions of the length changes within the
MTU, which substantially reduce the length change and vel-
ocity of the series muscle belly. The magnitude of the
decoupling of the muscle belly from the MTU by the tendon
is expressed by the ratio of MTU velocity and belly velocity
and has been termed tendon gearing [16]. Furthermore, the
rotation of the fibres (i.e. changes of pennation angle) during
muscle shortening and concomitant changes in muscle shape
decouple the fibre length change from the length change of
the muscle belly, further decreasing the fibre shortening
length and velocity [17]. The ratio of muscle belly velocity
and fibre velocity defines the effect of the fibre rotation mech-
anism on the shortening velocity, i.e. belly gearing (or
architectural gear ratio) [16,17]. Independent of the gearing
within the soleus MTU, muscle force is transmitted through
the Achilles tendon lever arm (i.e. the distance between the
tendon’s line of action and the centre of rotation of the ankle
joint). It has been shown that shorter lever arms of the Achilles
tendon are correlated with lower rates of energy consumption
during running [18,19]. The lower energy consumption has
been attributed to a greater energy storage and return by the
Achilles tendon due to the higher muscle force required for a
given joint moment at a smaller lever arm [18]. However,
the increased energy storage and release from the tendon is
associated with a higher muscle force, which in turn increases
the metabolic cost, counteracting or even deteriorating the
effects of increased energy storage and release [20]. Yet shorter
lever arms can reduce the fibre length excursions and fibre
shortening velocity of the soleus muscle at a given ankle joint
excursion during the stance phase, which can increase the
muscle force–length–velocity potential. Therefore, besides
the debated benefits in terms of energy storage and release, a
reduction of the fibre operating length changes and velocity
by a shorter lever arm of the Achilles tendon could be an
important mechanism for the improvement in running econ-
omy. Furthermore, the ankle joint excursion during the
stance phase of running may also influence the operating
length and velocity of the soleus fibres [21]. Although gearing
within the MTU contributes to the decoupling of fibre length
and MTU length trajectories, smaller ankle joint excursions
can decrease fibre length changes and velocities. In fact,
Cavagna & Kaneko [22] as well as Williams & Cavanagh [23]
reported reduced ankle joint excursions in runners with
higher running economy than others [22,23].
In the present study, we investigated the operating length
and velocity of the soleus muscle fascicles (i.e. bundles of
fibres) during running as a function of the experimentally
determined force–length and assessed force–velocity relation-
ships (i.e. force–length and force–velocity potential) and their
association to the energetic cost of running. We further
assessed tendon and belly gearing as well as Achilles tendon
lever arm and ankle joint excursions during the stance phase
of running as mediating factors for the fascicle operating
length and velocity. We hypothesized the force–length–
velocity potential to be associated with the energetic cost of
running, mainly due to the sensitivity of the force–velocity
potential to modulations of fascicle velocity. Finally, we
expected that gearing, tendon lever arm and joint excursion
would explain the majority of the fascicle velocity variability
in the soleus muscle during running.
2. Methods
(a) Experimental design
Nineteen healthy (age: 29 ± 6 years, height: 177 ± 9 cm, mass:
69 ± 9 kg, 7 female), ambitious runners who trained at least three
times per week participated in the present study. The ethics com-
mittee of the Humboldt-Universität zu Berlin approved the study
and the participants gave written informed consent in accordance
with the Declaration of Helsinki.
After familiarization, the participants ran on a treadmill at
2.5 m s−1 for 4 min. By integrating ultrasonography, electromyogra-
phy (EMG) and kinematic data, we measured muscle architectural
parameters (fascicle length, pennation angle and thickness) and
EMG activity, and assessed the MTU length of the soleus muscle
as well as the ankle joint angle of the right leg. Energetic cost of
running was determined by expired gas analysis during an
additional 10 min running trial at the same speed. In the second
part of the experiment, the individual force–fascicle length relation-
ship of the soleus was experimentally determined by means of
maximal isometric voluntary plantar flexion contractions (MVC)
of the right leg at different ankle joint angles on a dynamometer
in combination with ultrasound imaging of the soleus muscle fasci-
cles. The force applied to the Achilles tendon was calculated from
the ankle joint moment and the individual tendon lever arm. The
derived optimal fascicle length for force production was further
used to determine the force–velocity relationship of the soleus fas-
cicles. The order of the two parts of the experiments (running
and MVC) was randomized, yet the ultrasound probe and EMG
electrodes remained attached between both ultrasound measure-
ments. Based on the assessed force–length and force–velocity
relationships, it was possible to calculate the force–length and
force–velocity potential of the soleus muscle as a function of the fas-
cicle operating length and velocity during the stance phase of
running. The product of both potentials then gives the overall
force–length–velocity potential.
(b) Assessment of the soleus force–length and
force–velocity relationship
The participants were placed in prone position on the bench
of an isokinetic dynamometer (Biodex Medical, Syst. 3, Inc.,
Shirley, NY) with the knee in fixed flexed position (approx. 120°)
royalsocietypublishing.org/journal/rspb
Proc. R. Soc. B 286: 20192560
2
to restrict the contribution of the bi-articular m. gastrocnemius to
the plantar flexion moment [24] (figure 1a). Following a standard-
ized warm-up, MVCs were performed with the right leg in eight
different joint angles, including a plateau of around 3 s. The
angles ranged from 10° plantar flexion to the individual
maximum dorsiflexion angle, set in random order in uniformly dis-
tributed intervals. The moments at the ankle joint were calculated
taking into account the effects of gravitational and passive
moments and any misalignment between ankle joint axis and
dynamometer axis by means of an established inverse dynamics
approach [25] as well as the contribution of the antagonistic
muscles by means of electromyography (description in electronic
supplementary material; figure 1a). The force applied to the
Achilles tendon during the plantar flexion MVCs was calculated
as quotient of the joint moment and the individual tendon lever
arm (description in electronic supplementary material). Soleus
fascicle behaviour during the MVCs was synchronously captured
at 30 Hz by B-mode ultrasonography (Aloka Prosound Alpha 7,
Hitachi, Tokyo, Japan) with a 6 cm linear array probe (UST-
5713T, 13.3 MHz). The probe was mounted on the shank over the
medial aspect of the soleus muscle belly by means of a custom
made antiskid neoprene/plastic cast (figure 1a). The fascicle
length was post-processed from the ultrasound images (figure 1a)
using a self-developed semi-automatic tracking algorithm [26],
described in more detail in the electronic supplementary material.
Accordingly, an individual force–fascicle length relationship
was calculated for each participant based on a second-order poly-
nomial fit (figure 1b) and the maximum muscle force applied to the
tendon (Fmax) and optimal fascicle length for force generation (L0)
was derived, respectively. Furthermore, we assessed the force–
velocity relationship of the soleus using the classical Hill equation
[7], and the muscle-specific maximum fascicle shortening velocity
(Vmax) and constants of arel and brel. Vmax was derived from the
study of Luden et al. [27], which showed Vmax values for type 1
fibres of 0.77 L0 s−1 and 2.91 L0 s−1 for type 2 fibres of the human
soleus muscle measured in vitro at 15°C [27]. Considering the temp-
erature coefficient [28], Vmax can be predicted as 4.4 L0 s−1 for type
1 fibres and 16.8 L0 s−1 for type 2 fibres under physiological temp-
erature conditions (37°C). Using an average fibre type distribution
(type 1 fibres: 81%, type 2: 19%) of the human soleus muscle
reported in literature [27,29–31], Vmax can be calculated as
6.77 L0 s−1. arel was calculated as 0.1 + 0.4FT, where FT is the fast
twitch fibre type percentage (see above), which then equals to
0.175 [32,33]. The product of arel and Vmax then gives brel as 1.182
[34]. After rearrangement of the Hill equation and extension to
the eccentric component, the operating velocity normalized
to Vmax was used to calculate the individual force potential accord-
ing to the force–velocity relationship.
(c) Assessment of joint kinematics, muscle architecture
and electromyographic activity during running
During running on a treadmill (h/p cosmos mercury, Isny,
Germany, 2.5 m s−1), kinematic data of the right leg were captured
on the basis of anatomically-referenced reflective markers (greater
trochanter, lateral femoral epicondyle and malleolus, fifth meta-
tarsal and calcaneus) by a
Vicon motion capture system
(250 Hz). A 2 min warm-up and familiarization phase on the tread-
mill preceded the captured interval. The touchdown of the foot
and toe off were defined by the kinematic data as the first and
second peak in knee extension, respectively [35]. Ultrasonic
images of the soleus were obtained synchronously with a capture
frequency of 146 Hz and soleus fascicle length was measured as
mentioned above. At least nine steps (11.1 ± 1.5) were analysed
for each participant and averaged [8]. Pennation angle was calcu-
lated based on the angle between the deeper aponeurosis and
the reference fascicle and thickness as distance between both apo-
neuroses. The corresponding length changes of the soleus muscle
belly was calculated as the product of fascicle length and the
respective cosine of the pennation angle [36]. Note that this gives
not the length of the entire soleus muscle belly but the projection
of the instant fascicle length to the plane of the MTU, which can
be used to calculate the changes of the belly length. The length
change of the soleus MTU was calculated as the product of kin-
ematic data-based ankle angle changes and the individual
Achilles tendon lever arm [37], while the initial soleus MTU
length was determined at neutral ankle joint angle based on the
regression equation provided by Hawkins & Hull [38]. The vel-
ocities of MTU, fascicles and muscle belly were calculated as the
first derivative of the MTU, fascicle and belly lengths over the
time. From these data we calculated the MTU gearing (VMTU/
VFascicle [16]), tendon gearing (VMTU/VBelly [16]) and belly gearing
(VBelly/VFascicle [16,17]), where V is the stance phase-averaged vel-
ocity of the soleus MTU, fascicles and muscle belly in absolute (i.e.
positive) values. While belly gearing expresses the effects of fasci-
cle rotation, tendon gearing expresses the effects of tendon
compliance and MTU gearing is an overall expression of the effects
of both components on the fascicle velocity [16].
Surface EMG of soleus was measured by means of the wireless
EMG system according to the procedure described above (proces-
sing description in electronic supplementary material) and
normalized to the maximum processed EMG value obtained
from all the individual MVCs (EMGmax). All parameters were
averaged over the same steps as for the muscle fascicle assessment.
(d) Energetic cost of running
After detaching the ultrasound probe, the participants continued
with a 10 min running trial at the same speed (2.5 m s−1).
(a)
(b)
upper
aponeur.
deeper
aponeur.
1 cm
F
10
4500
500
1000
1500
2000
2500
3000
3500
4000
20
30
40
50
60
70
80
fascicle length (mm)
force (N)
q
Figure 1. Experimental set-up for the determination of the soleus force–fas-
cicle length relationship. (a) Maximum isometric plantar flexions (MVC) in
eight different joint angles were performed on a dynamometer. During
the MVCs, the soleus muscle fascicle length (F), pennation angle (Θ) and
muscle thickness were measured based on ultrasound images. (b) Exemplary
force–fascicle length relationship of the soleus muscle by the MVCs (squares)
and the respective second-order polynomial fit (dashed line).
royalsocietypublishing.org/journal/rspb
Proc. R. Soc. B 286: 20192560
3
A breath-by-breath cardio pulmonary exercise testing system
(MetaLyzer 3B – R2, Cortex Biophysik GmbH, Leipzig, Germany)
was used to record the percentage of concentration of both
oxygen and carbon dioxide expired and rate of oxygen consump-
tion ( _VO2) and carbon dioxide production ( _VCO2) was calculated
as average of the last 3 min. Running economy was expressed in
units of energy by
energetic cost ¼ 16:89 _VO2 þ 4:84 _VCO2,
ð2:1Þ
where the energetic cost is expressed in [W kg−1] and _VO2 and
_VCO2 in [ml s−1 kg−1] [3,39].
(e) Statistics
Differences between soleus MTU and soleus fascicle length
changes (absolute and normalized to L0) and velocities as well as
between belly gearing and tendon gearing were tested by means
of a paired t-test for dependent samples. The Pearson correlation
coefficient was calculated in order to assess the relationship of
the energetic cost of running and the force–velocity potential,
force–length–velocity potential and fascicle velocity during the
stance phase. As normality was not given for the force–length
potential, we used the Spearman correlation coefficient to assess
its relationship to the energetic cost. A Pearson correlation coeffi-
cient was also used to analyse the relationship of EMG activity
(mean and maximum) and force–length–velocity potential. We
further conducted a multiple regression analysis to assess the mag-
nitude of the effect of the four independent variables of stance
phase-averaged tendon gearing, belly gearing, angular velocity
of the ankle joint as well as Achilles tendon lever arm on the absol-
ute soleus fascicle velocity. The statistics were performed using
SPSS Statistics (IBM Corp., Version 20.0, Armonk, USA) and the
level of significance was set to α = 0.05. All values are reported as
means and standard deviations.
3. Results
The experimentally assessed L0 was on average 41.3 ± 5.2 mm
and corresponding Fmax was 2887.1 ± 724.2 N. The assessed
Vmax based on the values of arel = 0.175 and brel = 1.182 s−1
was 279.0 ± 34.9 mm s−1. Achilles tendon lever arm showed
an average length of 56.7 ± 7.4 mm.
The averaged stance and swing times during running
were 304 ± 23 ms and 439 ± 26 ms, respectively. During the
stance phase, the ankle joint showed angles between 17.0 ±
3.8° dorsiflexion and 14.5 ± 6.0° plantarflexion (figure 2),
and rotated with an average angular velocity of 164 ± 12°/s.
The average activation of soleus normalized to EMGmax
throughout the stance phase was 0.32 ± 0.19 EMGmax and
the maximum activation was 0.52 ± 0.18 EMGmax at 40 ± 6%
of the stance phase (figure 2). While the MTU showed a
lengthening–shortening behaviour during the stance phase,
the muscle fascicles shortened continuously with signifi-
cantly less length changes as the MTU ( p < 0.001; figure 2,
table 1). The pennation angle increased coincidentally with
fascicle
shortening
while
thickness
remained
almost
unchanged (figure 2, table 1). Operating range (i.e. minimum
to maximum) of the fascicles throughout the stance phase
covered the top of the ascending limb of the force–length
curve (0.75 ± 0.09 L0 to 1.01 ± 0.12 L0; figure 3) with a mean
fascicle operating length close to the optimal length (i.e.
0.90 ± 0.10 L0). Accordingly, the averaged force–length poten-
tial of the soleus fascicles was high; (i.e. 0.92 ± 0.07; figure 3).
The soleus fascicles operated between −0.078 ± 0.045
Vmax and 0.322 ± 0.071 Vmax with an average velocity of
0
20
40
60
80
100
–30
–20
–10
0
10
20
30
ankle angle (°)
0
20
40
60
80
100
250
300
350
400
MTU length (mm)
20
40
60
80
100
0
10
20
30
40
50
60
fascicle length (mm)
20
40
60
80
100
0
10
20
30
40
50
pennation angle (°)
20
40
60
80
100
0
5
10
15
20
25
30
thickness (mm)
20
40
60
80
100
stance phase (%)
0
0.2
0.4
0.6
0.8
1.0
EMGNorm
Figure 2. Ankle angle, soleus muscle–tendon unit (MTU) length, muscle
fascicle length, pennation angle, thickness and electromyographic (EMG)
activity (normalized to maximum voluntary isometric contraction) during
the stance phase of running (2.5 m s−1). Individual (n = 19) data are
shown in thin grey lines and group averages in thick black lines.
royalsocietypublishing.org/journal/rspb
Proc. R. Soc. B 286: 20192560
4
0.118 ± 0.039 Vmax throughout the stance phase (figure 3). The
decoupling of MTU and fascicle length trajectories (figure 2)
enabled a significantly lower absolute operating velocity
of the fascicles (40.0 ± 8.2 mm s−1) compared with the MTU
(166.5 ± 27.7 mm s−1, p < 0.001), resulting in a force–velocity
potential of 0.63 ± 0.09 (figure 3). The achieved total force–
length–velocity potential of the soleus muscle during the
stance phase of running was 0.58 ± 0.10. The calculated velocity
gearing ratios were 4.46 ± 0.98 for MTU gearing, 4.03 ± 0.89 for
tendon gearing and 1.11 ± 0.07 for belly gearing. The magni-
tude of tendon gearing was significantly greater (p < 0.001)
than belly gearing (figure 4).
The energetic cost of running in the investigated velocity
of 2.5 m s−1 was in average 10.69 ± 0.96 W kg−1. An inverse
correlation was observed for the energetic cost and the
overall force–length–velocity potential of the soleus muscle
(r = −0.520, p = 0.022; figure 5). Energetic cost and the force–
velocity potential were also inversely correlated (r = −0.565,
p = 0.012; figure 5), and energetic cost and shortening velocity
were positively correlated (r = 0.561, p = 0.012), indicative for
an association of the economy of running and the operating
velocity of the soleus fascicles during running. The force–
length potential did not show any significant correlation to
the energetic cost (rs = −0.076, p = 0.759; figure 5). A significant
inverse correlation was also observed for the force–length–vel-
ocity potential and the mean (r = −0.504, p = 0.028) and the
maximal EMG activation (r = −0.525, p = 0.021).
The multiple regression model for the assessment of the
fascicle velocity during the stance phase showed a significant
explanatory power ( p < 0.001, R2 = 0.928, adjusted R2 = 0.907)
and was expressed by the equation:
Fasicle velocity¼9:788 (tendon gearing) þ 0:716 (lever arm)
42:097 (belly gearing)
þ 0:209 (ankle angle velocity) þ 51:341:
The four included independent variables were all signifi-
cant predictors ( p < 0.001 for tendon gearing, tendon lever
arm and belly gearing and p = 0.002 for ankle angular vel-
ocity). Considering the standardized coefficients of −1.006
for tendon gearing, 0.638 for lever arm, −0.367 for belly gear-
ing and 0.310 for the ankle angular velocity, the model
showed that tendon gearing and Achilles tendon lever arm
had the greatest effect on the fascicle velocity.
4. Discussion
By mapping the operating length and velocity of the human
soleus muscle during running onto the individual force–
length and force–velocity curves, we investigated the associ-
ation between the energetic cost of locomotion and the soleus
fascicle force–length and force–velocity potential. The findings
showed that the soleus fascicles operated close to the optimal
length and with moderate continuous shortening during the
stance phase. The significant inverse relationship between
the energetic cost and the force–velocity potential provides
first direct experimental evidence that the fascicle shortening
Table 1. Average values (dimension) as well as changes (range) of the
ankle joint angle (minus indicates dorsiflexion), soleus muscle–tendon unit
(MTU) length and fascicle length (absolute and normalized to optimal
length), pennation angle and muscle thickness during the stance phase of
running (n = 19).
dimension
range
ankle angle
−6.1 ± 3.6°
31.5 ± 5.2°
MTU
321.4 ± 22.3 mm
32.2 ± 8.2 mm
(79.5 ± 22.9%L0)
fascicles
36.8 ± 4.2 mm
10.6 ± 3.0 mm*
(25.9 ± 7.8%L0*)
pennation angle
24.0 ± 5.1°
8.9 ± 3.1°
thickness
15.0 ± 3.3 mm
1.7 ± 1.0 mm
*Statistically significant difference to MTU (p < 0.05).
0.4
0.6
0.8
1.0
1.2
1.4
1.6
LNorm (L/L0)
0
0.2
0.4
0.6
0.8
1.0
1.2
–0.4 –0.2
0
0.2
0.4
0.6
0.8
1.0
1.2
VNorm (V/Vmax)
0
0.2
0.4
0.6
0.8
1.0
1.2
1.4
FNorm (F/Fmax)
FNorm (F/Fmax)
Figure 3. Operating length and velocity of soleus muscle fascicles during the stance phase of running mapped onto the averaged normalized force–length and
force–velocity curve. White circles indicate the average operating length and velocity of the stance phase of each participant and the black circle the respective group
average with the standard deviation of all participants (n = 19). The grey shaded areas illustrate the operating range (maximum to minimum) of the operating
length and velocity during the stance phase averaged for all participants. Force is normalized to the maximum force during the maximal isometric plantar flexion
contractions, fascicle length to the experimentally determined optimal fascicle length and fascicle velocity to the assessed maximum shortening velocity. Dotted lines
in the left graph indicate the standard deviation of the individually measured force–length relationships. Note that the data points do not lie on the average curves
because the individual force potentials were calculated for each percentage of the stance phase of each step and then averaged step-wise, which makes a difference
to the calculation using the overall subject-based average length or velocity due to the non-linearity of the curves.
royalsocietypublishing.org/journal/rspb
Proc. R. Soc. B 286: 20192560
5
velocity of the soleus muscle is a determining factor for the
economy of human running.
The triceps surae muscle group contributes substantially
to the overall energetic cost of running [20]. The soleus is the
largest muscle of the triceps surae [40] and the main muscle
to lift and accelerate the centre of mass during locomotion
[14,15]. During the stance phase of running, the fascicles of
the soleus muscle shorten when activated, contributing to the
ankle joint mechanical work/power output [41]. The fascicles
operated on the steep-rising part of the hyperbolic-shaped
force–velocity curve, in average at 11.8% of Vmax, where
already small changes in fascicle shortening velocity cause
relevant effects on the muscle force–velocity potential. As dic-
tated by the force–velocity relationship [7], an increase in
fascicle shortening velocity is accompanied by a decrease in
the muscle force potential. The decrease of the muscle force
potential requires an upregulation of the muscle activation to
maintain the same level of force to support and accelerate the
body mass [4,10]. The observed inverse relationship between
force–velocity potential and energetic cost confirmed our
hypothesis that the soleus fascicle shortening velocity is a key
factor for the energetic cost of running. This link may further
be supported by the observed inverse correlation of EMG
activation and force–length–velocity potential, although it
should be considered that active muscle volume cannot be
assessed accurately from EMG activity. The fascicles worked
in a small range on the upper portion of the ascending limb
of the force–length curve with a high force–length potential
of 0.92. An operating range on the ascending limb close to L0
(0.75–1.01 L0) was a quite consistent observation in the investi-
gated runners, despite notable differences in the optimal
fascicle length (L0 ranging from 33 to 51 mm). In our study,
we did not find any relationship between force–length poten-
tial and energetic cost of running. However, this does not
indicate that the force–length potential is not important for
running economy, but rather that the consistently observed
high force–length potential explained less of the detected varia-
bility in the energetic cost. Besides the favourable high force–
length potential for economical force production, operating
close to optimal length may additionally preserved from rela-
tively higher energetic cost that can arise when contracting at
shorter length. In vitro evidence showed that although force
is reduced at shorter sarcomere length, the ATPase rate seems
not to differ from the rate at optimal length, indicating
0
20
40
60
80
100
stance phase (%)
–100
–50
0
velocity (mm s–1)
belly
fascicle
0
20
40
60
80
100
stance phase (%)
–400
–200
0
200
velocity (mm s–1)
MTU
belly
. .
Figure 4. Operating velocity of the soleus muscle–tendon unit (MTU) and
muscle belly (top) as well as muscle belly and fascicles (bottom) over the
stance phase, illustrating the effect of tendon gearing and belly gearing,
respectively. Grey shadings indicate the standard deviations (n = 19).
7
8
9
10
11
12
13
14
r = –0.56*
force–velocity potential
energetic cost (W kg–1)
energetic cost (W kg–1)
energetic cost (W kg–1)
7
8
9
10
11
12
13
14
rs = –0.08
force–length potential
0.2
0.4
0.6
0.8
1.0
0.2
0.4
0.6
0.8
1.0
7
8
9
10
11
12
13
14
r = –0.52*
force–length–velocity potential
0.2
0.4
0.6
0.8
1.0
Figure 5. Association of the force–length–velocity potential, force–velocity
potential and force–length potential of the soleus muscle to the energetic
cost of running. *Statistically significant correlation ( p < 0.05).
royalsocietypublishing.org/journal/rspb
Proc. R. Soc. B 286: 20192560
6
comparably higher cost of contraction at shorter length [42,43].
However, this effect seems more pronounced at very short
lengths, a portion of the force–length curve that is probably
not covered by the soleus during running (operating range
0.75–1.01 L0). Furthermore, we showed that the soleus shor-
tened continuously during the stance phase of running,
which reflects a condition for force depression. However,
since a depression of force was shown to be accompanied by
a decrease in the ATPase activity [44], force depression would
have little or no effect on the energetic cost itself.
During the stance phase, the MTU showed length changes
of 80% L0 while the fascicles showed significantly lower length
changes (i.e. 26% L0). The regression model provided evidence
that the MTU–fascicle decoupling mechanisms of tendon and
muscle belly gearing together with the Achilles tendon lever
arm and ankle joint angular velocity determine the fascicle vel-
ocity. The R2 for the model was 0.928, demonstrating a high
goodness-of-fit and a high explanation of variance of the fasci-
cle velocity. The overall MTU gearing ratio of the soleus muscle
indicated a 4.5 fold reduction of the fascicle operating velocity
during the stance phase of running. The tendon gearing ratio
of 4.03 was notably greater than the belly gearing ratio of
1.11, resulting in a higher standardized regression coefficient
(−1.006 versus −0.367). The observed gearing ratios indicate
that the soleus fascicle velocity during the stance phase of run-
ning is mainly governed by the compliance of the series elastic
element. The high portion of tendon gearing in the soleus
muscle is the consequence of greater length changes of the
Achilles tendon and aponeurosis in relation to the muscle
belly length changes. The soleus produces mechanical work/
energy for the lift and acceleration of the body throughout
the entire stance phase. In the first half, where the MTU is
elongated, the fascicles actively shorten. This means that a
part of the mechanical energy of the human body is transferred
to the tendon. Also, in this setting the muscle fascicles produce
work under favourable conditions due to the force–length and
force–velocity relationships (both potentials in this phase were
very high) and save work as strain energy in the tendon. In the
second half, the tendon strain energy is returned and at the
same time the fascicles produce work by active shortening at
a reduced force–velocity potential (fascicle shortening velocity
is higher in this phase). The higher shortening velocity is
associated with a reduction in the EMG activity and an increase
in belly gearing. It has been suggested that increased gearing at
fast shortening velocities and lower forces is a mechanism that
allows particularly slower type fascicles to be more effective in
generating forces [16]. This supports the idea that the observed
activation pattern promoted an economical MTU interaction
during running.
Belly gearing (or the fascicle rotation component) reduced
the shorting velocity of the soleus significantly by 11% in
average throughout the stance phase (ratio = 1.11). The main
contribution of the fascicle rotation component was in the
second half of the stance phase. In situ experiments have
shown that belly gearing in pennate muscles is variable
with higher ratios during low muscle force to amplify belly
shortening at lower fascicle shortening velocity and lower
ratios during higher levels of muscle force to facilitate
muscle force transmission to the tendon in concentric contrac-
tions [17]. In accordance, we found an almost constant belly
gearing of ≈1 in the soleus muscle during the first half of
the stance phase were activation and consequently force
was increased. When the soleus activation level decreased
and the MTU shortened in the second half, the pennation
angle increased and enabled a greater contribution of the fas-
cicle
rotation
component
to
the
reduction
of
fascicle
shortening velocity (maximum belly gearing ratio of 1.18).
As proposed by the variable gearing concept, the low fascicle
rotation component shown by the soleus muscle during the
first half of the stance phase where muscle activation is
increased, facilitated the orientation of the line of action of
the fascicles to the line of action of the MTU [17].
Our results provide further evidence that the Achilles
tendon lever arm and ankle angular excursions during the
stance phase were important explanatory factors of the fascicle
shortening velocity. The lever arm is an anthropometric charac-
teristic and the results showed that shorter lever arms
translated into lower fascicle shortening velocities. The associ-
ation of the Achilles tendon lever arm and fascicle shortening
velocity in the current study provides first direct experimental
evidence that shorter lever arms increase the force–velocity
potential of the soleus muscle during running. Thus, the
reduced fascicle shortening velocity due to a smaller lever
arm is—in addition to tendon and belly gearing—a mechanism
that improves running economy. Further, the association of the
angular velocity of the ankle joint and fascicle shortening vel-
ocity during the stance phase shows that greater angular
excursions and velocities and in consequence greater length
changes of the soleus MTU lead to uneconomical higher
fascicle operating velocities.
Although the soleus probably contributes to a great portion
of the overall energetic cost during running, other limb muscles
that were not considered in the present study are involved.
However, the main energy source (positive work) is the ankle
joint (41%) [41] and the soleus is the greatest muscle among
the main plantar flexors with respect to physiological cross-sec-
tional area (soleus 63%, gastroc. med. 25%, gastroc. lat. 12%)
and volume (53%, 31% and 16% [40]). The key role of soleus
is further supported by the modelling study of Hamner and
Delp (2013), which showed that the soleus is by far the biggest
contributor to the vertical acceleration and fore-aft acceleration
of the centre of mass [14]. This function is achieved by active
shortening, which reduces the force–velocity potential and
consequently requires a greater active muscle volume. In con-
trast, the quadriceps muscle group, the main contributor
during early stance, decelerating and supporting body mass
[14,15], features more economical fascicle dynamics. Recently,
we showed that the fascicles of the vastus lateralis muscle as
a representative of the quadriceps muscle group operates
with a high force–length (i.e. 0.91) and force–velocity potential
(i.e. 0.97) during the stance phase of running [8]. Operating at
high force potentials minimizes the cost of this muscle, which
is energetically expensive due to its long fascicle length
(i.e. L0 = 94 mm [8]), by reducing active muscle volume. This
may indicate that the mechanical energy by muscular work
required for steady state running is generated by muscles
that are metabolically less expensive (i.e. due to shorter fascicle
length as the soleus muscle), probably to compensate for the
reduction of the force–velocity potential.
To assessthe force–velocity potential we used a biologically
funded value of Vmax, based on in vitro studies on the human
soleus, i.e. 6.77 L0 s−1 (279.0 ± 34.9 mm s−1). However, during
submaximal running in vivo the lower activation level and
selective slow fibre type recruitment may affect the actual
force–velocity potential of the soleus muscle. To evaluate the
effect of the choice of Vmax on the observed inverse correlation
royalsocietypublishing.org/journal/rspb
Proc. R. Soc. B 286: 20192560
7
of force–velocity potential and energetic cost, a sensitivity
analysis was conducted by decreasing and increasing Vmax in
10% intervals and calculating the correlation coefficients,
respectively. The results did not show any substantial effects
on the associations between force–velocity potential and ener-
getic cost until a value of Vmax of <2.0 L0 s−1 (i.e. Vmax+30%:
r = −0.577,
Vmax+20%:
r = −0.574,
Vmax+10%:
r = −0.570,
Vmax−10%:
r = −0.559,
Vmax−20%:
r = −0.552,
Vmax−30%:
r = −0.544,
Vmax−40%:
r = −0.534,
Vmax−50%:
r = −0.522,
Vmax−60%: r = −0.506, Vmax−70%: r = −0.484; p < 0.05), which
confirms and strengthens the overserved association. Further-
more, we assessed the force–length curve during maximal
isometric contractions and used it to calculate the force–
length potential of the soleus muscle during running at
submaximal activation. There is evidence from in vitro studies
that the force–length curve depends on muscle activation
[45,46]. However, in a recent in vivo study by Fontana &
Herzog [47] on the human vastus lateralis muscle, a rightward
shift of optimal length with submaximal activation was not
observed when force was normalized to the maximum EMG
signal. The authors suggested that the shift in optimal length
phenomenon might be related to the in vitro testing set-up
(e.g. non-physiological stimulation frequency range or Ca2+
concentrations). The discrepancy of the in vitro and in vivo evi-
dence clearly warrants future investigation to elucidate the
shifting length phenomenon in the context of in vivo submaxi-
mal locomotion. Given the current human in vivo evidence [47],
we can argue that mapping the submaximal fascicle operating
length onto the force–length curve in the present in vivo study
should not affect the findings.
In the present study we focused on the understanding of
the contribution of the force–length and force–velocity poten-
tial to the energetic cost of running and we showed that the
force–velocity potential is inversely related to the energetic
cost, explaining about one-third of its variance. We argue
that an increase of active muscle volume due to the decreased
force–velocity potential would increase the energetic cost of
running. However, it must be acknowledged that the energetic
cost of muscle contraction is complex and multifactorial.
Independent of active muscle volume, in higher shortening vel-
ocities the rate of cross-bridge cycling is increased and as a
consequence so is the consumed energy. In our study, shorten-
ing velocities of the soleus muscle were on average 0.118 Vmax
throughout the stance phase, a range where the rate of ATP
hydrolysis shows a steep increase [48]. Furthermore, in sub-
maximal intensity contractions as during our investigated
running velocity selective slow fibre type activation might
decrease the energetic cost by reducing the contribution of
energetically more expensive fast twitch fibres.
5. Conclusion
In conclusion, this study provides for the first time experimen-
tal evidence that the energetic cost of running is related to the
force–length–velocity potential of the soleus muscle with lower
shortening velocities of the fascicles as the main influencing
factor (i.e. higher force–velocity potential). The main mechan-
ism for the underlying reduction of the fascicle shortening
velocity during the stance phase was gearing within the
MTU, particularly greater tendon gearing, a shorter Achilles
tendon lever arm as well as, to a minor extent, a lower ankle
angular velocity.
Ethics. The ethics committee of the Humboldt-Universität zu Berlin
approved the study and the participants gave written informed con-
sent in accordance with the Declaration of Helsinki.
Data accessibility. The datasets generated and analysed during the cur-
rent study are available at https://dx.doi.org/10.6084/m9.figshare.
10119320.v1.
Authors’ contributions. S.B., F.M. and A.A. designed research; S.B., F.M.
and A.S. performed research; S.B. analysed data; S.B. and A.A.
drafted the manuscript and F.M. and A.S. made important intellectual
contributions during revision.
Competing interests. We declare we have no competing interests.
Funding. Funding for this research was supplied by the German
Federal Institute of Sport Science (grant no. ZMVI14-070604/17-18).
Acknowledgements. We acknowledge the support by Theresa Domroes,
Antonis Ekizos and Arno Schroll for data analysis.
References
1.
Bramble DM, Lieberman DE. 2004 Endurance
running and the evolution of Homo. Nature 432,
345–352. (doi:10.1038/nature03052)
2.
Joyner MJ. 1991 Modeling: optimal marathon
performance on the basis of physiological factors.
J. Appl. Physiol. 70, 683–687. (doi:10.1152/jappl.
1991.70.2.683)
3.
Kipp S, Byrnes WC, Kram R. 2018 Calculating
metabolic energy expenditure across a wide range
of exercise intensities: the equation matters. Appl.
Physiol. Nutr. Metab. 43, 639–642. (doi:10.1139/
apnm-2017-0781)
4.
Fletcher JR, MacIntosh BR. 2017 Running economy
from a muscle energetics perspective. Front. Physiol.
8, 433. (doi:10.3389/fphys.2017.00433)
5.
Kram R, Taylor CR. 1990 Energetics of running: a
new perspective. Nature 346, 265–267. (doi:10.
1038/346265a0)
6.
Gordon AM, Huxley AF, Julian FJ. 1966 The variation
in isometric tension with sarcomere length in
vertebrate muscle fibres. J. Physiol. 184, 170–192.
(doi:10.1113/jphysiol.1966.sp007909)
7.
Hill AV. 1938 The heat of shortening and
the dynamic constants of muscle. Proc. R. Soc.
Lond. B 126, 136–195. (doi:10.1098/rspb.
1938.0050)
8.
Bohm S, Marzilger R, Mersmann F, Santuz A,
Arampatzis A. 2018 Operating length and velocity of
human vastus lateralis muscle during walking and
running. Sci. Rep. 8, 5066. (doi:10.1038/s41598-
018-23376-5)
9.
Nikolaidou ME, Marzilger R, Bohm S, Mersmann F,
Arampatzis A. 2017 Operating length and velocity of
human M. vastus lateralis fascicles during vertical
jumping. R. Soc. open sci. 4, 170185. (doi:10.1098/
rsos.170185)
10. Roberts TJ, Marsh RL, Weyand PG, Taylor CR. 1997
Muscular force in running turkeys: the economy of
minimizing work. Science 275, 1113–1115. (doi:10.
1126/science.275.5303.1113)
11. Edman KA, Elzinga G, Noble MI. 1982 Residual force
enhancement after stretch of contracting frog single
muscle fibers. J. Gen. Physiol. 80, 769–784. (doi:10.
1085/jgp.80.5.769)
12. Abbott BC, Aubert XM. 1952 The force exerted by
active striated muscle during and after change of
length. J. Physiol. 117, 77–86. (doi:10.1113/
jphysiol.1952.sp004733)
13. Rubenson J, Pires NJ, Loi HO, Pinniger GJ, Shannon
DG. 2012 On the ascent: the soleus operating length
is conserved to the ascending limb of the force–
length curve across gait mechanics in humans.
J. Exp. Biol. 215, 3539–3551. (doi:10.1242/jeb.
070466)
royalsocietypublishing.org/journal/rspb
Proc. R. Soc. B 286: 20192560
8
14. Hamner SR, Delp SL. 2013 Muscle contributions
to fore–aft and vertical body mass center
accelerations over a range of running speeds.
J. Biomech. 46, 780–787. (doi:10.1016/j.jbiomech.
2012.11.024)
15. Dorn TW, Schache AG, Pandy MG. 2012 Muscular
strategy shift in human running: dependence of
running speed on hip and ankle muscle
performance. J. Exp. Biol. 215, 1944–1956. (doi:10.
1242/jeb.064527)
16. Wakeling JM, Blake OM, Wong I, Rana M, Lee SSM.
2011 Movement mechanics as a determinate of
muscle structure, recruitment and coordination. Phil.
Trans. R. Soc. B 366, 1554–1564. (doi:10.1098/rstb.
2010.0294)
17. Azizi E, Brainerd EL, Roberts TJ. 2008 Variable
gearing in pennate muscles. Proc. Natl Acad. Sci.
USA 105, 1745–1750. (doi:10.1073/pnas.
0709212105)
18. Scholz MN, Bobbert MF, Soest AJ, Clark JR, Heerden
J. 2008 Running biomechanics: shorter heels, better
economy. J. Exp. Biol. 211, 3266–3271. (doi:10.
1242/jeb.018812)
19. Raichlen DA, Armstrong H, Lieberman DE. 2011
Calcaneus length determines running economy:
implications for endurance running performance in
modern humans and Neandertals. J. Hum. Evol. 60,
299–308. (doi:10.1016/j.jhevol.2010.11.002)
20. Fletcher JR, MacIntosh BR. 2015 Achilles tendon
strain energy in distance running: consider the
muscle energy cost. J. Appl. Physiol. 118, 193–199.
(doi:10.1152/japplphysiol.00732.2014)
21. Rassier DE, MacIntosh BR, Herzog W. 1999 Length
dependence of active force production in skeletal
muscle. J. Appl. Physiol. 86, 1445–1457. (doi:10.
1152/jappl.1999.86.5.1445)
22. Cavagna GA, Kaneko M. 1977 Mechanical work and
efficiency in level walking and running. J. Physiol.
268, 467–481. (doi:10.1113/jphysiol.1977.
sp011866)
23. Williams KR, Cavanagh PR. 1987 Relationship
between distance running mechanics, running
economy, and performance. J. Appl. Physiol. 63,
1236–1245. (doi:10.1152/jappl.1987.63.3.1236)
24. Hof AL, van den Berg JW. 1977 Linearity between
the weighted sum of the EMGs of the human
triceps surae and the total torque. J. Biomech. 10,
529–539. (doi:10.1016/0021-9290(77)90033-1)
25. Arampatzis A, Morey-Klapsing G, Karamanidis K,
DeMonte G, Stafilidis S, Brüggemann G-P. 2005
Differences between measured and resultant joint
moments during isometric contractions at the ankle
joint. J. Biomech. 38, 885–892. (doi:10.1016/j.
jbiomech.2004.04.027)
26. Marzilger R, Legerlotz K, Panteli C, Bohm S,
Arampatzis A. 2018 Reliability of a semi-automated
algorithm for the vastus lateralis muscle
architecture measurement based on ultrasound
images. Eur. J. Appl. Physiol. 118, 291–301. (doi:10.
1007/s00421-017-3769-8)
27. Luden N, Minchev K, Hayes E, Louis E, Trappe T,
Trappe S. 2008 Human vastus lateralis and soleus
muscles display divergent cellular contractile
properties. Am. J. Physiol. Regul. Integr. Comp.
Physiol. 295, R1593–R1598. (doi:10.1152/ajpregu.
90564.2008)
28. Ranatunga KW. 1984 The force-velocity relation of
rat fast- and slow-twitch muscles examined at
different temperatures. J. Physiol. 351, 517–529.
(doi:10.1113/jphysiol.1984.sp015260)
29. Edgerton VR, Smith JL, Simpson DR. 1975 Muscle
fibre type populations of human leg muscles.
Histochem. J. 7, 259–266. (doi:10.1007/
BF01003594)
30. Larsson L, Moss RL. 1993 Maximum velocity of
shortening in relation to myosin isoform
composition in single fibres from human skeletal
muscles. J. Physiol. 472, 595–614. (doi:10.1113/
jphysiol.1993.sp019964)
31. Johnson MA, Polgar J, Weightman D, Appleton D.
1973 Data on the distribution of fibre types in
thirty-six human muscles. An autopsy study.
J. Neurol. Sci. 18, 111–129. (doi:10.1016/0022-
510x(73)90023-3)
32. Winters JM, Stark L. 1985 Analysis of fundamental
human movement patterns through the use of in-
depth antagonistic muscle models. IEEE Trans.
Biomed. Eng. 32, 826–839. (doi:10.1109/TBME.
1985.325498)
33. Winters JM, Stark L. 1988 Estimated mechanical
properties of synergistic muscles involved in
movements of a variety of human joints.
J. Biomech. 21, 1027–1041. (doi:10.1016/0021-
9290(88)90249-7)
34. Miller RH, Umberger BR, Caldwell GE. 2012
Sensitivity of maximum sprinting speed to
characteristic parameters of the muscle force–
velocity relationship. J. Biomech. 45, 1406–1413.
(doi:10.1016/j.jbiomech.2012.02.024)
35. Fellin RE, Rose WC, Royer TD, Davis IS. 2010
Comparison of methods for kinematic identification
of footstrike and toe-off during overground and
treadmill running. J. Sci. Med. Sport 13, 646–650.
(doi:10.1016/j.jsams.2010.03.006)
36. Fukunaga T, Kubo K, Kawakami Y, Fukashiro S,
Kanehisa H, Maganaris CN. 2001 In vivo behaviour
of human muscle tendon during walking.
Proc. R. Soc. Lond. B 268, 229–233. (doi:10.1098/
rspb.2000.1361)
37. Lutz GJ, Rome LC. 1996 Muscle function during
jumping in frogs. I. Sarcomere length change, EMG
pattern,andjumpingperformance.Am.J.Physiol. 271,
C563–C570. (doi:10.1152/ajpcell.1996.271.2.C563)
38. Hawkins D, Hull ML. 1990 A method for
determining lower extremity muscle-tendon lengths
during flexion/extension movements. J. Biomech.
23, 487–494. (doi:10.1016/0021-9290(90)90304-L)
39. Péronnet F, Massicotte D. 1991 Table of nonprotein
respiratory quotient: an update. Can. J. Sport Sci.
16, 23–29.
40. Albracht K, Arampatzis A, Baltzopoulos V. 2008
Assessment of muscle volume and physiological
cross-sectional area of the human triceps surae
muscle in vivo. J. Biomech. 41, 2211–2218. (doi:10.
1016/j.jbiomech.2008.04.020)
41. Novacheck TF. 1998 The biomechanics of running.
Gait Posture 7, 77–95. (doi:10.1016/S0966-
6362(97)00038-6)
42. Stephenson DG, Stewart AW, Wilson GJ. 1989
Dissociation of force from myofibrillar MgATPase
and stiffness at short sarcomere lengths in rat and
toad skeletal muscle. J. Physiol. 410, 351–366.
(doi:10.1113/jphysiol.1989.sp017537)
43. Hilber K, Sun Y-B, Irving M. 2001 Effects of
sarcomere length and temperature on the rate of
ATP utilisation by rabbit psoas muscle fibres.
J. Physiol. 531, 771–780. (doi:10.1111/j.1469-7793.
2001.0771h.x)
44. Joumaa V, Fitzowich A, Herzog W. 2017 Energy cost
of isometric force production after active shortening
in skinned muscle fibres. J. Exp. Biol. 220,
1509–1515. (doi:10.1242/jeb.117622)
45. Holt NC, Azizi E. 2014 What drives activation-
dependent shifts in the force–length curve? Biol.
Lett. 10, 20140651. (doi:10.1098/rsbl.2014.0651)
46. Rack PMH, Westbury DR. 1969 The effects of length
and stimulus rate on tension in the isometric cat
soleus muscle. J. Physiol. 204, 443–460. (doi:10.
1113/jphysiol.1969.sp008923)
47. Fontana HB, Herzog W. 2016 Vastus lateralis
maximum force-generating potential occurs at
optimal fascicle length regardless of activation level.
Eur. J. Appl. Physiol. 116, 1267–1277. (doi:10.1007/
s00421-016-3381-3)
48. Barclay CJ. 2015 Energetics of contraction. Compr.
Physiol. 5, 961–995. (doi:10.1002/cphy.c140038)
royalsocietypublishing.org/journal/rspb
Proc. R. Soc. B 286: 20192560
9
| The force-length-velocity potential of the human soleus muscle is related to the energetic cost of running. | 12-18-2019 | Bohm, Sebastian,Mersmann, Falk,Santuz, Alessandro,Arampatzis, Adamantios | eng |
PMC7379642 | Supplement Table 3. Change in VO2max (L·min-1 and ml·min-1·kg-1) from 1995-1997 to 2016-2017 in the total population and by age-group.
L·min-1
ml·min-1·kg-1
L·min-1
ml·min-1·kg-1
L·min-1
ml·min-1·kg-1
Year
n
Mean (SD)
Change
Mean (SD)
Change
n
Mean (SD)
Change
Mean (SD)
Change
n
Mean (SD)
Change
Mean (SD)
Change
95-97
1 354
3.19 (0.17)
Ref
43.9 (0.72)
Ref
2 195
2.82 (0.16)
Ref
38.0 (0.52)
Ref
1 025
2.44 (0.14)
Ref
32.5 (0.68)
Ref
98-99
1 840
3.19 (0.16)
-0,1%
43.7 (0.64)
-0,5%
2 849
2.80 (0.14)
-0,6%
37.5 (0.39)
-1,4%
1 854
2.42 (0.12)
-0,7%
32.3 (0.51)
-0,7%
00-01
3 469
3.16 (0.16)
-1,0%
43.0 (0.76)
-2,0%
5 248
2.82 (0.14)
-0,1%
37.5 (0.70)
-1,3%
3 828
2.40 (0.13)
-1,7%
31.8 (0.56)
-2,1%
02-03
6 563
3.09 (0.17)
-3,3%
42.3 (0.68)
-3,7%
9 429
2.78 (0.15)
-1,4%
36.8 (0.70)
-3,1%
6 637
2.35 (0.13)
-3,7%
31.2 (0.60)
-3,9%
04-05
9 617
3.09 (0.16)
-3,0%
42.2 (0.57)
-3,9%
16 294
2.78 (0.15)
-1,3%
36.7 (0.69)
-3,5%
11 509
2.35 (0.13)
-3,6%
31.1 (0.50)
-4,4%
06-07
9 743
3.08 (0.15)
-3,4%
41.7 (0.61)
-4,9%
16 867
2.79 (0.15)
-0,9%
36.5 (0.69)
-3,9%
11 909
2.37 (0.13)
-2,9%
31.2 (0.56)
-4,1%
08-09
11 268
3.07 (0.15)
-3,7%
41.6 (0.70)
-5,2%
18 652
2.82 (0.15)
-0,1%
36.6 (0.87)
-3,6%
13 559
2.39 (0.12)
-2,0%
31.2 (0.59)
-4,1%
10-11
10 340
3.06 (0.15)
-4,0%
41.3 (0.78)
-5,8%
17 618
2.83 (0.15)
0,2%
36.5 (0.86)
-3,8%
11 219
2.41 (0.13)
-1,2%
31.1 (0.58)
-4,2%
12-13
15 737
3.04 (0.14)
-4,7%
41.0 (0.82)
-6,7%
25 651
2.79 (0.14)
-1,1%
36.2 (0.94)
-4,7%
15 858
2.38 (0.12)
-2,3%
30.8 (0.56)
-5,2%
14-15
16 428
2.99 (0.14)
-6,4%
40.1 (0.67)
-8,5%
23 725
2.75 (0.13)
-2,5%
35.6 (0.90)
-6,4%
15 431
2.37 (0.12)
-2,7%
30.5 (0.65)
-6,2%
16-17
11 944
2.98 (0.14)
-6,5%
39.9 (0.77)
-9,2%
14 693
2.73 (0.13)
-3,2%
35.3 (0.84)
-7,1%
9 924
2.38 (0.13)
-2,3%
30.5 (0.71)
-6,1%
18-34 years
35-49 years
50-74 years
| Decline in cardiorespiratory fitness in the Swedish working force between 1995 and 2017. | 11-15-2018 | Ekblom-Bak, Elin,Ekblom, Örjan,Andersson, Gunnar,Wallin, Peter,Söderling, Jonas,Hemmingsson, Erik,Ekblom, Björn | eng |
PMC5889786 | SYSTEMATIC REVIEW
Effects of Strength Training on the Physiological Determinants
of Middle- and Long-Distance Running Performance:
A Systematic Review
Richard C. Blagrove1,2 • Glyn Howatson2,3 • Philip R. Hayes2
Published online: 16 December 2017
The Author(s) 2017. This article is an open access publication
Abstract
Background Middle- and long-distance running perfor-
mance is constrained by several important aerobic and
anaerobic parameters. The efficacy of strength training
(ST) for distance runners has received considerable atten-
tion in the literature. However, to date, the results of these
studies have not been fully synthesized in a review on the
topic.
Objectives
This systematic review aimed to provide a
comprehensive critical commentary on the current litera-
ture that has examined the effects of ST modalities on the
physiological determinants and performance of middle-
and long-distance runners, and offer recommendations for
best practice.
Methods Electronic databases were searched using a
variety of key words relating to ST exercise and distance
running. This search was supplemented with citation
tracking. To be eligible for inclusion, a study was required
to meet the following criteria: participants were middle- or
long-distance runners with C 6 months experience, a ST
intervention (heavy resistance training, explosive resis-
tance training, or plyometric training) lasting C 4 weeks
was applied, a running only control group was used, data
on one or more physiological variables was reported. Two
independent assessors deemed that 24 studies fully met the
criteria for inclusion. Methodological rigor was assessed
for each study using the PEDro scale.
Results PEDro scores revealed internal validity of 4, 5, or
6 for the studies reviewed. Running economy (RE) was
measured in 20 of the studies and generally showed
improvements (2–8%) compared to a control group,
although this was not always the case. Time trial (TT)
performance (1.5–10 km) and anaerobic speed qualities
also tended to improve following ST. Other parameters
[maximal oxygen uptake ( _VO2max), velocity at
_VO2max,
blood lactate, body composition] were typically unaffected
by ST.
Conclusion Whilst there was good evidence that ST
improves RE, TT, and sprint performance, this was not a
consistent finding across all works that were reviewed.
Several important methodological differences and limita-
tions are highlighted, which may explain the discrepancies
in findings and should be considered in future investiga-
tions in this area. Importantly for the distance runner,
measures relating to body composition are not negatively
& Richard C. Blagrove
[email protected]
Glyn Howatson
[email protected]
Philip R. Hayes
[email protected]
1
Faculty of Health, Education and Life Sciences, School of
Health Sciences, Birmingham City University, City South
Campus, Westbourne Road, Edgbaston, Birmingham
B15 3TN, UK
2
Division of Sport, Exercise and Rehabilitation, Faculty of
Health and Life Sciences, Northumbria University,
Northumberland Building, Newcastle-upon-Tyne NE1 8ST,
UK
3
Water Research Group, Northwest University,
Potchefstroom, South Africa
123
Sports Med (2018) 48:1117–1149
https://doi.org/10.1007/s40279-017-0835-7
impacted by a ST intervention. The addition of two to three
ST sessions per week, which include a variety of ST
modalities are likely to provide benefits to the performance
of middle- and long-distance runners.
Key Points
Strength training (ST) appears to provide benefits to
running economy, time trial performance and
maximal sprint speed in middle- and long-distance
runners of all abilities
Maximal oxygen uptake, blood lactate parameters,
and body composition appear to be unaffected by the
addition of ST to a distance runner’s program
Adding ST, in the form of heavy resistance training,
explosive resistance training, and plyometric training
performed, on 2–3 occasions per week is likely to
positively affect performance.
1 Introduction
Distance running performance is the consequence of a
complex interaction of physiological, biomechanical, psy-
chological, environmental, and tactical factors. From a
physiological perspective, the classic model [1, 2] identi-
fies three main parameters that largely influence perfor-
mance:
maximal
oxygen
uptake
( _VO2max),
running
economy (RE), and fractional utilization (sustainable per-
centage of _VO2max). Collectively, these determinants are
capable of predicting 16 km performance with more than
95% accuracy in well-trained runners [3]. The velocity
associated with _VO2max (v _VO2max) also provides a com-
posite measure of _VO2max and RE, and has been used to
explain differences in performance amongst trained dis-
tance runners [3, 4]. Whilst _VO2max values differ little in
homogenous groups of distance runners, RE displays a
high degree of interindividual variability [5, 6]. Defined as
the oxygen or energy cost of sustaining a given sub-max-
imal running velocity, RE is underpinned by a variety of
anthropometric, physiological, biomechanical, and neuro-
muscular factors [7]. Traditionally, chronic periods of
running training have been used to enhance RE [8, 9];
however, novel approaches such as strength training (ST)
modalities have also been shown to elicit improvements
[10].
For middle-distance (800–3000 m) runners, cardiovas-
cular-related parameters associated with aerobic energy
production can explain a large proportion of the variance in
performance [11–17]. However a large contribution is also
derived from anaerobic sources of energy [14, 18].
Anaerobic capabilities can explain differences in physio-
logical profiles between middle- and longer-distance run-
ners
[14]
and
are
more
sensitive
to
discriminating
performance in groups of elite middle-distance runners
than
traditional
aerobic
parameters
[19].
Anaerobic
capacity and event-specific muscular power factors, such as
v _VO2max and the velocity achieved during a maximal
anaerobic running test (vMART) have also been proposed
as limiting factors for distance runners [12, 20, 21]. For an
800-m runner in particular, near-maximal velocities of
running are reached during the first 200 m of the race [22],
which necessitate a high capacity of the neuromuscular and
anaerobic system.
Both RE and anaerobic factors, (i.e., speed, anaerobic
capacity and vMART) rely on the generation of rapid force
during ground contact when running [23, 24]. Programs of
ST provide an overload to the neuromuscular system,
which improves motor unit recruitment, firing frequency,
musculotendinous stiffness, and intramuscular co-ordina-
tion, and therefore potentially provides distance runners
with a strategy to enhance their RE and event-specific
muscular power factors [19]. In addition, an improvement
in force-generating capacity would theoretically allow
athletes to sustain a lower percentage of maximal strength,
thereby reducing anaerobic energy contribution [25]. This
reduction in relative effort may therefore reduce RE and
blood lactate (BL) concentration. As v _VO2maxis a function
of RE, _VO2max and anaerobic power factors, it would also
be expected to show improvements following an ST
intervention. Several recent reviews in this area have pro-
vided compelling evidence that a short-term ST interven-
tion is likely to enhance RE [10, 26], in the order of * 4%
[10]. Whilst these reviews have provided valuable insight
into how ST specifically impacts RE, studies also typically
measure other important aerobic and anaerobic determi-
nants of distance running performance, which have not
previously been fully synthesized in a review. Body com-
position also appears to be an important determinant of
distance running performance, with low body mass con-
ferring an advantage [27, 28]. Resistance training (RT) is
generally associated with a hypertrophic response [29];
however, this is known to be attenuated when RT and
endurance training are performed concurrently within the
same program [30]. Changes in body composition as a
consequence of ST in distance runners have yet to be fully
addressed in reviews on this topic.
1118
R. C. Blagrove et al.
123
There are also a number of recent publications [31–38]
that have not been captured in previous reviews [10, 26] on
this topic, which potentially provide valuable additional
insight into the area. Previous papers that have reviewed
the impact of ST modalities on distance running perfor-
mance have done so alongside other endurance sports
[23, 39] or are somewhat outdated [40–42]. Furthermore,
although improvements in RE would likely confer a benefit
to distance running performance, the outcomes from
studies that have used time trials have not been compre-
hensively reviewed. Performance-related outcome mea-
sures provide high levels of external validity compared to
physiological parameters, therefore it is likely that a col-
lective summary of results would be of considerable
interest to coaches and athletes.
Consequently the aim of this review was to systemati-
cally analyze the evidence surrounding the use of ST on
distance running parameters that includes both aerobic and
anaerobic qualities, in addition to body composition and
performance-related outcomes. This work also provides a
forensic, critical evaluation that, unlike previous work,
highlights areas that future investigations should address to
improve methodological rigor, such as ensuring valid
measurement of physiological parameters and maximizing
control over potential confounding factors.
2 Methods
2.1 Literature Search Strategy
The PRISMA statement [43] was used as a basis for the
procedures described herein. Electronic database searches
were carried out in Pubmed, SPORTDiscus, and Web of
Science using the following search terms and Boolean
operators: (‘‘strength training’’ OR ‘‘resistance training’’
OR ‘‘weight training’’ OR ‘‘weight lifting’’ OR ‘‘plyo-
metric training’’ OR ‘‘concurrent training’’) AND (‘‘dis-
tance running’’ OR ‘‘endurance running’’ OR ‘‘distance
runners’’ OR ‘‘endurance runners’’ OR ‘‘middle distance
runners’’) AND (‘‘anaerobic’’ OR ‘‘sprint’’ OR ‘‘speed’’
OR ‘‘performance’’ OR ‘‘time’’ OR ‘‘economy’’ OR ‘‘en-
ergy cost’’ OR ‘‘lactate’’ OR ‘‘maximal oxygen uptake’’
OR ‘‘ _VO2max’’ OR ‘‘aerobic’’ OR ‘‘time trial’’). Searches
were limited to papers published in English and from 1
January 1980 to 6 October 2017.
2.2 Inclusion and Exclusion Criteria
For a study to be eligible, each of the following inclusion
criteria were met:
•
Participants were middle- (800–3000 m) or long-dis-
tance runners (5000 m–ultra-distance). Studies using
triathletes and duathletes were also included because
often these participants possess similar physiology to
distance runners and complete similar volumes of
running training.
•
A ST intervention was applied. This was defined as
heavy (less than 9 repetition maximum (RM) loads and/
or 80% of 1RM) or isometric resistance training (HRT),
moderate load (9–15 RM and/or 60–80% 1RM) RT,
explosive resistance training (ERT), reactive ST or
plyometric training (PT). Sprint training (SpT) could be
used in conjunction with one or more of the above ST
methods, but not exclusively as the only intervention
activity.
•
The intervention period lasted 4 weeks or longer. This
criteria was employed as neuromuscular adaptations
have been observed in as little as 4 weeks in non-
strength trained individuals [44, 45].
•
A running only control group was used that adopted
similar running training to the intervention group(s).
•
Data on one or more of the following physiological
parameters was reported: _VO2max, RE, velocity associ-
ated with v _VO2max, time trial (TT) performance, time to
exhaustion (TTE), BL response, anaerobic capacity,
maximal speed, measures of body composition.
•
Published in full in a peer-reviewed journal.
Studies were excluded if any of the following criteria
applied:
•
Participants were non-runners (e.g., students, untrai-
ned/less than 6 months running experience). Further
restrictions were not placed upon experience/training
status.
•
The running training and/or ST intervention was poorly
controlled and/or reported.
•
The intervention involved only SpT or was embedded
as part of running training sessions.
•
Participants were reported to be in poor health or
symptomatic.
•
Ergogenic aids were used as part of the intervention.
Using the mean _VO2max values provided within each
study, participants training status was considered as mod-
erately-trained (male _VO2max B 55 ml kg-1 min-1), well-
trained (male _VO2max 55–65 ml kg-1 min-1), or highly-
trained (male _VO2max C 65 ml kg-1 min-1) [10, 46]. For
female
participants, the
_VO2max
thresholds
were set
10 ml kg-1 min-1 lower [46]). In the absence of _VO2max
values, training status was based upon the training or
competitive
level
of
the
participants:
moderately-
trained = recreational
or
local
club,
well-
Effects of Strength Training on Distance Running
1119
123
trained = Collegiate or provincial, highly-trained = na-
tional or international.
2.3 Study Selection
Figure 1 provides a visual overview of the study selection
process. Search results were imported into a published software
for systematic reviews [47], which allowed a blind screening
process tobe performed bytwoindependent reviewers(RB and
PH). Any disagreements were resolved by consensus. The
initial search yielded 454 publications. Following the removal
of duplicates (n = 190), publications were filtered by reading
the title and abstract [inter-rater reliability (IRR): 95.3%,
Cohens k = 0.86] leaving 19 review articles or commentaries,
and 47 potentially relevant papers, which were given full con-
sideration. Five additional records were identified as being
potentially relevant via manual searches of previously
published reviews on this topic and the individual study cita-
tions. These 52 studies were considered in detail for appropri-
ateness, resulting in a further 26 papers [34, 37, 48–71] being
excluded (IRR: 94.2%, Cohens k = 0.88) for the following
reasons: not published in full in a peer-reviewed journal
[50, 52, 60, 61], absence of a running only control group
[48, 49, 54, 57, 59, 62–67, 69], participants were non-runners
[51, 53, 56, 68], no physiological parameters were measured
[55], dissimilar running training was applied between groups
[71], the ST interventionwaspoorly controlled[54],andSTdid
not involve one of the aforementioned types [34, 37, 58, 70].
2.4 Analysis of Results
The Physiotherapy Evidence Database (PEDro) scale was
subsequently used to assess the quality of the remaining 26
records [31–33, 36, 38, 72–92] by the two independent
reviewers. Two studies reported their results across two
papers [32, 38, 90, 92], therefore both are considered as
single studies hereafter, thus a total of 24 studies were
analyzed. The PEDro scale is a tool recommended for
assessing the quality of evidence when systematically
reviewing randomized-controlled trials [93]. Each paper is
scrutinized against 11 items relating to the scientific rigor
of the methodology, with items 2–11 being scored 0 or 1.
Papers are therefore awarded a rating from 0 to 10
depending upon the number of items which the study
methodology satisfies (10 = study possesses excellent
internal validity, 0 = study has poor internal validity). No
studies were not excluded based upon their PEDro scale
score and IRR was excellent (93.2%, Cohens k = 0.86).
Results are summarized as a percentage change and the p
value for variables relating to: strength outcomes, RE,
_VO2max, v _VO2max, BL response, time trial, anaerobic per-
formance, and body composition. Due to the heterogeneity
of outcome measures in the included studies and the limi-
tations associated with conditional probability, where pos-
sible, an effect size (ES) statistic (Cohens d) is also provided.
Effect size values are based upon those reportedin the studies
or were calculated using the ratio between the change score
(post-intervention value minus pre-intervention value) and a
pooled standard deviation at baseline for intervention and
control groups. Values are interpreted as trivial\0.2; small
0.2–0.6; moderate 0.6–1.2; and large[1.2.
3 Results
3.1 Participant Characteristics
A summary of the participant characteristics for the 24
studies which met the criteria for inclusion in this review is
presented in Table 1. A total of 469 participants (male
Fig. 1 Search, screening and selection process for suitable studies
1120
R. C. Blagrove et al.
123
Table 1 Participant characteristics and design of each study
Study
Participant characteristics
Study design
n (I/C)
Sex
Age (years)
_VO2max
(mL kg-1 min-1)
Training background (event
specialism)
Duration
(weeks)
Randomized?
Running
controlled?
ST added or
replace
running?
PEDro
score
Albracht &
Arampatzis
[84]
26 (13/13)
M
I = 27,
C = 25
–
Recreational (C 3 runs wk-1,
30–120 km wk-1)
14
No
No
Added
5
Beattie et al.
[33]
20 (11/9)
M = 19
F = 1
I = 29.5,
C = 27.4
I = 59.6,
C = 63.2
Collegiate and national level
(1500 m–10 km)
40
No
No
Added
4
Berryman
et al. [80]
28 (HRT
n = 12, PT
n = 11, C
n = 5)
M
HRT = 31,
PT = 29,
C = 29
HRT = 57.5,
PT = 57.5,
C = 55.7
3–7 runs wk-1. Provincial
level (5 km–marathon)
8
Yes
Yes
Added
5
Bertuzzi et al.
[85]
22 (RTWBV
n = 8, RT
n = 8, C
n = 6)
M
RTWBV = 34,
RT = 31,
C = 33
RTWBV = 56.3,
RT = 57.4,
C = 56.1
Local 10 km (35–45 min)
race competitors
6
Yes
No
(monitored)
Added
6
Bonacci et al.
[83]
8 (3/5)
M = 5
F = 3
21.6
–
Moderately-trained triathletes
(34.8 km wk-1)
8
Yes
No
(monitored)
Added
5
Damasceno
et al. [89]
18 (9/9)
M
I = 34.1,
C = 32.9
I = 54.3,
C = 55.8
Local 10 km (35–45 min)
race competitors
8
Yes
No
(monitored)
Added
6
Ferrauti et al.
[81]
20 (11/9)
M = 14
F = 6
40.0
I = 52.0,
C = 51.1
Experienced (8.7 years)
recreational (4.6 h wk-1)
8
Yes
No
(monitored)
Added
6
Fletcher et al.
[82]
12 (6/6)
M
I = 22.2,
C = 26.3
I = 67.3,
C = 67.6
Regional/national/
international level (1500 m–
marathon)
8
Yes
No
Added
6
Giovanelli
et al. [36]
25 (13/12)
M
I = 36.3,
C = 40.3
I = 55.2,
C = 55.6
Experienced
(11.7 years,[60 km wk-1)
ultra-distance competitors
12
Yes
No
(monitored)
Added
6
Johnston
et al. [72]
12 (6/6)
F
30.3
I = 50.5,
C = 51.5
[1 year experience, 20–30
miles wk-1, 4–5 days wk-1
10
Yes
No
(monitored)
Added
6
Karsten et al.
[31]
16 (8/8)
M = 11F = 5
I = 39,
C = 30
I = 47.3,
C = 47.0
Recreational triathletes
([2 years, 3–5 days wk-1,
180–300 min wk-1)
6
Yes
No
Added
6
Mikkola et al.
[78]
25 (13/12)
M = 18
F = 7
I = 17.3,
C = 17.3
I = 62.4,
C = 61.8
High-school runners
([2 years)
8
No
No
(monitored)
Replace (I:
19%, C: 4%)
4
Millet et al.
[74]
15 (7/8)
M
I = 24.3,
C = 21.4
I = 69.7,
C = 67.6
Experienced (6.8 years)
triathletes (n = 7 national/
international)
14
Yes
No
(monitored)
Added
6
Effects of Strength Training on Distance Running
1121
123
Table 1 continued
Study
Participant characteristics
Study design
n (I/C)
Sex
Age (years)
_VO2max
(mL kg-1 min-1)
Training background (event
specialism)
Duration
(weeks)
Randomized?
Running
controlled?
ST added or
replace
running?
PEDro
score
Paavolainen
et al. [73]
18 (10/8)
M
I = 23,
C = 24
I = 63.7,
C = 65.1
Experienced (8 years) cross-
country runners
(545 h year-1)
9
Unclear
(matched on
_VO2max and
5 km)
Yes
Replace (I:
32%, C: 3%)
4
Pellegrino
et al. [91]
22 (11/11)
M = 14
F = 8
I = 34.2,
C = 32.5
I = 48.0,
C = 47.7
Experienced recreational
(local clubs and races)
6
Yes
No
Added
6
Piacentini
et al. [86]
16 (HRT
n = 6, RT
n = 5, C
n = 5)
M = 6
F = 4
HRT = 44.2
RT = 44.8
C = 43.2
–
Local ([5 years,
4–5 days wk-1) masters
runners (10 km – marathon)
6
Yes
No
Added
4
Ramı´rez-
Campillo
et al. [87]
32 (17/15)
M = 9
F = 13
22.1
–
National/international
competitive level (1500 m –
marathon)
6
Yes
No
(monitored)
Added
6
Saunders
et al. [77]
15 (7/8)
M
I = 23.4,
C = 24.9
I = 67.7,
C = 70.4
National/international
competitive level (3 km)
9
Yes
No
(monitored)
Added (but C
matched
with
stretching/
CS)
6
Schumann
et al.
[90, 92]
27 (13/14)
M
33
–
Recreational
([12 months; C 2
runs wk-1)
24
Unclear
(matched by
performance)
Yes
Added
5
Skovgaard
et al. [88]
21 (12/9)
M
31.1
59.4
Experienced (7.5 years)
recreational (29.7 km wk-1,
3.3 runs wk-1)
8
Yes
Yes (I only)
Replace (I:
42%)
6
Spurrs et al.
[75]
17 (8/9)
M
25
I = 57.6,
C = 57.8
Experienced (10 years);
60–80 km wk-1
6
Yes
No
(monitored)
Added
6
Støren et al.
[79]
17 (8/9)
M = 9
F = 8
I = 28.6,
C = 29.7
I = 61.4,
C = 56.5
Well-trained (5 km:
M = 18.42, F = 19.23)
8
Yes
No
(monitored)
Added
6
Turner et al.
[76]
18 (10/8)
M = 8
F = 10
I = 31,
C = 27
I = 50.4,
C = 54.0
Basic training
([6 months; C 3
runs wk-1)
6
Yes
No
(monitored)
Added
6
Vikmoen
et al.
[32, 38]
19 (11/8)
F
I = 31.5,
C = 34.9
53.3
Well-trained (duathletes)
11
Yes
Yes
Added
5
C control group, CS core stability, F female, h hours, HRT heavy resistance training, I intervention group, M male, PT plyometric training, RT resistance training, RTWBV resistance training with
whole body vibration, _VO2max maximal oxygen uptake, wk week
1122
R. C. Blagrove et al.
123
n = 352, female n = 96) are included, aged between 17.3
and 44.8 years. Maximal oxygen uptake data was reported
for all but five studies [83, 84, 86, 87, 90, 92] and ranged
from 47.0 to 70.4 mL kg-1 min-1. Based upon weighted
mean values in the studies that reported participant char-
acteristics for each group, age (30.2 vs. 29.0 years), body
mass (68.1 vs. 70.0 kg), height (1.74 vs. 1.74 m), and
_VO2max (57.3 vs. 57.7 mL kg-1 min-1) appeared to differ
little at baseline for ST groups and control groups respec-
tively. Moderately trained or recreational level runners
were used in nine studies [31, 72, 76, 81, 83, 84, 86,
90–92], well-trained participants in ten studies [32, 33,
36, 38, 73, 75, 79, 80, 85, 88, 89], and highly-trained or
national/international runners were used in four studies
[74, 77, 82, 87]. National caliber junior runners were also
used in one investigation [78]. Participants took part or
competed in events ranging from the middle-distances to
ultra-marathons,
and
several
studies
used
triathletes
[31, 74, 83] or duathletes [32, 38].
3.2 Study Design and PEDro Scores
Table 1 also provides an overview of several important
features of study design, including PEDro scale scores.
Studies lasted 6–14 weeks with the exception of two
investigations, which lasted 24 [90, 92] and 40 weeks [33].
Fourteen studies provided detailed accounts of the running
training undertaken by the participants. However, these
were usually reported from monitoring records, thus only
three studies were deemed to have appropriately controlled
for the volume and intensity of running in both groups
[32, 38, 73, 80, 90, 92]. Six studies provided little or no
detail on the running training that participants performed
[31, 33, 82, 84, 86, 91]. Strength training in all but three
investigations [73, 78, 88] was supplementary to running
training, and one paper provided the control group with
alternative activities (stretching and core stability) matched
for training time [77].
Studies were all scored a 4, 5, or 6 on the PEDro scale.
All investigations had points deducted for items relating to
blinding of participants, therapists, and assessors. Differ-
ences in the scores awarded were mainly the result of
studies not randomly allocating participants to groups and
failing to obtain data for more than 85% of participants
initially allocated to groups; or this information not being
explicitly stated.
3.3 Training Programs
Table 2 provides a summary of the training characteristics
associated with the ST intervention and running training
used concurrently as part of the study period. The ST
activities used were RT or HRT [31, 32, 38, 72,
78, 79, 81, 82, 84–86, 89], PT [75, 76, 80, 87, 91], ERT
[80], or a combination of these methods [33, 36, 77,
83, 90, 92], which in some cases also included SpT
[73, 74, 88].
All studies utilized at least one multi-joint, closed
kinetic chain exercise with the exception of two studies that
used isometric contractions on the ankle plantarflexors
[82, 84]. One study employed only resistance machine
exercises for lower limb HRT [81], whereas all other
studies used free weights, bodyweight resistance or a
combination of machines and free weights. Strength
training (using lower limb musculature) was scheduled
once [33, 80, 81], twice [31–33, 38, 75, 78, 85–87,
89, 90, 92], three times [36, 72, 74–77, 79, 82, 83, 88], or
four times [84] per week. One study used 15 sessions over
a 6-week period [91] and one study reported 2.7 h of ST
activity per week [73].
Heavy RT was typically prescribed in 2–6 sets of 3–10
repetitions per exercise at relatively heavy loads (higher
than 70% 1RM or to repetition failure). Plyometric training
prescription consisted of 1–6 exercises performed over 1–6
sets of 4–10 repetitions, totaling 30–228 foot contacts per
session. Most studies applied the principle of progressive
overload and some authors reported periodized models for
the intervention period [32, 33, 36, 38, 77, 88, 89]. Studies
which included SpT tended to utilize short distances
(20–150 m),
over
4–12
sets
at
maximal
intensity
[73, 74, 88]. Strength training was supervised or part-su-
pervised across all studies with the exception of three, one
that was unsupervised [76] and two where it was unclear
from the report [73, 74].
Running
training
varied
considerably
(16–170 km
week-1, 3–9 sessions week-1) across the studies, with
various levels of detail provided regarding weekly volume
and intensity. Importantly, all studies that added ST
reported that running training did not differ between
groups.
3.4 Strength Outcomes
All but two studies [31, 83] measured at least one strength-
related parameter (Table 3). Across all studies that used
1RM testing [33, 72, 74, 78, 79, 85, 86, 88–90, 92], the
intervention produced a statistically significant improve-
ment (4–33%, ES: 0.7–2.4). Maximal voluntary contraction
(MVC) was also used to assess strength capacity in seven
papers, with the majority reporting improved (7–34%, ES:
0.38–1.65) scores following ST [73, 75, 78, 81, 84] but
others reporting no difference compared to a control group
[81, 82, 90, 92]. Performance on a jump test was shown to
improve
(3–9%,
ES:
0.25–0.65)
in
some
studies
[32, 73, 74, 80, 87]; however, other studies showed no
Effects of Strength Training on Distance Running
1123
123
Table 2 Intervention and running training variables
Study
Intervention
type
Main exercises
Frequency
Volume per session
Intensity
ST
supervised?
Recovery between
sessions
Running training
Albracht &
Arampatzis
[84]
HRT
(isometric)
Ankle plantarflexion
(5 dorsiflexion,
knee extended, 40
hip flexion)
4 per week
4 sets 9 4 reps (3 s loading,
3 s relaxation)
90% MVC
(adjusted
weekly)
Yes
–
I: 66 km wk-1
C: 62 km wk-1
Beattie et al.
[33]
HRT/ERT/
PT
PT: pogo jumps,
depth jumps, CMJ
HRT: back squat,
RDL, lunge
ERT: jump squats
Wk 1–20: 2
per week;
Wk 21–40:
1 per week
9–12 sets (2–3 sets per
exercise); PT: 4–5 reps,
HRT: 3–8 reps, ERT: 3
reps
Load progressed
when
competent
Yes
C48 h between
sessions (wk
1–20). Separate
session to
running
Not reported (usual running
training)
Berryman
et al. [80]
ERT and PT
ERT: concentric
squats
PT: DJ
1 per week
ERT and PT: 3–6 sets 9 8
reps
ERT:[95% PPO
PT: 20–60 cm so
rebound[95%
CMJ
Yes
–
2 9 AIT (1 9 peak speed,
1 9 80% peak speed)
1 9 LSD (30–60 min)
Bertuzzi et al.
[85]
RT and
RTWBV
Half-squats
2 per week
3–6 sets 9 4–10 reps
periodized
70–100% 1RM
over 12 wk
Yes
Different days to
runs
57–61 km wk-1
Bonacci et al.
[83]
PT/ERT
PT: CMJ, knee lifts,
ankle jumps,
bounds, skips,
hurdle jumps
ERT: Squat jumps,
back ext.,
hamstring curls
3 per week
PT: 1–5 sets 9 5–10 reps or
20–30 m
RT: 2–5 sets 9 8–15 reps
Max height/fast
velocity
Yes
–
Same as previous 3 months. I:
swim (7.3 km), cycle
(137.6 km), run (34.8 km)
C: swim (10.1 km), cycle
(147.5 km), run (29.0 km)
Damasceno
et al. [89]
HRT
Half-squat, leg press,
calf raise, knee ext
2 per week
2–3 sets 9 3–10 reps
10RM periodized
to 3RM
Yes
72 h between
HRT sessions.
Different days to
runs
36–41 km wk-1 @50–70%
_VO2max
Ferrauti et al.
[81]
HRT
Machines: leg press,
knee ext., knee
flexion, hip ext.,
ankle ext.; UB
exercises
1 per week
LB; 1 per
week UB
LB: 4 sets 9 3–5 reps
3–5 RM
Yes
–
I: 240 min wk-1, C:
276 min wk-1
Fletcher et al.
[82]
HRT
(isometric)
Plantarflexions
3 per week
4 sets 9 20 s
80% MVC
Yes
–
70–170 km wk-1
1124
R. C. Blagrove et al.
123
Table 2 continued
Study
Intervention
type
Main exercises
Frequency
Volume per session
Intensity
ST
supervised?
Recovery between
sessions
Running training
Giovanelli
et al. [36]
CS/RT
(4wk)
HRT/ERT/
PT (8wk)
CS: 6 exercises (e.g.,
planks)
RT/HRT: single leg
half-squat, step-up,
lunges
ERT: CMJ, split
squat
PT: jump rope, high
knees
3 per week
5–8 exercises, 1–3
sets 9 6–15 reps (30 s rest)
–
Partly (only
wk 1 and
2)
C48 h between
sessions. Not
day after races/
AIT
I: normal running training
C: 70–140 km wk-1, 5–7
sessions wk-1
Johnston
et al. [72]
HRT
Squats, lunge, heel
raises (straight- and
bent-leg), knee
ext./flexion, 8xUB
exercises
3 per week
3 sets 9 6 reps squat and
lunge; 2 sets 9 20/12 reps
bent–/straight–leg heel
raise; 3 sets 9 8 reps knee
ext./flexion
RM each set
Yes
C48 h between
HRT
sessions. C 5 h
between HRT
and running
sessions.
4–5 days wk-1,
32–48 km wk-1
Karsten et al.
[31]
HRT
RDL, squat, calf
raises, lunges
2 per week
4 sets 9 4 reps
80% 1RM
Yes
C48 h between
HRT sessions.
3–5 sessions/
180–300 min wk-1
Mikkola et al.
[78]
HRT
Hamstring curl, leg
press, seated press,
squat, leg ext., heel
raise
2 per week
3–5 sets 9 3–5 reps
[90% 1RM
(reassessed
every 3 wk)
Yes
Separate session
to running
Total: I = 7 h wk-1,
C = 6.6 h wk-1;
Running: I = 48 km wk-1,
C = 44 km wk-1
Millet et al.
[74]
SpT/PT/
ERT
PT: alternative, calf,
squat, hurdle jumps
ERT: Squat, calf
raise, hurdle jump,
leg ext./curl
3 per week
(each
intervention
type once)
SpT: 5–10 sets 9 30–150 m
PT/ERT: 2–3 sets 9 6–10
reps
PT: BW
ERT: low load,
high velocity
Unclear
–
I: 8.8 h wk-1,
C: 8.5 h wk-1
Paavolainen
et al. [73]
SpT/PT/
ERT
PT: alternative, drop
and hurdle jumps,
CMJ, hops
ERT: leg press, knee
ext. and flexion
Not reported;
2.7 h per
week
SpT: 5–10 sets 9 20–100 m
PT/ERT: 5–20 reps.set-1/
30–200 reps.session-1
PT: BW or
barbell
ERT: 0–40%
1RM
Unclear
–
I: 8.4 h wk-1
(9 sessions) C: 9.2 h wk-1 (8
sessions)
Pellegrino
et al. [91]
PT
Modified version of
Spurrs et al.
(jumps, bounds,
hops)
15 sessions
total
60–228 foot contacts
Progressively
increased
Yes
–
I: 34.4–36.2 km wk-1
C: 29.5–31.3 km wk-1
Piacentini
et al. [86]
HRT and RT
Squat, calf press,
lunges, eccentric
quad, calf raise, leg
press ? UB
exercises
2 per week
HRT: 4 sets 9 3–4 reps
RT: 3 sets 9 10 reps
HRT: 85–90%
1RM
RT: 70% 1RM
Yes
–
4–5 days wk-1, 50 km wk-1
Effects of Strength Training on Distance Running
1125
123
Table 2 continued
Study
Intervention
type
Main exercises
Frequency
Volume per session
Intensity
ST
supervised?
Recovery between
sessions
Running training
Ramı´rez-
Campillo
et al. [87]
PT
DJ
2 per week
60 contacts (6 sets 9 10
reps)
20 reps @20 cm,
20 reps
@40 cm, 20
reps @60 cm
Yes
C48 h between
PT sessions.
Performed
before runs.
I: 64.7 km.wk-1
C: 70.0 km.wk-1 (AIT
preferred)
Saunders
et al. [77]
PT/HRT
PT: CMJ, ankle
jumps, bounds,
skips, hurdle
jumps, scissor
jumps
HRT: back ext., leg
press, hamstring
curls
3 per week
PT: Progress from 1 to 6
sets 9 6–10 reps/10–30 m
HRT: 1–5 sets 9 6–10 reps
(except back ext.)
PT: fast GCT
HRT: Leg press
60% 1RM
Yes
–
107 km.wk-1 (3x AIT,
1 9 LSD 60–150 min,
3 9 LSD 30–60 min,
3–6 9 LSD 20–40 min)
Schumann
et al.
[90, 92]
HRT/ERT/
PT
HRT: leg press, knee
flexion, calf raise
?UB/core
exercises
ERT: Squat jumps,
step-ups
PT: Drop jumps,
hurdle jumps
2 per week
HRT (wk 5–24): 5–12 reps
per set
HRT (wk 5–24):
60–85% 1RM
ERT: 20–30%
1RM
Yes
Same session as
running.
[48 h between
sessions
Weekly: 2x run (35–45 min/
65–85% HRmax), 2 9 LSD
(35–40 min & 70–125 min/
60–65% HRmax),
1–2 9 AIT and HIIT
Skovgaard
et al. [88]
SpT/HRT
HRT: squat, deadlift,
leg press
SpT 9 2 per
week
HRT 9 1 per
week
SpT: 4–12 sets 9 30 s
(3 min rest)
HRT: 3–4 sets 9 6–8 reps
wk 1–4; 4 sets 9 4 reps wk
5–8
SpT: maximal
effort
HRT: 15RM to
8RM wk 1–4;
4RM wk 5–8
Yes
3–4 d between
SpT/HRT
sessions.
Different days to
runs
I: AIT (4 9 4 ? 2 min @85%
HRmax); 50 min @75–85%
HRmax
C: 40 km total (4 km AIT)
Spurrs et al.
[75]
PT
Jumps, bounds, hops
2–3 per week
60–180 foot contacts
Bilateral
progressed to
unilateral and
greater height
Yes
Separate session
to running
60–80 km per week
Støren et al.
[79]
HRT
Half-squats
3 per week
4 sets 9 4 reps
4RM
Yes
–
I: 253 min wk-1 (? 119 min
other ET)
C: 154 min wk-1 (?120 min
other ET)
Turner et al.
[76]
PT
Vertical jumps and
hops (continuous
and intermittent),
split jumps, uphill
jumps
3 per week
40–110 foot contacts (5–30 s
per exercise)
Bodyweight,
short contact
time
No
(logbooks)
Performed in
running sessions
Continued regular running
(C 3 runs wk-1, C 10
miles wk-1)
1126
R. C. Blagrove et al.
123
change compared to a control group [33, 76–78, 90–92]
and in one study the control group improved to a greater
extent than the intervention group [86]. Changes in an
ability to produce force rapidly also showed mixed results,
with some studies showing improvements in peak power
output [80] and rate of force development [78, 79] and
others showing no change in these parameters [36, 75, 77].
Similarly, stiffness, when measured directly or indirectly
(using reactive strength index) during non-running tasks,
has been shown to improve (ES: 0.43–0.90) [75, 84, 86, 87]
and remain unchanged [33, 74, 89] following ST. Vertical
or leg stiffness during running showed improvements
(10%, ES: 0.33) at relatively slow speeds [36] and also at
3 km race pace (ES: 1.2) following ST [74].
3.5 Running Economy
An assessment of RE was included in all but four
[31, 85, 87, 90, 92] of the studies in this review (Table 3).
Running economy was quantified as the oxygen cost of
running at a given speed in every case, except in three
studies where a calculation of energy cost was used
[82, 84, 91]. Statistically significant improvements (2–8%,
ES: 0.14–3.22) in RE were observed for at least one speed
in 14 papers. A single measure of RE was reported in four
of these papers [31, 79, 80, 88], and a further four studies
assessed RE across multiple different speeds and found
improvements across all measures taken [72, 74, 75, 84].
Six papers reported a mixture of significant and non-sig-
nificant results from the intensities they used to evaluate
RE [36, 73, 76–78, 86]. Six studies failed to show any
significant improvements in RE compared to a control
group [32, 81–83, 89, 91].
3.6 Maximal Oxygen Uptake
No statistically significant changes were reported in
_VO2max or _VO2peak for any group in the majority of studies
that assessed this parameter [31, 32, 36, 72, 74, 75,
77–80, 85, 88, 89]. Three papers observed improvements
for _VO2max in the intervention group, but the change in
score did not differ significantly from that of the control
group [33, 81, 91]. One study detected a significant
improvement (4.9%) in
_VO2max for the control group
compared to the intervention group [73].
3.7 Velocity Associated with _VO2max
Nine studies provided data on v _VO2max or a similar metric
[31–33, 36, 74, 78, 80, 85, 89]. Just two of these papers
reported statistically significant improvements (3–4%, ES:
0.42–0.49) in the ST group compared to the control group
Table 2 continued
Study
Intervention
type
Main exercises
Frequency
Volume per session
Intensity
ST
supervised?
Recovery between
sessions
Running training
Vikmoen
et al.
[32, 38]
HRT
Machines: Half-
squats, unilateral
leg press, cable hip
flexion, calf raises
2 per week
3 sets 9 4–10 reps
(periodized 3wk cycles)
Sets performed to
RM failure
Partly (1
session
per wk
3–11)
HRT first session
or performed on
different days
4.3 sessions wk-1; 3.7 h
@60–82% HRmax, 1.1 h
@83–87% HRmax, 0.8 h
@[87% HRmax
AIT aerobic interval training, BW body weight, CMJ counter-movement jump, C control group, CS core stability, DJ drop jump, ERT explosive resistance training, ET endurance training (e.g.,
cycling, swimming, roller skiing), GCT ground contact time, h hours, HIIT high-intensity interval training, HRmax maximum heart rate (predicted from 220-age), HRT heavy resistance training,
I intervention group, LB lower body, LSD long slow distance run, MVC maximum voluntary contraction, PPO peak power output, PT plyometric training, RDL Romanian deadlift, RM
repetition maximum, RT resistance training, SpT sprint training, ST strength training, UB upper body, RTWBV resistance training with whole body vibration
Effects of Strength Training on Distance Running
1127
123
Table 3 Outcomes of the studies. Percentage changes, effect size (ES) and p value only reported for statistically significant group results or ES[0.2. All results presented are for the
intervention (I) group unless stated (e.g., C = control). Variables measured where no-significance (NS) difference for time (pre- vs. post-score) and no group 9 time (G 9 T) interaction was
detected, are also listed
Study
Main strength outcomes
Economy
_VO2max= _VO2peak
v _VO2max
Blood lactate
Time trial
Anaerobic
measures
Body composition
Albracht and
Arampatzis
[84]
Plantarflexion MVC
(6.7%, ES = 0.56,
p = 0.004), max
Achilles tendon force
(7.0%, ES = 0.55,
p\0.01), Tendon
stiffness (15.8%,
ES = 0.90, p\0.001)
[email protected] km h-1 (5.0%,
ES = 0.79)
@12.6 km h-1 (3.4%,
ES = 0.51)
[email protected] km h-1 (4.6%,
ES = 0.61)
@12.6 km h-1 (3.5%,
ES = 0.50), all p\0.05
–
–
[email protected] and
12.6 km h-1, NS
–
–
Body mass, NS
Beattie et al.
[33]
1RM back squat (wk
0–20: 19.3%,
ES = 1.2, p = 0.001)
DJRSI (wk 0–20: 7.3%,
ES = 0.3, NS G 9 T;
wk 0–40: 14.6%,
ES = 0.5, NS G 9 T)
CMJ (wk 0–20: 11.5%,
ES = 0.5, NS G 9 T;
wk 0–40: 11.5%,
ES = 0.6, NS G 9 T)
Ave. of 5 speeds
Wk 0–20: 5.0%, ES = 1.0,
p = 0.01.
Wk 0–40: 3.5%, ES = 0.6,
NS.
Wk 0–20: 0.1%,
ES = 0.1,
p = 0.013.
Wk 0-40, I:
7.4%,
ES = 0.5,
p = 0.003, C:
2.8%,
ES = 0.6, NS
Wk 0-20:
3.5%,
ES = 0.7,
NS.
Wk 0-40:
4.0%,
ES = 0.9,
NS
v2 mmol L-1,
v4 mmol L-1, NS
–
–
Body mass, fat and
lean muscle, NS
Berryman
et al. [80]
PPO (ERT: 15.4%,
ES = 0.98, p\0.01;
PT: 3.4%, ES = 0.24,
p\0.01).
CMJ (ERT: 4.5%,
ES = 0.25, p\0.01;
PT: 6.0%, ES = 0.52,
p\0.01)
@12 km h-1
ERT: 4%, ES = 0.62,
p\0.01.
PT: 7%, ES = 1.01,
p\0.01
NS
ERT: 4.2%,
ES = 0.43,
p\0.01.
PT: 4.2%,
ES = 0.49,
p\0.01
–
3 km TT
ERT: 4.1%,
ES = 0.37.
PT: 4.8%,
ES = 0.46.
C: 3.0%,
ES = 0.20;
all p\0.05,
G 9 T NS
–
Body mass, NS
Bertuzzi et al.
[85]
1RM half squat (RT:
17%, p B 0.05;
RTWBV: 18%,
p B 0.05)
–
NS
NS
–
–
–
–
Bonacci et al.
[83]
–
@12 km h-1 (after 45 min
AIT cycle) NS
–
–
–
–
Body mass, skinfolds,
thigh and calf girth,
NS
1128
R. C. Blagrove et al.
123
Table 3 continued
Study
Main strength outcomes
Economy
_VO2max= _VO2peak
v _VO2max
Blood lactate
Time trial
Anaerobic
measures
Body composition
Damasceno
et al. [89]
1RM half–squat (23%,
ES = 1.41, p\0.05),
DJRSI, wingate test NS
@12 km h-1 NS
NS
v _VO2max
(2.9%,
ES = 0.42,
p\0.05)
–
10 km TT
(2.5%,
p = 0.039),
increased
speed in
final 7 laps
(p\0.05)
30 s Wingate
test, NS
Body mass and
skinfold, NS
Ferrauti et al.
[81]
Leg extension MVC
(33.9%, ES = 1.65,
p\0.001); leg flexion
MVC (9.4%,
ES = 0.38, NS)
@LT (ES = 0.40, p\0.05,
NS G 9 T)
@8.6 and 10.1 km h-1, NS
[email protected] km h-1
(ES = 0.61, p = 0.05 G
9 T)
5.6%,
ES = 0.40,
NS G 9 T
–
[email protected] km h-1 (I:
13.1%, C: 12.1%,
NS G 9 T).
v4 mmol L-1 (I:
4.2%, C: 2.6%, NS
G 9 T).
–
–
Body mass, NS
Fletcher et al.
[82]
Isometric MVC (I:
21.6%, C: 13.4%), NS
G 9 T
EC@75,85,95% sLT, NS
–
–
BL@ 75,85,95% sLT,
NS.
–
–
–
Giovanelli
et al. [36]
SJ PPO, NS
kleg@10 km h-1, (9.5%,
ES = 0.33,
p = 0.034),
@12 km h-1 (10.1%,
ES = 0.33,
p = 0.038).
kvert
@8,10,12,14 km h-1,
NS
@8 km h-1 (6.5%,
ES = 0.43, p = 0.005),
@10 km h-1 (3.5%,
ES = 0.48, p = 0.032),
@12 km h-1 (4.0%,
ES = 0.34, p = 0.020),
@14 km h-1 (3.2%,
ES = 0.35, p = 0.022),
@RCP NS
NS
NS
–
–
–
Body mass, FFM, fat
mass, NS
Johnston
et al. [72]
1RM squat (40%,
p\0.05), knee flexion
(27%, p\0.05)
@12.8 km h-1 (4.1%,
ES = 1.76, p\0.05),
@13.8 km h-1 (3.8%,
ES = 1.61, p\0.05)
NS
–
–
–
–
Body mass, fat mass,
FFM, limb girth,
NS
Karsten et al.
[31]
–
–
NS
NS
–
5 km TT
(3.5%,
ES = 1.06,
p = 0.002)
ARD, NS
–
Effects of Strength Training on Distance Running
1129
123
Table 3 continued
Study
Main strength outcomes
Economy
_VO2max= _VO2peak
v _VO2max
Blood lactate
Time trial
Anaerobic
measures
Body composition
Mikkola et al.
[78]
MVC (8%), 1RM (4%),
RFD (31%) on leg
press; all p\0.05.
CMJ and 5–bounds, NS
@14 km h-1 (2.7%,
ES = 0.32, p\0.05),
@10,12,13 km h-1, NS
NS
NS
BL@12 km h-1
(12%, p\0.05),
@14 km h-1 (11%,
p\0.05)
–
vMART
(3.0%,
p\0.01),
v30 m sprint
(1.1%,
p\0.01)
Body mass (2%,
ES = 0.32,
p\0.01).
Thickness of QF (I:
3.9%, ES = 0.35,
p\0.01; C: 1.9%,
ES = 0.10,
p\0.05); fat %,
lean mass, NS
Millet et al.
[74]
1RM half–squat (25%,
p\0.01), 1RM heel
raise (17%, p\0.01),
hop height (3.3%,
p\0.05)
kleg@3 km pace
(ES = 1.2, p\0.05)
GCT, hop stiffness, NS
@75% v _VO2max (7.4%,
ES = 1.14, p\0.05)
@ * 92% _VO2max (5.9%,
ES = 1.15, p\0.05)
NS
2.6%,
ES = 0.57,
p\0.01,
NS G 9 T
–
–
–
Body mass, NS
Paavolainen
et al. [73]
MVC knee extension
(7.1%, p\0.01), 5BJ
(4.6%, p\0.01)
@15 km h-1 (8.1%,
ES = 3.22, p\0.001)
@13.2 km h-1, NS
_VO2@LT, NS
C: (4.9%,
p\0.05)
_VO2max
demand
(3.7%,
p\0.05, NS
G 9 T)
–
–
5 km TT
(3.1%,
p\0.05)
v20 m (3.4%,
ES = 0.77,
p\0.01)
vMART
(ES = 1.98,
p\0.001)
Body mass, fat %,
calf and thigh girth,
NS
Pellegrino
et al. [91]
CMJ (5.2%, p = 0.045,
NS G 9 T)
@10.6 km h-1 (1.3%,
p\0.05 group) NS G 9
T @7.7, 9.2, 12.1, 13.5,
15.0, 16.4 km h-1, NS.
5.2%,
ES = 0.49,
p = 0.03, NS
G 9 T
–
sLT, NS
3 km TT
(2.6%,
ES = 0.20,
p = 0.04)
–
–
Piacentini
et al. [86]
1RM leg press (HRT:
17%, ES = 0.69,
p\0.05), CMJ (C: 7%,
ES = 0.63, p\0.05),
SJ (C: 13%,
ES = 0.83, p\0.01),
Stiffness (RT: 13%,
ES = 0.64, p\0.05)
@10.75 km h-1/marathon
pace (HRT: 6.2%,
p\0.05).
@9.75,11.75 km h-1, NS
–
–
–
–
–
Body mass, fat mass,
FFM, RMR, NS
Ramı´rez-
Campillo
et al. [87]
CMJ (8.9%, ES = 0.51,
p\0.01), DJ @20 cm
(12.7%, ES = 0.43,
p\0.01), DJ @40 cm
(16.7%, ES = 0.6,
p\0.05)
–
–
–
–
2.4 km TT
(3.9%,
ES = 0.4,
p\0.05)
20 m sprint
(2.3%,
ES = 0.3,
p\0.01)
Body mass, NS
1130
R. C. Blagrove et al.
123
Table 3 continued
Study
Main strength outcomes
Economy
_VO2max= _VO2peak
v _VO2max
Blood lactate
Time trial
Anaerobic
measures
Body composition
Saunders
et al. [77]
SJ RFD and peak force,
NS.
5CMJ, NS
@18 km h-1 (4.1%,
ES = 0.35, p\0.05)
@14,16 km h-1, NS
NS
–
BL
@14,16,18 km h-1,
NS
–
–
Body mass, NS
Schumann
et al.
[90, 92]
1RM leg press (I: NS, C:
–4.7%, p = 0.011),
MVC leg flexion (–
9.7%, p = 0.031,
ES = 0.96, NS G 9
T), MVC leg press NS,
MVC knee ext. NS,
CMJ NS
–
–
–
BL during 6 9 1 km
(I: NS, C:, 21%, NS
G 9 T)
v4 mmol L-1 (I: 6%,
C: 8%, NS G 9 T).
1 km TT after
5x 1 km,
60 s rec. (I:
9%, C: 13%,
NS G 9 T)
–
Body mass, NS;
CSA vastus lateralis
(group diff. I: 7%,
C: -6%, NS G 9 T);
Total and leg lean
mass (I: 2%, NS G
9 T)
Skovgaard
et al. [88]
1RM squat (wk 4: 3.8%,
wk 8: 12%, p\0.001);
1RM leg press (wk 4:
8%, p\0.05; wk 8:
18%, p\0.001), 5RM
deadlift (wk 4: 14%,
wk8: 22%, p\0.001)
@12 km h-1 (wk 8: 3.1%,
ES = 1.53, p\0.01)
NS
–
–
10 km TT
(wk 4:
3.8%,
ES = 1.50,
p\0.05)
1500 m TT
(wk 8:
5.5%,
ES = 0.67,
p\0.001)
–
Body mass, NS
Spurrs et al.
[75]
MTS @75% MVC (left:
14.9%, right: 10.9%,
p\0.05), Calf MVC
(left: 11.4%, right:
13.6%, p\0.05).
RFD NS
@12 km h-1 (6.7%,
ES = 0.45), 14 km h-1
(6.4%, ES = 0.45),
16 km h-1 (4.1%,
ES = 0.30), all p\0.01
NS
–
–
3 km TT
(2.7%,
ES = 0.13,
p\0.05, NS
G 9 T)
–
Body mass, NS
Støren et al.
[79]
1RM (33.2%, p\0.01)
and RFD (26%,
p\0.01) half–squat
@70% _VO2max (5%,
ES = 1.03, p\0.01)
NS
–
sLT, LT % _VO2max,
NS
–
–
Body mass, NS
Turner et al.
[76]
CMJ and SJ, NS
Ave. of 3 speeds: M = 9.6,
11.3, 12.9, F = 8.0, 9.6,
11.3 km h-1 (2–3%,
p B 0.05)
@9.6 km h-1, NS
–
–
–
–
–
–
Effects of Strength Training on Distance Running
1131
123
[80, 89]. One study [74] reported a 2.6% improvement (ES:
0.57) and another [33] a 4.0% increase (ES: 0.9) after a
40-week intervention; however, these changes were not
significantly different to the control group.
3.8 Blood Lactate Parameters
Blood lactate value was measured at fixed velocities in six
studies [77, 78, 81, 82, 84, 92] and velocity assessed for
fixed concentrations of BL (2–4 mmol L-1) or lactate
threshold (LT) in six studies [32, 33, 79, 81, 90, 91]. One
study using young participants observed significantly
greater improvements (11–12%) at two speeds compared to
the control group [78]. Other studies found no significant
changes following the intervention [32, 33, 77, 79,
82, 84, 91] or a change which was not superior to the
control group [81, 90, 92].
3.9 Time-Trial Performance
To assess the impact of ST directly upon distance running
performance, studies utilized time trials over 1000 m
(preceded by 5 9 1 km) [90, 92], 1500 m [88], 2.4 km
[87], 3 km [75, 80, 91], 5 km [31, 73], 10 km [88, 89],
5 min [32], and 40 min [38]. There were similarities to
competitive scenarios in most studies, including perfor-
mances taking place under race conditions [31, 75,
87, 90–92], on an outdoor athletics track [31, 87–89], on
an indoor athletics track [73, 75, 80, 90–92], and fol-
lowing a prolonged (90-min) submaximal run [38]. Per-
formance
improvements
were
statistically
significant
compared to a control group for eight of the 12 trials. The
exceptions were a 40-min time trial [38], a 1000-m rep-
etition [90, 92], and two studies that used a 3 km time trial
[75, 80]. Statistically significant 3 km improvements were
observed for all groups in one case [80]; however, the ES
was larger for the two intervention groups (0.37 and 0.46)
compared to the control group (0.20). Improvements over
middle-distances (1500–3000 m) were generally moderate
(3–5%, ES: 0.4–1.0). Moderate to large effects (ES:[1.0)
were observed for two studies [31, 88] that evaluated
performance over longer distances (5–10 km); however,
the relative improvements were quite similar (2–4%) over
long
distances
compared
to
shorter
distances
[31, 73, 88, 89].
3.10 Anaerobic Outcomes
Tests relating to anaerobic determinants of distance run-
ning performance were used in five investigations. Sprint
speed over 20 m [73, 87] and 30 m [78] showed statis-
tically significant improvements following ST (1.1–3.4%).
Two
studies
provided
evidence
for
enhancement
of
Table 3 continued
Study
Main strength outcomes
Economy
_VO2max= _VO2peak
v _VO2max
Blood lactate
Time trial
Anaerobic
measures
Body composition
Vikmoen
et al.
[32, 38]
1RM half–squat (45%,
ES = 2.4, p\0.01), SJ
(8.9%, ES = 0.83,
p\0.05), CMJ (5.9%,
ES = 0.65, p\0.05)
@10 km h-1, NS
NS
NS
v3.5 mmol L-1, NS
5 min TT
(4.7%,
ES = 0.95,
p\0.05).
40 min TT,
NS
I: Leg mass (3.1%,
ES = 1.69,
p=p\0.05), body
mass, NS
C: Leg mass (-2.2%),
body mass decrease
(-1.2%, p\0.05)
ARD anaerobic running distance, BJ broad jump, BL blood lactate, CMJ counter-movement jump, C control group, DJ drop jump, DJRSI drop jump reactive strength index, EC energy cost,
EMG electromyography, ERT explosive resistance training, FFM fat-free mass, FU fractional utilization, GCT ground contact time, GRF ground reaction force, HR heart rate, HRT heavy
resistance training, I intervention group, kleg leg stiffness, kvert vertical stiffness, (s)LT (speed at) lactate threshold, MAS maximal aerobic speed, MTS musculotendinous stiffness, MVC
maximum voluntary contraction, PPO peak power output, PT plyometric training, QF quadriceps femoris, RCP respiratory compensation point (VE/VCO2), RFD rate of force development, RM
repetition maximum, RMR resting metabolic rate, RT resistance training, RTWBV resistance training with whole body vibration, SJ squat jump, TT time trial, TTE time to exhaustion, v velocity,
vMART velocity during maximal anaerobic running test, _VO2 oxygen uptake, _VO2max= _VO2peak highest oxygen uptake associated with a maximal aerobic exercise test, v _VO2max velocity
associated with _VO2max, wk week
1132
R. C. Blagrove et al.
123
vMART [73, 78], and one further study showed no
change in anaerobic running distance after 6 weeks of
HRT [31]. A 30-s Wingate test was also used in one
paper; however, no differences in performance were noted
[89].
3.11 Body Composition
Body mass did not change from baseline in 18 of the
studies [32, 33, 36, 38, 72–75, 77, 79–81, 83, 84, 86–89];
however, one investigation reported a significant increase
(2%, ES: 0.32) following ST [78]. This study also docu-
mented changes in the thickness of quadriceps femoris
muscle in both the intervention (3.9%, ES: 0.35) and
control group (1.9%, ES: 0.10) [78]. Similarly, an increase
in total lean mass (3%) and leg lean mass (3%) was found
following 12 weeks of ST despite little alteration in cross-
sectional area of the vastus lateralis and body mass being
noted [90, 92]. Another study observed a significant
decrease (- 1.2%) in body mass in the control group, with
no change in the intervention group [32]. A significant
increase in leg mass (3.1%, ES: 1.69) was also noted in this
study [32, 38]. Other indices of body composition that
exhibited
no
significant
changes
were:
fat
mass
[33, 36, 72, 73, 78, 86], fat-free mass [36, 72, 86], lean
muscle mass [33, 78], skinfolds [83, 89], and limb girth
measurements [72, 73, 83].
4 Discussion
The aim of this systematic review was to identify and
evaluate current literature which investigated the effects of
ST exercise on the physiological determinants of middle-
and long-distance running performance. The addition of
new research published in this area, and the application of
more liberal criteria provided results for 50% more par-
ticipants (n = 469) compared to a recent review on RE
[10]. Based upon the data presented herein, it appears that
ST activities can positively affect performance directly and
provide benefits to several physiological parameters that
are important for distance running. However, inconsisten-
cies exist within the literature, that can be attributed to
differences in methodologies and characteristics of study
participants, thus practitioners should be cautious when
applying generalized recommendations to their athletes.
Despite the moderate PEDro scores (4, 5, or 6), the quality
of the works reviewed in this paper are generally consid-
ered acceptable when the unavoidable constraints imposed
by a training intervention study (related to blinding) are
taken into account.
4.1 Running Economy
Running economy, defined as the oxygen or energy cost to
run at a given sub-maximal velocity, is influenced by a
variety of factors, including force-related and stretch–
shortening cycle qualities, which can be improved with ST
activities.
In
general,
an
ST
intervention,
lasting
6–20 weeks, added to the training program of a distance
runner appears to enhance RE by 2–8%. This finding is in
agreement with previous meta-analytical reviews in this
area that show concurrent training has a beneficial effect
(* 4%) on RE [10, 26]. In real terms, an improvement in
RE of this magnitude should theoretically allow a runner to
operate at a lower relative intensity and thus improve
training and/or race performance. No studies attempted to
demonstrate this link directly, although inferences were
made in studies, which noted improvements in RE and
performance separately [73, 80, 88]. Other works provide
evidence that small alterations in RE (* 1.1%) directly
translate to changes (* 0.8%) in sub-maximal [94] and
maximal running performance [95]. The typical error of
measurement of RE has been reported to be 1–2% [96–99]
and the smallest worthwhile change * 2% [94, 98, 100],
which is thought to represent a ‘‘real’’ improvement and
not simply a change due to variability of the measure.
Taken together, it is therefore likely that the improvements
seen in RE following a period of concurrent training would
represent a meaningful change in performance.
Improvements were observed in moderately-trained
[72, 76, 84, 86], well-trained [33, 36, 73, 75, 79, 80, 88]
and highly-trained participants [74, 77], suggesting runners
of any training status can benefit from ST. Different modes
of ST were utilized in the studies, with RT or HRT
[72, 78, 79, 84, 86], ERT [80], PT [75, 76, 80], and a
combination of these activities [33, 36, 77], all augmenting
RE to a similar extent. Single-joint isometric RT may also
provide a benefit if performed at a high frequency (4
day week-1) [84]. Several studies adopted a periodized
approach to the types of ST prioritized during each 3- to
6-week cycle [33, 36, 77, 88], which is likely to provide the
best strategy to optimize gains long-term [101].
Six studies [32, 81–83, 89, 91] failed to show any
improvement in RE and a further six [36, 73, 76–78, 86]
observed both improvements and an absence of change at
various velocities. This implies benefits are more likely to
occur under specific conditions relating to the choice of
exercises, participant characteristics, and velocity used to
measure RE. In most studies that observed a benefit,
exercises
with
free
weights
were
utilized
[33, 36, 72, 74, 86, 88]. Multi-joint exercises using free
weights are likely to provide a superior neuromuscular
stimulus compared to machine-based or single-joint exer-
cises as they demand greater levels of co-ordination, multi-
Effects of Strength Training on Distance Running
1133
123
planar control, activation of synergistic muscle groups
[102, 103] and usually require force to be produced from
closed-kinetic chain positions. These types of exercise also
have a greater biomechanical similarity to the running
action so are therefore likely to provide a greater level of
specificity and hence transfer of training effect [104]. An
insufficient overload or a lack of movement pattern
specificity may therefore be the reason for the absence of
an effect in studies that used only resistance machines
[32, 81] or a single-joint exercise [82]. These studies were
also characterized by a lower frequency of sessions com-
pared to studies that used similar RT exercises but did
observe an improvement in RE [78, 84].
Moderately-trained runners were used in three of the six
studies showing an absence of effect [81, 83, 91] and one
used triathletes who performed a relatively low volume of
running (34.8 km week-1) as part of their training [83].
However, a similar number of studies who used recre-
ational athletes did show a positive effect [72, 76, 84, 86],
suggesting that training level is unlikely to be the reason
for the lack of response in these studies. This is also con-
firmed by recent observations that showed improvement in
RE following a period of concurrent training was similar
across individuals irrespective of training status and the
number of sessions per week ST was performed [10].
The velocity used to assess RE may also explain the
discrepancies in results across studies. It has been sug-
gested that runners are most economical at the speeds they
practice at most [98], and for investigations that utilized
PT, stretch–shortening cycle improvements are likely to
manifest at high running speeds where elastic mechanisms
have greatest contribution [83, 105]. Therefore a velocity-
specific measurement of RE may be the most valid strategy
to establish whether an improvement has occurred. For
example,
Saunders
and
associates
[77]
observed
an
improvement (p = 0.02, ES: 0.35) at 18 km h-1 in elite
runners, but an absence of change at slower speeds. Sim-
ilarly, Millet and colleagues [74] noted large (ES:[1.1)
improvements
at
speeds
faster
than
75%
v _VO2max
(* 15 km h-1) in highly-trained triathletes, and Paavo-
lainen et al. [73] detected changes at 15 km h-1 but not
slower speeds in well-trained runners. Furthermore, Pia-
centini and co-workers [86] found improvement at race-
pace in recreational marathon runners but not at a slower
and a faster velocity. Improvements observed at faster
compared to slower speeds may also reflect improvements
in motor unit recruitment as a consequence of ST. As
running speed increases there is a requirement for greater
peak vertical forces due to shorter ground contact times,
which elevates metabolic cost [25]. To produce higher
forces, yet overcome a reduction in force per motor unit as
a consequence of a faster shortening velocity, more motor
unit recruitment is required [106]. Thus, an increase in
absolute motor unit recruitment following a period of ST
would result in a lower relative intensity reducing the
necessity to recruit higher threshold motor units during
running [25]. Several studies that failed to show any
response used a single velocity to assess RE [32, 83, 89],
perhaps indicating that the velocity selected was unsuit-
able to capture an improvement. Furthermore, only a small
number of studies used relative speeds [33, 74, 79, 81, 82],
with most choosing to assess participants at the same
absolute intensity. A given speed for one runner may rep-
resent a high relative intensity, whereas for another runner
it may be a relatively low intensity. Therefore selecting the
same absolute speed in a group heterogeneous with respect
to _VO2max, may not provide a true reflection of any changes
which take place following an intervention. Moreover, this
may also confound any potential improvements observed
in fractional utilization of _VO2max.
Several common procedural issues exist in the studies
reviewed, which may influence the interpretation of results
and therefore conclusions drawn. The majority of studies
quantified RE and _VO2max as a ratio to body mass; how-
ever, oxygen uptake does not show a linear relationship
with increasing body size [107]. It is also known that the
relationship between body size and metabolic response
varies across intensities, with a trend for an increasing size
exponent as individuals move from low-intensity towards
maximal exercise [108, 109]. Moreover, allometric scaling
is likely to decrease interindividual variability [110],
potentially improving the reliability of observations [99].
Ratio-scaling RE for all velocities to body mass is therefore
theoretically and statistically inappropriate [111]. Just two
studies [79, 80] used an appropriate allometric scaling
exponent (0.75) to account for the non-linearity associated
with oxygen uptake response to differences in body mass,
both establishing a large ES in their results. The unsuit-
ability of ratio-scaling as a normalization technique when
processing physiological data is likely to have influenced
the statistical outcomes of some studies and thus inaccurate
conclusions may have been generated.
Running economy was expressed as oxygen cost in all
but three studies [82, 84, 91], which quantified RE using
the energy cost method. As the energy yield from the
oxidation of carbohydrates and lipids differs, subtle alter-
ations in substrate utilization during exercise can confound
measurement of RE when expressed simply as an oxygen
uptake value. Energy cost is therefore the more valid
[112, 113] and reliable [99] metric for expressing econ-
omy, compared to traditional oxygen cost, as metabolic
energy expenditure can be calculated using the respiratory
exchange ratio, thus accounting for differences in substrate
utilization. Despite attempts to control for confounding
1134
R. C. Blagrove et al.
123
variables such as diet and lifestyle in most studies, equiv-
alence in inter-trial substrate utilization cannot be guaran-
teed, which may have impacted upon the measurement of
RE.
4.2 Maximal Oxygen Uptake
Maximal oxygen uptake is widely regarded as one of the
most important factors in distance running success [114],
therefore the objective for any distance runner is to maxi-
mize their aerobic power [9]. An individual’s _VO2max is
limited by their ability to uptake, transport and utilize
oxygen in the mitochondria of working muscles. Endur-
ance training involving prolonged continuous bouts of
exercise or high intensity interval training induces adap-
tations primarily within the cardiovascular and metabolic
systems that results in improvements in _VO2max [9, 115].
Conversely, ST is associated with a hypertrophy response
that increases body mass and has been reported to decrease
capillary density, oxidative enzymes and mitochondrial
density [116–118], which would adversely impact aerobic
performance. Theoretically there is therefore little basis for
ST as a strategy to enhance aerobic power. However it is
important to address whether in fact _VO2max is negatively
affected when distance running is performed concurrently
with ST.
Thirteen works in this review found no change in
_VO2max following the intervention period, demonstrating
that although ST does not appear to positively influence
_VO2max, it also does not hinder aerobic power. Although
ST in most studies was supplementary to running training,
it appears that the additional physiological stimulus pro-
vided by ST was insufficient to elicit changes in cardio-
vascular-related
parameters
[119].
Three
studies
did
observe significant increases in aerobic power that did not
differ to the change observed in the control group [33, 81,
91], and one further study found an improvement in
_VO2max in the control group only [78]. It is perhaps sur-
prising that more studies did not find an increase in _VO2max
(in any group) given that participants continued their nor-
mal running training through the study period. Improve-
ments in _VO2max of 5–10% have been shown following
relatively short periods (\6 weeks) of endurance training
[9]; however, the magnitude of changes is dependent upon
a variety of factors including the initial fitness level of
individuals and the duration and nature of the training
programoo [120]. Maximal oxygen uptake is known to
have an innate upper limit for each individual, therefore in
highly-trained and elite runners, long-term performance
improvement is likely to result from enhancement of other
physiological
determinants,
such
as
RE,
fractional
utilization and v _VO2max [4, 121, 122]. A number of studies
used moderately-trained participants [23, 72, 76, 81, 91],
who would be the most likely to show an improvement in
_VO2max following a 6- to 14-week period of running, with
two investigations demonstrating improvements for both
groups [81, 91]. The absence of _VO2max improvement in
other papers suggests that the duration of the study and/or
the training stimulus, was insufficient to generate an
improvement [120]. Indeed, one study of 40 weeks’ dura-
tion in Collegiate level runners observed similar improve-
ments (ES: 0.5–0.6) in
_VO2max in both groups [33],
suggesting a longer time period may be required to detect
changes in runners with a higher training status. High-in-
tensity aerobic training ([80% _VO2max) is a potent stim-
ulus for driving changes in _VO2max[123]; however, some
studies reported runners predominantly utilized low-inten-
sity (\70%
_VO2max) continuous running [74, 78, 89],
which may also explain the lack of changes observed.
4.3 Velocity Associated with _VO2max
An individual’s v _VO2max is influenced by their _VO2max, RE
and anaerobic factors including neuromuscular capacity [4,
124]. The amalgamation of several physiological qualities
into this single determinant appears to more accurately
differentiate performance, particularly in well-trained run-
ners [3, 98, 125, 126], therefore v _VO2max has been labelled
as an important endurance-specific measure of muscular
power [127].
Improvements for v _VO2max (3–4%, ES: 0.42–0.49) were
found in two investigations [80, 89], with a further two
studies observing improvements (2.6–4.0%, ES: 0.57–0.9)
that could not be ascribed to the training differences
between the groups [33, 74]. A number of studies also
found little change in v _VO2max following an intervention
[31, 32, 36, 78, 85]. As v _VO2max is the product of the
interaction between aerobic and anaerobic variables, a
small improvement in one area of physiology may not
necessarily result in an increase in v _VO2max. Damasceno
et al. [89] found an improvement in v _VO2max (2.9%,
p\0.05, ES: 0.42) despite detecting no change in _VO2max,
RE or Wingate performance, therefore attributed the
change to the large improvements (23%, ES: 1.41) in the
force-producing ability they observed in participants.
Conversely, Berryman and associates [80] found changes
in v _VO2max (4.2%, ES: 0.43–0.49) alongside improvements
in RE (4–7%, ES: 1.01), moderate increases in power
output, and no change in _VO2max scores. Beattie and co-
workers [33] credited the change in v _VO2max they observed
(20-weeks:
3.5%,
ES:
0.7)
to
the
accumulation
of
Effects of Strength Training on Distance Running
1135
123
improvements in RE, _VO2max and anaerobic factors; how-
ever, these were not sufficiently large enough to provide a
significant group 9 time interaction. Millet and colleagues
[74] found notable improvements in RE (7.4%, ES: 1.14);
however, changes in RE could not explain the changes
observed in v _VO2max (r = - 0.46, p = 0.09). It may also
be the case that longer periods of ST are required before an
improvement in v _VO2max is detected, as studies showing an
improvement (2.6–4.0%, ES: 0.57–0.9) from baseline las-
ted 14 weeks or more [33, 74], and studies showing little
change tended to be 6–8 weeks in duration [31, 78, 85].
The conflicting results could also be explained by the
inconsistency in methods used to define v _VO2max. A
number of different protocols and predictive methods have
been suggested to assess v _VO2max [4], including determi-
nation from the _VO2-velocity relationship [128] and the
peak running speed attained during a maximal test using
speed increments to achieve exhaustion [21, 127]. All
studies that measured v _VO2max in this review did so via an
incremental run to exhaustion progressed using velocity.
Velocity at _VO2max was taken as the highest speed that
could be maintained for a full 60-s stage [78, 80, 85], an
average of the final 30-s [31, 36], the mean velocity in the
final 120-s [32], or the minimum velocity that elicited
_VO2max [33, 74]. Although a direct approach to the mea-
surement of v _VO2max has been recommended [4], due to
the velocity increments (0.5–1.0 km h-1) used in these
investigations, this may not provide sufficient sensitivity to
detect a change following a short- to medium-term inter-
vention.
Damasceno
and
associates
[89]
calculated
v _VO2max using a more precise method based upon the
fractional time participants reached through the final stage
of the test multiplied by the increment rate. This perhaps
provided a greater level of accuracy which allowed the
authors to identify the differences in changes which existed
between the groups. Taken together, there is weak evidence
that v _VO2max can be improved following an ST interven-
tion, despite constituent physiological qualities often
exhibiting change. Differences in the protocols used to
determine v _VO2max makes comparison problematic; how-
ever, a more precise measurement of v _VO2max
that
accounts for partial completion of a final stage is likely to
provide the sensitivity to identify subtle changes that may
occur.
The critical velocity model, which represents exercise
tolerance in the severe intensity domain, potentially offers
an alternative to measurement of v _VO2max that is currently
uninvestigated in runners [35, 129]. Two main parameters
can be assessed using the critical velocity model; critical
velocity itself, which is defined as the lower boundary of
the severe intensity domain which when maintained to
exhaustion leads to attainment of _VO2max, and the curva-
ture constant of the velocity–time hyperbola above critical
velocity, which is represented by the total distance that can
be covered prior to exhaustion at a constant velocity [130].
Middle-distance running performance (800 m) is strongly
related to critical velocity models (r = 0.83–0.94) in
trained runners [131], and may be more important than RE
in well-trained runners [35]. Evidence from studies using
untrained participants has demonstrated that the total
amount of work that can be performed above critical power
during
high-intensity
cycling
exercise
is
improved
(35–60%) following 6–8 weeks of RT [132, 133]. Future
investigations should therefore address the dearth in liter-
ature around how ST might positively influence parameters
related to the critical velocity model [35].
4.4 Blood Lactate Markers
A runner’s velocity at a reference point on the lactate-
velocity curve (e.g., LT) or BL for a given running speed
are important predictors of distance running performance
[134–136]. A runners LT also corresponds to the fractional
utilization of _VO2max that can be sustained for a given
distance [114], therefore an increase in LT also allows a
greater proportion of aerobic capacity to be accessed.
In contrast to RE, ST appears to have little impact upon
BL markers. This is quite surprising as an improvement in
RE should theoretically result in an enhancement in speed
for a fixed BL concentration. This suggests that adaptations
to RE can occur independently to changes in metabolic
markers of performance. An absence of change in BL also
implies that ST does not alter anaerobic energy contribu-
tion during running, thus assuming aerobic energy cost of
running is reduced following ST, it can be inferred that
total energy cost (aerobic plus anaerobic energy) is also
likely to be reduced. Previous studies have shown as little
as 6 weeks of endurance training can improve BL levels or
the velocity corresponding to an arbitrary BL value in
runners [137–139]. The intensity of training is important to
elicit improvement in BL parameters [140], therefore it
appears that the running training prescription may have
been insufficient to stimulate improvements, or the training
status of participants meant a longer period was required to
realize a meaningful change. In addition, the inter-session
reliability of BL measurement between 2–4 mmol L-1
is * 0.2 mmol L-1 [99], therefore over a short study
duration this metric may not provide sufficient sensitivity
to detect change.
Training at an intensity above the LT is likely to result
in a reduction in the rate of BL production (and therefore
accumulation), or an improved lactate clearance ability
from the blood [9]. Short duration high-intensity bouts of
1136
R. C. Blagrove et al.
123
activity generate high levels of BL so drive metabolic
adaptations which can result in an improvement in per-
formance [141–143]. Studies that have utilized high-repe-
tition, low-load RT in endurance athletes therefore have the
potential to produce high BL concentrations so may pro-
vide an additional stimulus to improve performance via BL
parameters. This theory is supported by works that have
demonstrated improvements in BL-related variables in
endurance athletes following an intervention that uses a
strength-endurance style of conditioning with limited rest
between sets [54, 62, 144]. The ST prescription in the
studies reviewed was predominantly low-repetition, high-
intensity RT or PT, which is unlikely to have provided a
metabolic environment sufficient to directly enhance
adaptations related to BL markers.
4.5 Time-Trial Performance
Physiological parameters such as _VO2max, v _VO2max, RE
and LT are clearly important determinants that can be
quantified in a laboratory; however, for a runner, TT per-
formance possesses a far higher degree of external validity.
Similar improvements in TT performance were observed
for middle-distance events (3–5%, ES: 0.4–1.0) and long-
distance events up to 10 km (2–4%, ES: 1.06–1.5). In the
majority of these studies, time trials took place in a similar
environment and under comparable conditions to a race,
therefore these findings have genuine applicability to ‘‘real-
life’’ scenarios. These improvements are likely to be a
consequence of significant enhancements in one or more
determinants of performance. Interestingly, Damasceno
and co-authors [89] found an improvement in 10 km TT
performance due to the attainment of higher speeds in the
final 3 km, despite observing no change in RE during a
separate assessment. This suggests that greater levels of
muscular strength may result in lower levels of relative
force production per stride, thereby delaying recruitment of
higher threshold muscle fibers and thus providing a fatigue
resistant effect [145]. This subsequently manifests in a
superior performance during the latter stages of long-dis-
tance events [89].
Four studies observed no difference in performance
change compared to a control group [38, 75, 80, 90, 92].
Vikmoen and colleagues [38] attributed a lack of effect in
their 40 min TT to the slow running velocity caused by the
5.3% treadmill inclination used in the test. This was also
the only study to use a treadmill set to a pre-determined
velocity which participants could control once the test had
commenced. The absence of natural self-pacing may
therefore have prevented participants achieving their true
potential on the test. Spurrs et al. [75] and Berryman et al.
[80] both found improvements in 3 km performance
compared to a pre-training measure of a comparable
magnitude to other studies (2.7–4.8%, ES: 0.13–0.46);
however, changes were not significantly different to a
control group, suggesting ST provided no additional benefit
or there was a practice effect associated with the test.
It could be possible that enhancement of physiological
qualities in some studies could be attributed to RT being
positioned immediately after low-intensity, non-depleting
running sessions [146]. This arrangement of activities in
concurrent training programs has been shown to provide a
superior stimulus for endurance adaptation compared to
performing separate sessions, and without compromising
the signaling response regulating strength gains [147, 148].
This, however, appears not to be the case, as most studies
reported ST activities took place on different days to run-
ning sessions [85, 88, 89] or were at least performed as
separate sessions within the same day [33, 36, 38, 72, 75,
78]. Only three studies performed ST and running imme-
diately after one another, with one positioning PT before
running [87] and one lacking clarity on sequencing [76].
Schumann and colleagues [90, 92] observed no additional
benefit to both strength and endurance outcomes compared
to a running only group, when ST was performed imme-
diately following an incremental running session (65–85%
maximal heart rate), citing residual fatigue which com-
promised quality of ST sessions as the reason.
4.6 Anaerobic Running Performance
The contribution of anaerobic factors to distance running
performance is well established [127, 149]. In particular,
anaerobic capacity and neuromuscular capabilities are
thought to play a large role in discriminating performance
in runners who are closely matched from an aerobic per-
spective [124, 150]. An individual’s v _VO2max perhaps
provides the most functional representation of neuromus-
cular power in distance runners; however, measures of
maximal running velocity and anaerobic capacity are also
potentially important [127].
Tests for pure maximal sprinting velocity (20–30 m)
were used in three studies [73, 78, 87] and showed
improvements (1.1–3.4%) following ST in every case. This
confirms results from previous studies that have shown
sprinting performance can be positively affected by an ST
intervention in shorter-distance specialists [151–153]. This
finding has important implications for distance runners, as
competitive events often involve mid-race surges and
outcomes are frequently determined in sprint-finishes,
particularly at an elite level [154–157]. Middle-distance
runners also benefit from an ability to produce fast running
speeds at the start of races [158], therefore improving
maximum speed allows for a greater ‘‘anaerobic speed
Effects of Strength Training on Distance Running
1137
123
reserve’’ [159], resulting in a lower relative work-rate, and
thus decreasing anaerobic energy contribution [41]. Inter-
estingly, endurance training in cyclists has been shown to
improve critical power [160] but reduce work capacity for
short duration exercise [161, 162]. It is unknown whether
long-term aerobic training has a similar effect on anaerobic
running qualities; however, ST offers a strategy to avoid
this potential negative consequence.
The velocity attained during a maximal anaerobic run-
ning test provides an indirect measure of anaerobic and
neuromuscular performance, and has a strong relationship
(r = 0.85) to v _VO2max [19]. The vMART is particularly
relevant to middle-distance runners because it requires
athletes to produce fast running speeds under high-levels of
fatigue caused by the acidosis and metabolites derived
from glycolysis [163]. Both studies that included this test
observed significant improvements in vMART (1.1–3.4%),
which can be attributed to changes observed in neuro-
muscular power as a result of the ST intervention [73, 78].
One study showed no alteration in the predicted distance
achieved on an anaerobic running test following 6 weeks
of HRT; however, the validity and reliability of the test was
questioned by the authors [31]. Performance on a 30 s
Wingate test was also unchanged following 8 weeks of
running training combined with HRT in recreational par-
ticipants [89]. This finding perhaps underlines the impor-
tance of selecting tests which are specific to the training
which has been performed in the investigation.
4.7 Strength Outcomes
Changes in strength outcomes were evident in most studies
despite all but one [78] observing no change in body mass.
Since strength changes can be ascribed to both neurological
and morphological adaptations [164], it is therefore likely
that improvements are primarily underpinned by alterations
in intra- and inter-muscular co-ordination. It is also known
that initial gains in strength in non-strength trained indi-
viduals are the consequence of neural adaptations rather
than structural changes [118]. An improvement in force
producing capability is perhaps expected in individuals
who have little or no strength-training experience [165];
however, concurrent regimens of training have consistently
been shown to attenuate strength-related adaptation [30].
The seminal paper published by Hickson et al. [48] was
the first to identify the potential for endurance exercise to
mitigate strength gains, when both training modalities were
performed concurrently within the same program. Follow-
up
investigations
have
since
shown
mixed
results
[166–171], but evidence from this review clearly demon-
strates that, for the distance runner at least, strength-related
improvements are certainly possible following a concurrent
period of training. Nevertheless, the study designs adopted
by the works under review did not include a strength-only
training group, thus it is not possible to determine whether
strength adaptation was in fact negated under a concurrent
regimen. One study using well-trained endurance cyclists
with no ST experience, observed a blunted strength
response in a group who added ST to their endurance
training compared to a group who only performed ST
[170]. Based upon this finding and other similar observa-
tions [167, 172, 173] it seems likely that although distance
runners can significantly improve their strength using a
concurrent approach to training, strength outcomes are
unlikely to be maximized. Moreover, the degree of inter-
ference with strength-adaptation also appears to be exac-
erbated when volumes of endurance training are increased
and the duration of concurrent training programs is longer
[30, 146].
4.8 Body Composition
Resistance training performed 2–3 times per week is
associated with increases in muscle cross-sectional area as
a principal adaptation [174]. Although gains in gross body
mass may appear to be an unfavorable outcome for dis-
tance runners, the addition of muscle mass to proximal
regions of the lower limb (i.e., gluteal muscles) should
theoretically provide an advantage, via increases in hip
extension forces, minimizing moment of inertia of the
swinging limb, and reducing absolute energy usage [25]. It
is somewhat surprising that virtually all studies demon-
strated an absence of change in body mass, fat-free mass,
lean muscle mass, and limb girths. Other than one inves-
tigation [33], the duration of the studies that observed no
effect on measures of body composition was\14 weeks,
suggesting this may not have been sufficiently long to
demonstrate a clear hypertrophic response. There is also a
possibility that small increases in muscle mass within
specific muscle groups (e.g., gluteals) were present, and
contributed to the improvements observed in RE, but these
may not have been detectable using a gross measure of
mass. Evidence for this may have occurred in the Schu-
mann et al. study [90, 92], who observed increases in total
lean mass (3%) despite noting no significant change in
body mass or cross-sectional area of the vastus lateralis
compared to baseline measures.
The interference effect observed during concomitant
integration of endurance and ST as part of the same pro-
gram may also provide an explanation for the lack of
change in measures of mass. Following a bout of exercise,
a number of primary and secondary signaling messengers
are up regulated for 3–12 h [175], which initiate a series of
molecular events that serve to activate or suppress specific
1138
R. C. Blagrove et al.
123
genes. The signaling messengers which are activated, relate
to the specific stress which is imposed on the physiological
systems involved in an exercise bout. Strength training
causes mechanical perturbation to the muscle cell, which
elicits a multitude of signaling pathways that lead to a
hypertrophic response [176]. In particular, the secretion of
insulin-like growth factor-1 as a result of intense muscular
contraction is likely to cause a cascade of signaling events
which increase activity of phosphoinositide-3-dependent
kinase (Pl-3 k) and the mammalian target of Rapamycin
(mTOR) [177–179]. There is strong evidence that mTOR is
responsible for mediating skeletal muscle hypertrophy via
activation of ribosome proteins which up regulate protein
synthesis [180]. Prolonged exercise bouts, such as those
associated with endurance training, activate metabolic
signals related to energy depletion, uptake and release of
calcium ions from the sarcoplasmic reticulum and oxida-
tive stress in cells [181]. Adenosine monophosphate acti-
vated kinase (AMPK) is a potent secondary messenger
which functions to monitor energy homeostasis [182] and
when activated, modulates the release of peroxisome pro-
liferator
co-activator-1a,
which
along
with
calcium-
calmodulin-dependent
kinases
increase
mitochondrial
function to enhance aerobic function [181, 183, 184].
Crucially though, AMPK also acts to inhibit the Pl-3 k/
mTOR stage of the pathway via activation of the tuberous
sclerosis complex thereby suppressing the ST induced up
regulation of protein synthesis [185, 186]. This conflict
arising at a molecular signaling level therefore appears to
impair the muscle fiber hypertrophy response to ST and
attenuate increases in body mass [186].
4.9 Muscle–Tendon Interaction Mechanisms
The
potential
mechanisms
for
the
positive
changes
observed in physiological parameters underpinning running
performance were directly investigated in three studies [82,
84, 91], and were inferred from gait measures [36, 73–75,
77] and strength outcomes in others. It is well documented
that muscle–tendon unit stiffness correlates well with RE
[187–189]. Tendons are also highly adaptable to mechan-
ical loading and have been shown to increase in stiffness in
response to HRT and PT [84, 190, 191]. Despite observing
no statistical effect for HRT on RE, Fletcher and col-
leagues [82] also found a relationship between the change
in RE and the changes observed in Achilles tendon stiff-
ness. Despite these associations, it is likely that improve-
ments in RE are a consequence of the interaction between
adaptations to tendon properties and improvements in
motor unit activation which influence behavior of force–
length-velocity properties of muscles [25]. It tends to be
assumed that improved tendon stiffness allows the body to
store and return elastic energy more effectively, which
results in a reduction in muscle energy cost due to a greater
contribution from the elastic recoil properties of tendons
[192]. Indeed, authors of studies in the present review have
argued that the improvements observed in RE following a
period of ST are due to an enhanced utilization of elastic
energy during running [36, 73–75]. An alternative pro-
posal, based upon more recent evidence, suggests the
Achilles tendon provides a very small contribution to the
total energy cost of running therefore improvements in
stiffness provide a negligible reduction in energy cost [193,
194]. Instead, a tendon with an optimal stiffness con-
tributes to reducing RE by minimizing the magnitude and
velocity of muscle shortening, thus allowing muscle fas-
cicles to optimize their length and remain closer to an
isometric state [25]. A reduction in the amount and velocity
of fiber shortening therefore reduces the level of muscle
activation required and hence the energy cost of running
[193].
The improvements observed in maximal and explosive
strength, which can be attributed to increases in motor unit
recruitment and firing frequency, enable the lower limb to
resist eccentric forces during the early part of ground
contact [165] and thus contribute to the attainment of a
near isometric state during stance. As the force required to
sustain speed during distance running performance is
submaximal, the level of motor unit activation needed can
be minimized when fascicles contract isometrically [25].
This enables the Achilles tendon in particular to accom-
modate a greater proportion of the muscle–tendon unit
length change during running thereby reducing metabolic
cost [194]. Variables which provide an indirect measure of
the neuromuscular systems ability to produce force rapidly
and utilize tendon stiffness were found to improve in other
studies that showed improvements in running performance
and/or key determinants [73, 74, 78–80, 87]. However,
some studies found improvements in running-related
parameters despite observing no alterations in jump per-
formance [33, 76–78, 91], rate of force development [36,
75, 77], or stiffness [33, 74, 89] illustrating that measures
were insufficiently sensitive to detect change, or a combi-
nation of mechanisms is likely to be contributing towards
the enhancements observed.
Heavy RT causes a shift in muscle fiber phenotype, from
the less efficient myosin heavy chain (MHC) IIx to more
oxidative MHC IIa, [195, 196]. A higher proportion of
MHC IIa has been shown to relate to better running
economy [91, 197, 198]; however, whether changes to
MHC properties as a result of ST contribute to an
improvement in RE and performance remains to be deter-
mined. One previous study provided evidence that 4 weeks
of sprint running (30-s bouts) improve RE and also the
percentage of MHC IIx [199]; however, the absence of
endurance
training
may
partly
explain
the
shift
in
Effects of Strength Training on Distance Running
1139
123
phenotype. Over a longer period (6 weeks), Pellegrino and
co-workers [91] found no measurable changes in MHC
isoforms following a PT intervention despite a significant
improvement in 3 km TT performance, suggesting that a
contribution from this mechanism is unlikely for distance
running.
It could also be speculated that improvements in RE due
to improved strength might have resulted in subtle changes
to running kinematics, thus enabling participants to per-
form less work for a given submaximal speed [72]. There is
currently little direct support for this conjecture; however,
previous work has shown that running technique is an
important component of RE [200, 201], and improving hip
strength can reduce undesirable frontal and transverse
plane motion in the lower limb during running [202]. One
study in this review did observe a reduction in EMG
amplitude in the superficial musculature of the lower limb
following ST; however, this wasn’t accompanied by an
improvement in RE [83]. This suggests that favorable
adaptations in neuromuscular control do not necessarily
translate to reducing the metabolic cost of running. Addi-
tionally,
two
studies
showed
significant
increases
(3.0–4.4%) in ground contact time during submaximal
running after an ST intervention [36, 81]; however, only
Giovanelli and colleagues [36] found a corresponding
improvement in RE. Several papers have demonstrated an
inverse relationship between RE and ground contact times
[201, 203, 204], since a lower peak vertical force is
required to generate the same amount of impulse during
longer compared to short ground contacts [25]. Although
there is currently minimal evidence to suggest an ST
intervention increases ground contact time during sub-
maximal running, this mechanism may in part explain the
improvements in RE.
4.10 Strength-Training Prescription
4.10.1 Modality and Exercise Selection
The works included in this review used a variety of ST
modalities; however, the most effective type of training is
currently difficult to discern. Adaptations are specific to the
demands placed upon the body, therefore it would be
expected that HRT, ERT and PT produce somewhat dif-
ferent outcomes [205]. This can be observed in the study by
Berryman and co-workers [80], who observed larger
improvements in explosive concentric power in a group
following an ERT program compared to a group who used
PT. The opposite result occurred for the counter-movement
jump, which places a greater reliance on a plyometric
action; the PT group displayed greater improvements than
the ERT group [80]. Heavy RT, which is characterized by
slow velocities of movement, is likely to improve agonist
muscle activation via enhanced recruitment of the motor
neuron pool, whereas ERT, which involves lighter loads
being moved rapidly, tends to enhance firing frequency and
hence improve rate of force development [164, 165]. Ply-
ometric training develops properties related to the stretch–
shortening cycle function [206], and uses movements pat-
terns which closely mimic the running action (e.g., hopping
and skipping). It is therefore likely that although a variety
of ST methods are capable of improving physiological
parameters relating to distance running performance, the
mechanisms underpinning the response may differ.
In less strength-trained individuals, such as those used in
the studies reviewed, any novel ST stimulus is likely to
provide a sufficient overload to the neuromuscular system
to induce an adaptation in the short term [207]. This is
perhaps why ST is effective even in highly-trained distance
runners [74, 77, 87]. Studies that have attempted to com-
pare ST techniques in distance runners have generally
shown HRT to be superior to ERT or a mixed methods
approach at improving aerobic parameters [57, 63] and
maximal anaerobic running speed [62]. Plyometric training
has also shown superiority to ERT for improvement of RE
in moderately trained runners [80]. Other investigations
have found no differences in the physiological changes
between groups using HRT, ERT or a mixture of modali-
ties [62, 65]. A number of studies have also shown HRT
and/or ERT to be more beneficial to a muscular endurance
style of ST [59, 64, 65, 67, 86]. The addition of whole body
vibration to RT also provides no extra benefit [85].
Although ERT and PT may have more appeal compared to
HRT due to their higher-level of biomechanical similarity
to running, an initial period of HRT is likely to provide an
advantage long-term in terms of reducing injury risk [208]
and eliciting a more pronounced training effect [209].
Taken together, it seems that long-term, a mixed modality
approach to ST is most effective, as this provides the
variety and continual overload required to ensure the
neuromuscular system is constantly challenged. One study
that used a longer intervention period lends support to this
notion, as significant improvements were observed in
strength and physiological measures after 20 and 40 weeks
with a periodized methodology that used several types of
ST [33]. Further research is required to ascertain the long-
term benefits of various ST modalities and the relative
merits of different approaches to sequencing and pro-
gressing these modalities.
As discussed in Sect. 4.1, the exercises selected in an ST
program can potentially influence the magnitude of neu-
romuscular adaptation and thus the impact on physiological
determinants of performance. Exercises using free weights,
which require force to be generated from the leg extensor
muscles in a close-kinetic chain position, are the most
likely to positively transfer to running performance [210].
1140
R. C. Blagrove et al.
123
Examples of RT exercises commonly used include: barbell
squat, deadlifts, step-ups and lunging movement patterns
[31, 33, 36, 72, 79, 85, 88]. Isometric HRT may also have
value for the plantarflexors [84]. Explosive RT, by its very
nature, should avoid a deceleration phase, therefore exer-
cises such as squat jumps and Olympic weightlifting
derivatives should be utilized [33, 80]. To maximize
transfer to distance running performance, particularly at
faster speeds, PT exercises should exhibit short ground
contact times (\0.2 s) [36, 72], which approximates the
contact times observed in competitive middle- [211] and
long-distance running [212], and encourages a rapid exci-
tation–contraction coupling sequence and improved mus-
culotendinous stiffness
[36, 73–75]. Exercises which
possess a low to moderate eccentric demand such as depth
jumps (from a 20–30 cm box), skipping, hopping, speed
bounding appear most suitable [33, 73, 75, 77, 80, 83].
4.10.2 Intra-Session Variables
For non-strength trained individuals, exercise prescription
and gradual progression is important to avoid injury and
overtraining [213]. Most studies initially used 1–2 sets and
progressed to 3–6 sets over the course of the intervention
period for HRT, ERT and PT, which appears appropriate to
circumvent these risks. Several studies utilized a low (3–5)
repetition range in every HRT session [31, 79, 81, 86] at
loads which approached maximum (C 80% 1RM or repe-
tition failure), but did not observe superior benefits com-
pared to investigations that prescribed RT at moderate
loads (60–80% 1RM) and higher repetition ranges (5–15
repetitions). Sets were performed to RM in a number of
studies [32, 38, 72, 79, 81, 88, 89], which was likely
employed as a means of standardizing the intensity of each
set in the absence of 1RM data for participants. Performing
sets which leads to repetition failure induces a high level of
metabolic and neuromuscular fatigue, which may delay
recovery [214]. Although training to repetition failure may
be more important than the load lifted for inducing a
hypertrophy response [215], this is both unfavorable and
unnecessary to optimize gains in strength compared to a
non-repetition failure strategy [216]. Not working to rep-
etition failure also appears to become a more important
feature of RT as ST status increases [216]. Participants
were often instructed to move the weights as rapidly as
possible when performing the concentric phase of RT
exercises, which increases the likelihood of maximizing
neuromuscular adaptations [217]. Plyometric training is
characterized by high eccentric forces compared to running
and RT, therefore repetitions per set were typically low
(4–10 repetitions). Total foot contacts progressed from 30
to 60 repetitions in the first week of an intervention up to
110–228 repetitions after 6–9 weeks [73, 75, 76, 91].
Plyometric exercises were all performed without additional
external resistance in all but one study [73] and in many
cases a short ground contact time [76, 77, 83] and maximal
height [80, 83] were cued to amplify the intensity. An inter-
set recovery period of 2–3 min was typical for HRT, ERT
and PT, which is in line with recommendations for these
training techniques [213]. Where SpT was incorporated
into
ST
programs,
repetition
distances
were
short
(20–150 m) and performed at or close to maximal running
speed [73, 74, 88].
4.10.3 Inter-Session Variables
The majority of studies that demonstrated improvements in
running physiology scheduled ST 2–3 times per week,
which is in line with the guidelines for non-strength trained
individuals [213]. One study used just one session per week
(ERT or PT) and achieved moderate improvements in
strength outcomes and RE after 8 weeks of training [80].
Beattie and associates [33] observed small improvements
(ES: 0.3) in RE using a single ST session (mixed activities)
each week for 20 weeks; however, the participants had
already experienced moderate improvement (ES: 1.0) in
this parameter using a twice weekly program in the
20 weeks prior. For well-trained runners who complete
8–13 running sessions per week [73, 77], it would be useful
to establish the minimal ST dosage required to elicit a
beneficial effect to reduce the risk of overtraining. Equally,
for the recreational runner, ST may take up valuable leisure
time that could be spent running, therefore identifying the
optimal volume and frequency of ST to achieve an
improvement in performance would be desirable. A pre-
vious meta-analysis indicated that two or three sessions per
week provides a large effect on strength, but for the non-
strength trained individual, three sessions is superior to two
sessions per week [218]. More recently, a weak relation-
ship was established between improvement in RE and
weekly frequency of ST sessions in 311 endurance runners
[10]. This suggests that higher weekly volumes of ST
would not necessarily provide greater RE improvements,
therefore two sessions per week is likely to be sufficient
[10].
Given the volume of endurance training participants
were exposed to and the duration of each study, it seems
likely that an attenuation of strength-related adaptation
would have occurred. To minimize this interference phe-
nomenon, it is therefore recommended that a recovery
period of[3 h is provided following high-intensity run-
ning training before ST takes place [146]. In many studies
running training and ST took place on different days [33,
36, 85, 88, 89], and several papers noted a gap of[3 h
between running and ST on the same day [32, 38, 72, 78,
79]. This feature of concurrent training prescription
Effects of Strength Training on Distance Running
1141
123
therefore appears important in ensuring sufficient strength-
adaptations are realized but without compromising running
training. Although there is very little evidence that the
dosage of ST prescribed impaired any endurance-related
adaptations, recent work has highlighted that acute bouts of
RT may cause fatigue sufficient to impair subsequent
running performance, which long term may result in sub-
optimal adaptation [219]. It is therefore recommended that
this potential fatigue is accounted for by allowing at least
24 h recovery between an ST session and an intensive
running session [33, 85, 88, 89].
The results provide compelling evidence that a rela-
tively short period (6 weeks) of ST can enhance physio-
logical qualities related to distance running performance.
Improvements in RE [57] and 10 km TT performance [88]
have also been shown in as little as 4 weeks. A relationship
between intervention duration and improvement in RE has
previously been reported [10], suggesting that longer
periods of ST provide a larger benefit. The same may be
true for v _VO2max; however, more research using longer
periods of ST is required to establish if this is indeed the
case. The benefits to performance also seem to be depen-
dent
on
study
duration
as
most
short
interventions
(6 weeks) tended to produce small TT improvements
(2.4–2.7%, ES: 0.13–0.4) [75, 87, 91], whereas longer
programs (8–11 weeks) resulted in moderate or large per-
formance effects (3.1–5.5%, ES: 0.67–1.50) [32, 73, 88]. It
would seem reasonable to assume that highly-trained dis-
tance runners would require a higher volume of ST to
achieve the same benefit as less experienced runners;
however, this does not appear to be the case. Relatively
short (6–9 weeks) periods of ST improved RE and TT
performance to a similar extent in highly-trained individ-
uals [77, 87] and recreational runners [76, 86, 91]. It is
therefore recommended that future investigations use
periods of 10 weeks or longer to provide further insight
into how ST modalities may impact physiological param-
eters long-term in different types of distance runner.
The time of year or phase of training when the research
was conducted was not reported in the majority of studies.
Several papers indicated that the intervention formed part
of an off-season preparation period [73, 74, 78, 82, 86], but
others scheduled the intervention within the competition
period [32, 38, 87]. Based upon the literature reviewed, it is
currently not possible to provide specific recommendations
for ST in different phases of a runners training macrocycle,
as most studies found at least some physiological or per-
formance benefits to concurrent training. Importantly
though, evidence suggests that choosing to exclude ST
following a successful intervention period results in a
detraining effect which causes improvements to return to
baseline
levels
within
6 weeks
[31].
The
40-week
intervention conducted by Beattie and colleagues [33]
provides evidence that reducing ST volume from two
sessions per week (both with a lower limb HRT emphasis)
during the preparatory phase to one weekly session (ERT
and PT emphasis) during the in-season racing period is
sufficient to at least maintain previous strength and phys-
iological gains. This finding corroborates with a mainte-
nance effect observed in cyclists [220, 221] and soccer
players [222] showing one ST session per week is sufficient
to preserve the strength qualities developed during a pre-
ceding phase of training. Therefore, runners can decrease
ST volume from 2–3 sessions per week (each with a lower
limb focus) in preparatory phases of training to a single
session each week during the competitive season without
fearing a loss of adaptation as a consequence of the
reduction in training density.
It is currently uncertain what volume and intensity of
running and ST are most likely to avoid the interference
effect associated with concurrent training practices. One
option to minimize attenuation of strength development is
to organize activities into periods that concentrate on
developing either strength or endurance adaptation [223].
This polarized approach to planning seems unnecessary
and counterintuitive for distance runners who generally
possess little ST experience, therefore require a minimal
stimulus to create an adaptation. Indeed, studies that
replaced running training with ST [73, 78, 88] found no
greater benefit than those which included ST in a supple-
mentary manner.
4.10.4 Training Supervision
In most studies, the ST routine was supervised and tightly
monitored; however, similar controls were often absent for
the running training participants performed. It seems rea-
sonable to assume that any errors in participants training
logbooks would be similar across intervention and control
groups; however, validity of findings would be improved if
the running component of training had been more tightly
defined. Where supervision of the ST exercises was not
included [76] or only included for the first 2 weeks [36],
strength measures did not improve following the inter-
vention period. This indicates that a suitably qualified
coach is an important feature of an ST programme for a
distance runner who lacks ST experience.
4.11 Limitations
In addition to the limitations already highlighted in this
review,
there
are
other
weaknesses
that
should
be
acknowledged. For many of the studies reviewed, calcu-
lation of an ES was possible for the variables measured,
which provides insight into the meaningfulness and
1142
R. C. Blagrove et al.
123
substantiveness of results. However, despite the qualitative
nature of this review, interpretation of findings was pre-
dominantly based upon reported probability values, which
can be misleading due to low sample sizes and the
heterogeneity in the pool of participants studied. A rela-
tively large number of studies have been included in this
review; however, several parameters (e.g., v _VO2max and
BL) were measured in only a small number of studies,
which increases the possibility that false conclusions may
be drawn.
There was also a lack of detail concerning several
important confounding variables in studies, such as the
nature of running training prescription and participant’s
previous experience in ST. All but seven studies [31, 73,
74, 76, 84, 86, 90, 92] identified that participants had not
been engaged in a program of ST for at least 3 months prior
to the study commencing. Although it is perhaps unlikely
that participants in these seven studies were strength-
trained, this cannot be discounted and may therefore have
influenced findings in these investigations.
5 Conclusion and Future Research
This review is the most comprehensive to date surrounding
the potential impact of ST on the physiological determi-
nants of distance running. The research reviewed suggests
that supplementing the training of a distance runner with
ST is likely to provide improvements to RE, TT perfor-
mance and anaerobic parameters such as maximal sprint
speed. Improvements in RE in the absence of changes in
_VO2max, BL and body composition parameters suggests
that the underlying mechanisms predominantly relate to
alterations in intra-muscular co-ordination and increases in
tendon stiffness which contribute to optimizing force–
length-velocity properties of muscle. Nevertheless, it is
clear that the inclusion of ST does not adversely affect
_VO2max or BL markers. The addition of two to three
supervised ST sessions per week is likely to provide a
sufficient stimulus to augment parameters within a 6- to
14-week period, and benefits are likely to be larger for
interventions of a longer duration. A variety of ST
modalities can be used to achieve similar outcomes
assuming runners are of a non-strength trained status;
however, to maximize long-term adaptations, it is sug-
gested that a periodized approach is adopted with HRT
prioritized initially. Although changes in fat-free mass
were not observed in the majority of studies, a targeted RT
program, which aims to increase muscle mass specifically
around the proximal region of the lower limb may enhance
biomechanical and physiological factors which positively
influence RE.
A number of methodological issues are likely to have
contributed towards the discrepancies in results and should
be acknowledged in future research conducted in this area.
In particular, the measurement of RE should be quantified
as energy cost (rather than oxygen cost) and a variety of
speeds assessed which are relative to the maximum steady
state of each participant. Furthermore, when quantifying
RE and
_VO2max, differences in body size should be
accounted for by using scaling exponents which are
appropriate for the cohort under investigation. Although a
direct measure of v _VO2max has obvious validity, the dis-
crete increments utilized during a maximal test may not
provide the sensitivity required to detect changes which
exist in this parameter following a relatively short inter-
vention. Alternative strategies to quantifying v _VO2max may
provide a solution. It is therefore recommended that future
studies focus their time and efforts on investigating the
effects of ST on physiological variables other than _VO2max
and BL responses, such as RE, v _VO2max and parameters
associated with the critical power model. The nature of the
running training undertaken by participants and strength
training history potentially confounds the outcomes of
studies in this area, therefore attempts should also be made
to control these variables as much as possible.
Although the interference phenomenon is likely to have
blunted the strength adaptations observed, the extent to
which this occurs is currently uncertain due to the absence
of a strength-only training group in the studies reviewed.
For longer term interventions, where improvements inevi-
tably plateau, minimizing attenuation to strength outcomes
(and equally augmenting aerobic adaptation) potentially
becomes more important. Therefore the organization of ST
around running training provides a further avenue for
investigation. Similarly, it would be useful for practitioners
to understand the optimal sequencing of ST modalities
within a long-term program in order to optimize training
outcomes and facilitate a peaking response. Finally, very
few investigations have examined the effect of ST on
specific populations of runners such as young [78], female
[32, 38, 72], and masters’ age [86] competitors, therefore
future research should attempt to address this dearth in
literature.
Compliance with ethical standards
Conflict of interest Richard Blagrove, Glyn Howatson and Philip
Hayes declare that they have no conflict of interest. No funding was
provided to support the preparation of this manuscript.
Open Access This article is distributed under the terms of the
Creative Commons Attribution 4.0 International License (http://
creativecommons.org/licenses/by/4.0/), which permits unrestricted
use, distribution, and reproduction in any medium, provided you give
appropriate credit to the original author(s) and the source, provide a
Effects of Strength Training on Distance Running
1143
123
link to the Creative Commons license, and indicate if changes were
made.
References
1. di Prampero PE, Atchou G, Bruckner JC, Moia C. The ener-
getics of endurance running. Euro J Appl Physiol Occ Physiol.
1986;55(3):259–66.
2. Joyner MJ. Modeling: optimal marathon performance on the
basis
of
physiological
factors.
J
Appl
Physiol.
1991;70(2):683–7.
3. McLaughlin JE, Howley ET, Bassett DR Jr, et al. Test of the
classic model for predicting endurance running performance.
Med Sci Sports Exerc. 2010;42(5):991–7.
4. Billat LV, Koralsztein JP. Significance of the velocity at _VO2max
and
time
to
exhaustion
at
this
velocity.
Sports
Med.
1996;22(2):90–108.
5. Conley DL, Krahenbuhl GS. Running economy and distance
running performance of highly trained athletes. Med Sci Sports
Exerc. 1980;12(5):357–60.
6. Morgan DW, Craib M. Physiological aspects of running econ-
omy. Med Sci Sports Exerc. 1992;24(4):456–61.
7. Saunders PU, Pyne DB, Telford RD, Hawley JA. Factors
affecting running economy in trained distance runners. Sports
Med. 2004;34(7):465–85.
8. Svedenhag J, Sjodin B. Physiological characteristics of elite
male runners in and off-season. Can J Appl Sport Sci.
1985;10(3):127–33.
9. Jones AM, Carter H. The effect of endurance training on
parameters of aerobic fitness. Sports Med. 2000;29(6):373–86.
10. Denadai BS, de Aguiar RA, de Lima LC, et al. Explosive
training and heavy weight training are effective for improving
running economy in endurance athletes: a systematic review and
meta-analysis. Sports Med. 2017;47(3):545–54.
11. Lacour JR, Padilla-Magunacelaya S, Barthelemy JC, Dormois
D. The energetics of middle-distance running. Euro J App
Physiol Occ Physiol. 1990;60(1):38–43.
12. Brandon LJ, Boileau RA. Influence of metabolic, mechanical
and physique variables on middle distance running. J Sports
Med Phys Fit. 1992;32(1):1–9.
13. Padilla S, Bourdin M, Barthelemy JC, Lacour JR. Physiological
correlates of middle-distance running performance. A compar-
ative study between men and women. Euro J App Physiol Occ
Physiol. 1992;65(6):561–6.
14. Brandon LJ. Physiological factors associated with middle dis-
tance running performance. Sports Med. 1995;19(4):268–77.
15. Abe D, Yanagawa K, Yamanobe K, Tamura K. Assessment of
middle-distance running performance in sub-elite young runners
using energy cost of running. Euro J App Physiol Occ Physiol.
1998;77(4):320–5.
16. Ingham SA, Whyte GP, Pedlar C, et al. Determinants of 800-m
and 1500-m running performance using allometric models. Med
Sci Sports Exerc. 2008;40(2):345–50.
17. Rabada´n M, Dı´az V, Caldero´n FJ, et al. Physiological deter-
minants of speciality of elite middle- and long-distance runners.
J Sports Sci. 2011;29(9):975–82.
18. Busso T, Chatagnon M. Modelling of aerobic and anaerobic
energy production in middle-distance running. Eur J App
Physiol. 2006;97(6):745–54.
19. Paavolainen L, Nummela A, Rusko H. Muscle power factors
_VO2maxand as determinants of horizontal and uphill running
performance. Scand J Med Sci Sports. 2000;10(5):286–91.
20. Houmard JA, Costill DL, Mitchell JB, et al. The role of
anaerobic ability in middle distance running performance. Euro
J App Physiol Occ Physiol. 1991;62(1):40–3.
21. Billat V, Renoux JC, Pinoteau J, et al. Reproducibility of run-
ning time to exhaustion at _VO2max in subelite runners. Med Sci
Sports Exerc. 1994;26(2):254–7.
22. Reardon J. Optimal pacing for running 400-and 800-m track
races. Am J Phys. 2013;81(6):428–35.
23. Beattie K, Kenny IC, Lyons M, Carson BP. The effect of
strength training on performance in endurance athletes. Sports
Med. 2014;44(6):845–65.
24. Moore IS. Is there an economical running technique? A review
of modifiable biomechanical factors affecting running economy.
Sports Med. 2016;46(6):793–807.
25. Fletcher JR, MacIntosh BR. Running economy from a muscle
energetics perspective. Front Physiol. 2017;8:433.
26. Balsalobre-Fernandez
C,
Santos-Concejero
J,
Grivas
GV.
Effects of strength training on running economy in highly
trained runners: a systematic review with meta-analysis of
controlled trials. J Strength Cond Res. 2016;30(8):2361–8.
27. Cavanagh PR, Pollock ML, Landa J. A biomechanical com-
parison of elite and good distance runners. Ann NY Acad Sci.
1977;301:328–45.
28. Coetzer P, Noakes TD, Sanders B, et al. Superior fatigue
resistance of elite black South African distance runners. J Appl
Physiol. 1993;75(4):1822–7.
29. Schoenfeld BJ, Ogborn D, Krieger JW. Effects of resistance
training frequency on measures of muscle hypertrophy: a sys-
tematic
review
and
meta-analysis.
Sports
Med.
2016;46(11):1689–97.
30. Wilson JM, Marin PJ, Rhea MR, et al. Concurrent training: a
meta-analysis examining interference of aerobic and resistance
exercises. J Strength Cond Res. 2012;26(8):2293–307.
31. Karsten B, Stevens L, Colpus M, et al. The effects of sport-
specific maximal strength and conditioning training on critical
velocity, anaerobic running distance, and 5-km race perfor-
mance. Int J Sports Physiol Perf. 2016;11(1):80–5.
32. Vikmoen O, Raastad T, Seynnes O, et al. Effects of heavy
strength training on running performance and determinants of
running performance in female endurance athletes. PLoS One.
2016;11(3):e0150799.
33. Beattie K, Carson BP, Lyons M, et al. The effect of strength
training on performance indicators in distance runners. J
Strength Cond Res. 2017;31(1):9–23.
34. Clark AW, Goedeke MK, Cunningham SR, et al. Effects of
pelvic and core strength training on high school cross-country
race times. J Strength Cond Res. 2017;31(8):2289–95.
35. Denadai BS, Greco CC. Resistance training and exercise toler-
ance during high-intensity exercise: moving beyond just running
economy and muscle strength. J Appl Physiol. 2017. https://doi.
org/10.1152/japplphysiol.00800.2017 (jap 00800 2017).
36. Giovanelli N, Taboga P, Rejc E, Lazzer S. Effects of strength,
explosive and plyometric training on energy cost of running in
ultra-endurance athletes. Eur J Sport Sci. 2017;17(7):805–13.
37. Stohanzl M, Balas J, Draper N. Effects of minimal dose of
strength training on running performance in female recreational
runners. J Sports Med Phys Fit. 2017. https://doi.org/10.23736/
s0022-4707.17.07124-9.
38. Vikmoen O, Ronnestad BR, Ellefsen S, Raastad T. Heavy
strength training improves running and cycling performance
following prolonged submaximal work in well-trained female
athletes. Physiol Rep. 2017; 5(5). https://doi.org/10.14814/phy2.
13149.
39. Berryman N, Mujika I, Arvisais D, et al. Strength training for
middle- and long-distance performance: a meta-analysis. Int J
1144
R. C. Blagrove et al.
123
Sports Physiol Perf. 2017; pp 1–27. https://doi.org/10.1123/
ijspp.2017-0032
40. Tanaka H, Swensen T. Impact of resistance training on endur-
ance performance. A new form of cross-training? Sports Med.
1998;25(3):191–200.
41. Jung AP. The impact of resistance training on distance running
performance. Sports Med. 2003;33(7):539–52.
42. Yamamoto LM, Lopez RM, Klau JF, et al. The effects of
resistance training on endurance distance running performance
among highly trained runners: a systematic review. J Strength
Cond Res. 2008;22(6):2036–44.
43. Moher D, Liberati A, Tetzlaff J, et al. Preferred reporting items
for systematic reviews and meta-analyses: the PRISMA state-
ment. Bri Med J. 2009;339:b2535. https://doi.org/10.1136/bmj.
b2535.
44. Mayhew TP, Rothstein JM, Finucane SD, Lamb RL. Muscular
adaptation to concentric and eccentric exercise at equal power
levels. Med Sci Sports Exerc. 1995;27(6):868–73.
45. Baroni BM, Rodrigues R, Franke RA, et al. Time course of
neuromuscular adaptations to knee extensor eccentric training.
Int J Sports Med. 2013;34(10):904–11.
46. Jones AM. Middle- and long-distance running. In: Winter EM,
Jones AM, Davidson RC, et al., editors. Sport and exercise
physiology testing guidelines: volume i–sport testing: The Bri-
tish Association of Sport and Exercise Sciences Guide. London,
UK: Routledge; 2006. p. 152.
47. Ouzzani M, Hammady H, Fedorowicz Z, Elmagarmid A. Ray-
yan-a web and mobile app for systematic reviews. Syst Rev.
2016;5(1):210.
48. Hickson RC. Interference of strength development by simulta-
neously training for strength and endurance. Euro J Appl Physiol
Occ Physiol. 1980;45(2–3):255–63.
49. Hickson RC, Dvorak BA, Gorostiaga EM, et al. Potential for
strength and endurance training to amplify endurance perfor-
mance. J Appl Physiol. 1988;65(5):2285–90.
50. Spurrs R, Murphy A, Watsford M. Plyometric training improves
distance running performance: a case study. J Sci Med Sport.
2002;5(4):41.
51. Glowacki SP, Martin SE, Maurer A, et al. Effects of resistance,
endurance, and concurrent exercise on training outcomes in
men. Med Sci Sports Exerc. 2004;36(12):2119–27.
52. Saunders PU, Pyne DB, Telford RD, et al. Nine weeks of ply-
ometric training improves running economy in highly trained
distance runners. Med Sci Sports Exerc. 2004;36(5):S254.
53. Chtara M, Chamari K, Chaouachi M, et al. Effects of intra-
session concurrent endurance and strength training sequence on
aerobic
performance
and
capacity.
Brit
J
Sports
Med.
2005;39(8):555–60.
54. Hamilton RJ, Paton CD, Hopkins WG. Effect of high-intensity
resistance training on performance of competitive distance
runners. Int J Sports Physiol Perf. 2006;1(1):40–9.
55. Esteve-Lanao J, Rhea MR, Fleck SJ, Lucia A. Running-specific,
periodized strength training attenuates loss of stride length
during intense endurance running. J Strength Cond Res.
2008;22(4):1176–83.
56. Kelly CM, Burnett AF, Newton MJ. The effect of strength
training on three-kilometer performance in recreational women
endurance runners. J Strength Cond Res. 2008;22(2):396–403.
57. Guglielmo LG, Greco CC, Denadai BS. Effects of strength
training
on
running
economy.
Int
J
Sports
Med.
2009;30(1):27–32.
58. Sato K, Mokha M. Does core strength training influence running
kinetics, lower-extremity stability, and 5000-m performance in
runners? J Strength Cond Res. 2009;23(1):133–40.
59. Taipale RS, Mikkola J, Nummela A, et al. Strength training in
endurance runners. Int J Sports Med. 2010;31(7):468–76.
60. Childs D, Ryan M, Reneau P. The effects of core strength
training on maximal running performance in middle distance
running. Med Sci Sports Exerc. 2011;43(5):775.
61. Hasegawa H, Yamauchi T, Kawasaki T, et al. Effects of plyo-
metric training using a portable self-coaching system on running
performance and biomechanical variables in jump exercises. J
Strength Cond Res. 2011;25:S110–1.
62. Mikkola J, Vesterinen V, Taipale R, et al. Effect of resistance
training regimens on treadmill running and neuromuscular per-
formance in recreational endurance runners. J Sports Sci.
2011;29(13):1359–71.
63. Barnes KR, Hopkins WG, McGuigan MR, et al. Effects of
resistance training on running economy and cross-country per-
formance. Med Sci Sports Exerc. 2013;45(12):2322–31.
64. Sedano S, Marin PJ, Cuadrado G, Redondo JC. Concurrent
training in elite male runners: the influence of strength versus
muscular endurance training on performance outcomes. J
Strength Cond Res. 2013;27(9):2433–43.
65. Taipale RS, Mikkola J, Vesterinen V, et al. Neuromuscular
adaptations during combined strength and endurance training in
endurance runners: maximal versus explosive strength training
or a mix of both. Euro J Appl Physiol. 2013;113(2):325–35.
66. Mac´kała K, Stodo´łka J. Effects of explosive type strength
training on selected physical and technical performance char-
acteristics in middle distance running-a case report. Pol J Sport
Tourism. 2014;21(4):228–33.
67. Taipale RS, Mikkola J, Salo T, et al. Mixed maximal and
explosive strength training in recreational endurance runners. J
Strength Cond Res. 2014;28(3):689–99.
68. Bluett KA, De Ste Croix MB, Lloyd RS. A preliminary inves-
tigation into concurrent aerobic and resistance training in youth
runners. Isokinet Exerc Sci. 2015;23(2):77–85.
69. Roschel H, Barroso R, Tricoli V, et al. Effects of strength
training associated with whole-body vibration training on run-
ning economy and vertical stiffness. J Strength Cond Res.
2015;29(8):2215–20.
70. Tong TK, McConnell AK, Lin H, et al. Functional inspiratory
and core muscle training enhances running performance and
economy. J Strength Cond Res. 2016;30(10):2942–51.
71. Vorup J, Tybirk J, Gunnarsson TP, et al. Effect of speed
endurance and strength training on performance, running econ-
omy and muscular adaptations in endurance-trained runners.
Euro J Appl Physiol. 2016;116(7):1331–41.
72. Johnston RE, Quinn TJ, Kertzer R, Vroman NB. Strength
training in female distance runners: impact on running economy.
J Strength Cond Res. 1997;11(4):224–9.
73. Paavolainen L, Ha¨kkinen K, Ha¨ma¨la¨inen I, et al. Explosive-
strength training improves 5-km running time by improving
running
economy
and
muscle
power.
J
Appl
Phys.
1999;86(5):1527–33.
74. Millet GP, Jaouen B, Borrani F, Candau R. Effects of concurrent
endurance and strength training on running economy and VO(2)
kinetics. Med Sci Sports Exerc. 2002;34(8):1351–9.
75. Spurrs RW, Murphy AJ, Watsford ML. The effect of plyometric
training on distance running performance. Euro J Appl Physiol.
2003;89(1):1–7.
76. Turner AM, Owings M, Schwane JA. Improvement in running
economy after 6 weeks of plyometric training. J Strength Cond
Res. 2003;17(1):60–7.
77. Saunders PU, Telford RD, Pyne DB, et al. Short-term plyo-
metric training improves running economy in highly trained
middle and long distance runners. J Strength Cond Res.
2006;20(4):947–54.
78. Mikkola J, Rusko H, Nummela A, et al. Concurrent endurance
and explosive type strength training improves neuromuscular
Effects of Strength Training on Distance Running
1145
123
and anaerobic characteristics in young distance runners. Int J
Sports Med. 2007;28(7):602–11.
79. Storen O, Helgerud J, Stoa EM, Hoff J. Maximal strength
training improves running economy in distance runners. Med
Sci Sports Exerc. 2008;40(6):1087–92.
80. Berryman N, Maurel DB, Bosquet L. Effect of plyometric vs.
dynamic weight training on the energy cost of running. J
Strength Cond Res. 2010;24(7):1818–25.
81. Ferrauti A, Bergermann M, Fernandez-Fernandez J. Effects of a
concurrent strength and endurance training on running perfor-
mance and running economy in recreational marathon runners. J
Strength Cond Res. 2010;24(10):2770–8.
82. Fletcher JR, Esau SP, MacIntosh BR. Changes in tendon stiff-
ness and running economy in highly trained distance runners.
Euro J Appl Physiol. 2010;110(5):1037–46.
83. Bonacci J, Green D, Saunders PU, et al. Plyometric training as
an intervention to correct altered neuromotor control during
running after cycling in triathletes: a preliminary randomised
controlled trial. Phys Ther Sport. 2011;12(1):15–21.
84. Albracht K, Arampatzis A. Exercise-induced changes in triceps
surae tendon stiffness and muscle strength affect running
economy
in
humans.
Euro
J
Appl
Physiol.
2013;113(6):1605–15.
85. Bertuzzi R, Pasqua LA, Bueno S, et al. Strength-training with
whole-body vibration in long-distance runners: a randomized
trial. Int J Sports Med. 2013;34(10):917–23.
86. Piacentini MF, De Ioannon G, Comotto S, et al. Concurrent
strength and endurance training effects on running economy in
master
endurance
runners.
J
Strength
Cond
Res.
2013;27(8):2295–303.
87. Ramirez-Campillo R, Alvarez C, Henriquez-Olguin C, et al.
Effects of plyometric training on endurance and explosive
strength performance in competitive middle- and long-distance
runners. J Strength Cond Res. 2014;28(1):97–104.
88. Skovgaard C, Christensen PM, Larsen S, et al. Concurrent speed
endurance and resistance training improves performance, run-
ning economy, and muscle NHE1 in moderately trained runners.
J Appl Physiol. 2014;117(10):1097–109.
89. Damasceno MV, Lima-Silva AE, Pasqua LA, et al. Effects of
resistance training on neuromuscular characteristics and pacing
during 10-km running time trial. Euro J Appl Physiol.
2015;115(7):1513–22.
90. Schumann M, Mykkanen OP, Doma K, et al. Effects of endur-
ance training only versus same-session combined endurance and
strength training on physical performance and serum hormone
concentrations in recreational endurance runners. Appl Physiol
Nutr Metab. 2015;40(1):28–36.
91. Pellegrino J, Ruby BC, Dumke CL. Effect of plyometrics on the
energy cost of running and MHC and titin isoforms. Med Sci
Sports Exerc. 2016;48(1):49–56.
92. Schumann M, Pelttari P, Doma K, et al. Neuromuscular adap-
tations to same-session combined endurance and strength
training in recreational endurance runners. Int J Sports Med.
2016;37(14):1136–43.
93. Maher CG, Sherrington C, Herbert RD, et al. Reliability of the
PEDro scale for rating quality of randomized controlled trials.
Phys Ther. 2003;83(8):713–21.
94. Hoogkamer W, Kipp S, Spiering BA, Kram R. Altered running
economy directly translates to altered distance-running perfor-
mance. Med Sci Sports Exerc. 2016;48(11):2175–80.
95. Frederick EC, Daniels JT, Hayes JW. The effect of shoe weight
on the aerobic demands of running. In: Bachl N, Prokop L,
Suckert R, editors. Curr top sports med. Vienna: Urban &
Schwarzenberg; 1984. p. 616–25.
96. Morgan DW, Martin PE, Krahenbuhl GS, Baldini FD. Vari-
ability in running economy and mechanics among trained male
runners. Med Sci Sports Exerc. 1991;23(3):378–83.
97. Pereira MA, Freedson PS. Intraindividual variation of running
economy in highly trained and moderately trained males. Int J
Sports Med. 1997;18(2):118–24.
98. Saunders PU, Pyne DB, Telford RD, Hawley JA. Reliability and
variability of running economy in elite distance runners. Med
Sci Sports Exerc. 2004;36(11):1972–6.
99. Blagrove RC, Howatson G, Hayes PR. Test-retest reliability of
physiological parameters in elite junior distance runners fol-
lowing allometric scaling. Eur J Sport Sci. 2017; pp 1–10.
https://doi.org/10.1080/17461391.2017.1364301.
100. Shaw AJ, Ingham SA, Fudge BW, Folland JP. The reliability of
running economy expressed as oxygen cost and energy cost in
trained
distance
runners.
Appl
Physiol
Nutr
Metab.
2013;38(12):1268–72.
101. Rhea MR, Alderman BL. A meta-analysis of periodized versus
nonperiodized strength and power training programs. Res Q
Exerc Sport. 2004;75(4):413–22.
102. McCaw ST, Friday JJ. A comparison of muscle activity between
a free weight and a machine bench press. J Strength Cond Res.
1994;8:259–64.
103. Schwanbeck S, Chilibeck PD, Binsted G. A comparison of free
weight squat to Smith machine squat using electromyography. J
Strength Cond Res. 2009;23(9):2588–91.
104. Young WB. Transfer of strength and power training to sports
performance. Int J Sports Physiol Perf. 2006;1(2):74–83.
105. Cavagna GA, Kaneko M. Mechanical work and efficiency in
level walking and running. J Physiol. 1977;268(2):467–81.
106. Barnes KR, Kilding AE. Running economy: measurement,
norms,
and
determining
factors.
Sports
Med
Open.
2015;1(1):8–15.
107. Bergh U, Sjodin B, Forsberg A, Svedenhag J. The relationship
between body mass and oxygen uptake during running in
humans. Med Sci Sports Exerc. 1991;23(2):205–11.
108. Batterham AM, Jackson AS. Validity of the allometric cascade
model at submaximal and maximal metabolic rates in exercising
men. Respir Physiol Neurobiol. 2003;135(1):103–6.
109. Markovic G, Vucetic V, Nevill AM. Scaling behaviour of
_VO2max in athletes and untrained individuals. Ann Hum Biol.
2007;34(3):315–28.
110. Helgerud J. Maximal oxygen uptake, anaerobic threshold and
running economy in women and men with similar performances
level
in
marathons.
Eur
J
Appl
Physiol
Occ
Physiol.
1994;68(2):155–61.
111. Curran-Everett D. Explorations in statistics: the analysis of
ratios
and
normalized
data.
Adv
Physiol
Educ.
2013;37(3):213–9.
112. Fletcher JR, Esau SP, Macintosh BR. Economy of running:
beyond the measurement of oxygen uptake. J Appl Physiol.
2009;107(6):1918–22.
113. Shaw AJ, Ingham SA, Folland JP. The valid measurement of
running
economy
in
runners.
Med
Sci
Sports
Exerc.
2014;46(10):1968–73.
114. Bassett DR, Howley ET. Limiting factors for maximum oxygen
uptake and determinants of endurance performance. Med Sci
Sports Exerc. 2000;32(1):70–84.
115. Pate RR, Branch JD. Training for endurance sport. Med Sci
Sports Exerc. 1992;24(9 Suppl):S340–3.
116. Tesch PA, Komi PV, Hakkinen K. Enzymatic adaptations con-
sequent to long-term strength training. Int J Sports Med.
1987;8(1 Suppl):66–9.
117. Dudley GA. Metabolic consequences of resistive-type exercise.
Med Sci Sports Exerc. 1988;20(5 Suppl):S158–61.
1146
R. C. Blagrove et al.
123
118. Kraemer WJ, Fleck SJ, Evans WJ. Strength and power training:
physiological mechanisms of adaptation. Exerc Sport Sci Rev.
1996;24:363–97.
119. Hurley BF, Seals DR, Ehsani AA, et al. Effects of high-intensity
strength training on cardiovascular function. Med Sci Sports
Exerc. 1984;16(5):483–8.
120. Wenger HA, Bell GJ. The interactions of intensity, frequency
and duration of exercise training in altering cardiorespiratory
fitness. Sports Med. 1986;3(5):346–56.
121. Martin DE, Vroon DH, May DF, Pilbeam SP. Physiological
changes in elite male distance runners training for Olympic
competition. Phys Sportsmed. 1986;14(1):152–206.
122. Morgan DW, Baldini FD, Martin PE, Kohrt WM. Ten kilometer
performance and predicted velocity
_VO2max at among well-
trained male runners. Med Sci Sports Exerc. 1989;21(1):78–83.
123. Midgley AW, McNaughton LR, Wilkinson M. Is there an
optimal training intensity for enhancing the maximal oxygen
uptake of distance runners?: empirical research findings, current
opinions, physiological rationale and practical recommenda-
tions. Sports Med. 2006;36(2):117–32.
124. Paavolainen LM, Nummela AT, Rusko HK. Neuromuscular
characteristics and muscle power as determinants of 5-km run-
ning performance. Med Sci Sports Exerc. 1999;31(1):124–30.
125. Noakes TD, Myburgh KH, Schall R. Peak treadmill running
velocity during the _VO2max test predicts running performance. J
Sports Sci. 1990;8(1):35–45.
126. Stratton E, O’Brien BJ, Harvey J, et al. Treadmill velocity best
predicts
5000-m
run
performance.
Int
J
Sports
Med.
2009;30(1):40–5.
127. Noakes TD. Implications of exercise testing for prediction of
athletic performance: a contemporary perspective. Med Sci
Sports Exerc. 1988;20(4):319–30.
128. Daniels JT, Scardina N, Hayes J, et al. Elite and subelite female
middle-and long-distance runners. In: Landers DM, editor. Sport
and elite performers. Champaign: Human Kinetics; 1984. p. 57–72.
129. Denadai BS, Greco CC. Can the critical power model explain
the increased peak velocity/power during incremental test after
concurrent strength and endurance training? J Strength Cond
Res. 2017;31(8):2319–23.
130. Jones AM, Vanhatalo A, Burnley M, et al. Critical power:
implications for determination of and exercise tolerance. Med
Sci Sports Exerc. 2010;42(10):1876–90.
131. Bosquet L, Duchene A, Lecot F, et al. Vmax estimate from
three-parameter critical velocity models: validity and impact on
800 m running performance prediction. Euro J Appl Physiol.
2006;97(1):34–42.
132. Bishop D, Jenkins DG. The influence of resistance training on
the critical power function and time to fatigue at critical power.
Aust J Sci Med Sport. 1996;28(4):101–5.
133. Sawyer BJ, Stokes DG, Womack CJ, et al. Strength training
increases endurance time to exhaustion during high-intensity
exercise despite no change in critical power. J Strength Cond
Res. 2014;28(3):601–9.
134. Farrell PA, Wilmore JH, Coyle EF, et al. Plasma lactate accu-
mulation and distance running performance. Med Sci Sports.
1979;11(4):338–44.
135. Fay L, Londeree BR, LaFontaine TP, Volek MR. Physiological
parameters related to distance running performance in female
athletes. Med Sci Sports Exerc. 1989;21(3):319–24.
136. Yoshida T, Udo M, Iwai K, Yamaguchi T. Physiological char-
acteristics related to endurance running performance in female
distance runners. J Sports Sci. 1993;11(1):57–62.
137. Tanaka K, Matsuura Y, Matsuzaka A, et al. A longitudinal
assessment of anaerobic threshold and distance-running perfor-
mance. Med Sci Sports Exerc. 1984;16(3):278–82.
138. Carter H, Jones AM, Doust JH. Effect of 6 weeks of endurance
training
on
the
lactate
minimum
speed.
J
Sports
Sci.
1999;17(12):957–67.
139. Billat V, Sirvent P, Lepretre PM, Koralsztein JP. Training effect
on performance, substrate balance and blood lactate concentra-
tion at maximal lactate steady state in master endurance-runners.
Pflugers Arch. 2004;447(6):875–83.
140. Londeree BR. Effect of training on lactate/ventilatory thresh-
olds:
a
meta-analysis.
Med
Sci
Sports
Exerc.
1997;29(6):837–43.
141. Jacobs I, Esbjornsson M, Sylven C, et al. Sprint training effects
on muscle myoglobin, enzymes, fiber types, and blood lactate.
Med Sci Sports Exerc. 1987;19(4):368–74.
142. Harmer AR, McKenna MJ, Sutton JR, et al. Skeletal muscle
metabolic and ionic adaptations during intense exercise fol-
lowing
sprint
training
in
humans.
J
Appl
Physiol.
2000;89(5):1793–803.
143. Burgomaster KA, Heigenhauser GJ, Gibala MJ. Effect of short-
term sprint interval training on human skeletal muscle carbo-
hydrate metabolism during exercise and time-trial performance.
J Appl Physiol. 2006;100(6):2041–7.
144. Marcinik EJ, Potts J, Schlabach G, et al. Effects of strength
training on lactate threshold and endurance performance. Med
Sci Sports Exerc. 1991;23(6):739–43.
145. Hayes PR, Bowen SJ, Davies EJ. The relationships between
local muscular endurance and kinematic changes during a run to
exhaustion
at
v
_VO2max.
J
Strength
Cond
Res.
2004;18(4):898–903.
146. Baar K. Using molecular biology to maximize concurrent
training. Sports Med. 2014;44(Suppl 2):S117–25.
147. Coffey VG, Pilegaard H, Garnham AP, et al. Consecutive bouts
of diverse contractile activity alter acute responses in human
skeletal muscle. J Appl Physiol. 2009;106(4):1187–97.
148. Wang L, Mascher H, Psilander N, et al. Resistance exercise
enhances the molecular signaling of mitochondrial biogenesis
induced by endurance exercise in human skeletal muscle. J Appl
Physiol. 2011;111(5):1335–44.
149. Green HJ, Patla AE. Maximal aerobic power: neuromuscular
and
metabolic
considerations.
Med
Sci
Sports
Exerc.
1992;24(1):38–46.
150. Bulbulian R, Wilcox AR, Darabos BL. Anaerobic contribution
to distance running performance of trained cross-country ath-
letes. Med Sci Sports Exerc. 1986;18(1):107–13.
151. Blazevich AJ, Jenkins DG. Effect of the movement speed of
resistance training exercises on sprint and strength performance
in concurrently training elite junior sprinters. J Sports Sci.
2002;20(12):981–90.
152. Satkunskiene D, Rauktys D. Stanislovaitis A The effect of
power training on sprint running kinematics. Educ Physical
Train Sport. 2009;72:116–22.
153. Kamandulis S, Skurvydas A, Brazaitis M, et al. Effect of a
periodized power training program on the functional perfor-
mances and contractile properties of the quadriceps in sprinters.
Res Q Exerc Sport. 2012;83(4):540–5.
154. Tucker R, Lambert MI, Noakes TD. An analysis of pacing
strategies during men’s world-record performances in track
athletics. Int J Sports Physiol Perf. 2006;1(3):233–45.
155. Hanley B. Senior men’s pacing profiles at the IAAF World
Cross
Country
Championships.
J
Sports
Sci.
2014;32(11):1060–5.
156. Hanley B. Pacing profiles and pack running at the IAAF World
Half
Marathon
Championships.
J
Sports
Sci.
2015;33(11):1189–95.
157. Sandford GN, Pearson S, Allen SV, et al. Tactical behaviours in
men’s 800 m Olympic and World Championship medallists: a
Effects of Strength Training on Distance Running
1147
123
changing of the guard. Int J Sports Physiol Perform. 2017; pp
1–13. https://doi.org/10.1123/ijspp.2016-0780.
158. Turnes T, Salvador AF, Lisboa FD, et al. A fast-start pacing
strategy speeds pulmonary oxygen uptake kinetics and improves
supramaximal
running
performance.
PLoS
One.
2014;9(10):e111621.
159. Bundle MW, Hoyt RW, Weyand PG. High-speed running per-
formance: a new approach to assessment and prediction. J Appl
Physiol. 2003;95(5):1955–62.
160. Vanhatalo A, Jones AM, Burnley M. Application of critical
power in sport. Int J Sports Physiol Perform. 2011;6(1):128–36.
161. Jenkins DG, Quigley BM. Endurance training enhances critical
power. Med Sci Sports Exerc. 1992;24(11):1283–9.
162. Vanhatalo A, Doust JH, Burnley M. A 3-min all-out cycling test
is sensitive to a change in critical power. Med Sci Sports Exerc.
2008;40(9):1693–9.
163. Rusko HK. Measurement of maximal and submaximal anaero-
bic power: an introduction. Int J Sports Med. 1996;17(Suppl
2):S89–90.
164. Folland JP, Williams AG. The adaptations to strength training:
morphological and neurological contributions to increased
strength. Sports Med. 2007;37(2):145–68.
165. Sale DG. Neural adaptation to resistance training. Med Sci
Sports Exerc. 1988;20(5 Suppl):S135–45.
166. Sale D, MacDougall J, Jacobs I, Garner S. Interaction between
concurrent strength and endurance training. J App Physiol.
1990;68(1):260–70.
167. Kraemer WJ, Patton JF, Gordon SE, et al. Compatibility of high-
intensity strength and endurance training on hormonal and
skeletal muscle adaptations. J Appl Physiol. 1995;78(3):976–89.
168. McCarthy JP, Agre JC, Graf BK, et al. Compatibility of adaptive
responses with combining strength and endurance training. Med
Sci Sports Exerc. 1995;27(3):429–36.
169. McCarthy JP, Pozniak MA, Agre JC. Neuromuscular adapta-
tions to concurrent strength and endurance training. Med Sci
Sports Exerc. 2002;34(3):511–9.
170. Rønnestad BR, Hansen EA, Raastad T. High volume of endur-
ance training impairs adaptations to 12 weeks of strength
training in well-trained endurance athletes. Eur J Appl Physiol.
2012;112(4):1457–66.
171. Lundberg TR, Fernandez-Gonzalo R, Gustafsson T, Tesch PA.
Aerobic exercise does not compromise muscle hypertrophy
response to short-term resistance training. J Appl Physiol.
2013;114(1):81–9.
172. Hennessy LC, Watson AW. The interference effects of training
for strength and endurance simultaneously. J Strength Cond Res.
1994;8(1):12–9.
173. Chtara M, Chaouachi A, Levin GT, et al. Effect of concurrent
endurance and circuit resistance training sequence on muscular
strength
and
power
development.
J
Strength
Cond
Res.
2008;22(4):1037–45.
174. Hakkinen K. Neuromuscular and hormonal adaptations during
strength and power training. A review. J Sports Med Phys Fit.
1989;29(1):9–26.
175. Yang Y, Creer A, Jemiolo B, Trappe S. Time course of myo-
genic and metabolic gene expression in response to acute
exercise
in
human
skeletal
muscle.
J
Appl
Physiol.
2005;98(5):1745–52.
176. Spiering BA, Kraemer WJ, Anderson JM, et al. Resistance
exercise biology: manipulation of resistance exercise pro-
gramme variables determines the responses of cellular and
molecular signalling pathways. Sports Med. 2008;38(7):527–40.
177. Glass DJ. Skeletal muscle hypertrophy and atrophy signaling
pathways. Int J Biochem Cell Biol. 2005;37(10):1974–84.
178. Song Y-H, Godard M, Li Y, et al. Insulin-like growth factor
I-mediated skeletal muscle hypertrophy is characterized by
increased mTOR-p70S6 K signaling without increased Akt
phosphorylation. J Investig Med. 2005;53(3):135–42.
179. Vary TC. IGF-I stimulates protein synthesis in skeletal muscle
through multiple signaling pathways during sepsis. Am J Physiol
Regul Integr Comp Physiol. 2006;290(2):R313–21.
180. Bodine SC. mTOR signaling and the molecular adaptation to
resistance exercise. Med Sci Sports Exerc. 2006;38(11):1950–7.
181. Irrcher I, Adhihetty PJ, Joseph AM, et al. Regulation of mito-
chondrial biogenesis in muscle by endurance exercise. Sports
Med. 2003;33(11):783–93.
182. Hardie DG, Sakamoto K. AMPK: a key sensor of fuel and
energy status in skeletal muscle. Physiol. 2006;21:48–60.
183. Horman S, Browne G, Krause U, et al. Activation of AMP-
activated protein kinase leads to the phosphorylation of elon-
gation factor 2 and an inhibition of protein synthesis. Curr Biol.
2002;12(16):1419–23.
184. Rose AJ, Hargreaves M. Exercise increases Ca2 ? -calmodulin-
dependent protein kinase II activity in human skeletal muscle. J
Physiol. 2003;553(Pt 1):303–9.
185. Baar K. Training for endurance and strength: lessons from cell
signaling. Med Sci Sports Exerc. 2006;38(11):1939–44.
186. Nader GA. Concurrent strength and endurance training: from
molecules to man. Med Sci Sports Exerc. 2006;38(11):1965–70.
187. Dalleau G, Belli A, Bourdin M, Lacour JR. The spring-mass
model and the energy cost of treadmill running. Eur J Appl
Physiol Occup Physiol. 1998;77(3):257–63.
188. Arampatzis A, De Monte G, Karamanidis K, et al. Influence of
the muscle-tendon unit’s mechanical and morphological prop-
erties
on
running
economy.
J
Exp
Biol.
2006;209(Pt
17):3345–57.
189. Dumke CL, Pfaffenroth CM, McBride JM, McCauley GO.
Relationship between muscle strength, power and stiffness and
running economy in trained male runners. Int J Sports Physiol
Perform. 2010;5(2):249–61.
190. Kubo K, Kanehisa H, Fukunaga T. Effects of resistance and
stretching training programmes on the viscoelastic properties of
human tendon structures in vivo. J Physiol. 2002;538(Pt
1):219–26.
191. Foure A, Nordez A, Cornu C. Plyometric training effects on
Achilles tendon stiffness and dissipative properties. J Appl
Physiol. 2010;109(3):849–54.
192. Kyrolainen H, Belli A, Komi PV. Biomechanical factors
affecting
running
economy.
Med
Sci
Sports
Exerc.
2001;33(8):1330–7.
193. Fletcher JR, Groves EM, Pfister TR, Macintosh BR. Can muscle
shortening alone, explain the energy cost of muscle contraction
in vivo? Eur J Appl Physiol. 2013;113(9):2313–22.
194. Fletcher JR, MacIntosh BR. Achilles tendon strain energy in
distance running: consider the muscle energy cost. J Appl
Physiol. 2015;118(2):193–9.
195. Staron RS, Malicky ES, Leonardi MJ, et al. Muscle hypertrophy
and fast fiber type conversions in heavy resistance-trained
women. Eur J Appl Physiol Occup Physiol. 1990;60(1):71–9.
196. Staron RS, Karapondo DL, Kraemer WJ, et al. Skeletal muscle
adaptations during early phase of heavy-resistance training in
men and women. J Appl Physiol. 1994;76(3):1247–55.
197. Kyrolainen H, Kivela R, Koskinen S, et al. Interrelationships
between muscle structure, muscle strength, and running econ-
omy. Med Sci Sports Exerc. 2003;35(1):45–9.
198. Hunter GR, McCarthy JP, Carter SJ, et al. Muscle fiber type,
Achilles tendon length, potentiation, and running economy. J
Strength Cond Res. 2015;29(5):1302–9.
199. Iaia FM, Hellsten Y, Nielsen JJ, et al. Four weeks of speed
endurance training reduces energy expenditure during exercise
and maintains muscle oxidative capacity despite a reduction in
training volume. J Appl Physiol. 2009;106(1):73–80.
1148
R. C. Blagrove et al.
123
200. Williams KR, Cavanagh PR. Relationship between distance
running mechanics, running economy, and performance. J Appl
Physiol. 1987;63(3):1236–45.
201. Folland JP, Allen SJ, Black MI, et al. Running technique is an
important component of running economy and performance.
Med Sci Sports Exerc. 2017;49(7):1412–23.
202. Snyder KR, Earl JE, O’Connor KM, Ebersole KT. Resistance
training is accompanied by increases in hip strength and changes
in lower extremity biomechanics during running. Clin Biomech.
2009;24(1):26–34.
203. Chapman RF, Laymon AS, Wilhite DP, et al. Ground contact
time as an indicator of metabolic cost in elite distance runners.
Med Sci Sports Exerc. 2012;44(5):917–25.
204. Di Michele R, Merni F. The concurrent effects of strike pattern
and ground-contact time on running economy. J Sci Med Sport.
2014;17(4):414–8.
205. Crewther B, Cronin J, Keogh J. Possible stimuli for strength and
power adaptation: acute mechanical responses. Sports Med.
2005;35(11):967–89.
206. Markovic G, Mikulic P. Neuro-musculoskeletal and perfor-
mance adaptations to lower-extremity plyometric training.
Sports Med. 2010;40(10):859–95.
207. Cormie P, McGuigan MR, Newton RU. Adaptations in athletic
performance after ballistic power versus strength training. Med
Sci Sports Exerc. 2010;42(8):1582–98.
208. Lauersen JB, Bertelsen DM, Andersen LB. The effectiveness of
exercise interventions to prevent sports injuries: a systematic
review and meta-analysis of randomised controlled trials. Br J
Sports Med. 2014;48(11):871–7.
209. Cormie P, McGuigan MR, Newton RU. Influence of strength on
magnitude and mechanisms of adaptation to power training.
Med Sci Sports Exerc. 2010;42(8):1566–81.
210. Gamble P. Implications and applications of training specificity
for coaches and athletes. Strength Cond J. 2006;28(3):54.
211. Hayes P, Caplan N. Foot strike patterns and ground contact
times during high-calibre middle-distance races. J Sports Sci.
2012;30(12):1275–83.
212. Hasegawa H, Yamauchi T, Kraemer WJ. Foot strike patterns of
runners at the 15-km point during an elite-level half marathon. J
Strength Cond Res. 2007;21(3):888.
213. Kraemer WJ, Ratamess NA. Fundamentals of resistance train-
ing: progression and exercise prescription. Med Sci Sports
Exerc. 2004;36(4):674–88.
214. Izquierdo M, Ibanez J, Gonzalez-Badillo JJ, et al. Differential
effects of strength training leading to failure versus not to failure
on hormonal responses, strength, and muscle power gains. J
Appl Physiol. 2006;100(5):1647–56.
215. Morton RW, Oikawa SY, Wavell CG, et al. Neither load nor
systemic
hormones
determine
resistance
training-mediated
hypertrophy or strength gains in resistance-trained young men. J
Appl Physiol. 2016;121(1):129–38.
216. Davies T, Orr R, Halaki M, Hackett D. Effect of training leading
to repetition failure on muscular strength: a systematic review
and meta-analysis. Sports Med. 2016;46(4):487–502.
217. Pareja-Blanco F, Rodriguez-Rosell D, Sanchez-Medina L, et al.
Effect of movement velocity during resistance training on neu-
romuscular performance. Int J Sports Med. 2014;35(11):916–24.
218. Rhea MR, Alvar BA, Burkett LN, Ball SD. A meta-analysis to
determine the dose response for strength development. Med Sci
Sports Exerc. 2003;35(3):456–64.
219. Doma K, Deakin GB, Bentley DJ. Implications of impaired
endurance performance following single bouts of resistance
training: an alternate concurrent training perspective. Sports
Med. 2017. https://doi.org/10.1007/s40279-017-0758-3.
220. Rønnestad BR, Hansen EA, Raastad T. In-season strength
maintenance training increases well-trained cyclists’ perfor-
mance. Eur J Appl Physiol. 2010;110(6):1269–82.
221. Rønnestad BR, Hansen J, Hollan I, Ellefsen S. Strength training
improves performance and pedaling characteristics in elite
cyclists. Scand J Med Sci Sports. 2015;25(1):e89–98.
222. Rønnestad BR, Nymark BS, Raastad T. Effects of in-season
strength maintenance training frequency in professional soccer
players. J Strength Cond Res. 2011;25(10):2653–60.
223. Garcia-Pallares J, Izquierdo M. Strategies to optimize concur-
rent training of strength and aerobic fitness for rowing and
canoeing. Sports Med. 2011;41(4):329–43.
Effects of Strength Training on Distance Running
1149
123
| Effects of Strength Training on the Physiological Determinants of Middle- and Long-Distance Running Performance: A Systematic Review. | [] | Blagrove, Richard C,Howatson, Glyn,Hayes, Philip R | eng |
PMC10648636 | Citation: Martinez-Torremocha, G.;
Sanchez-Sanchez, J.; Alonso-Callejo,
A.; Martin-Sanchez, M.L.; Serrano, C.;
Gallardo, L.; Garcia-Unanue, J.;
Felipe, J.L. Physical Demands in the
Worst-Case Scenarios of Elite Futsal
Referees Using a Local Positioning
System. Sensors 2023, 23, 8662.
https://doi.org/10.3390/s23218662
Academic Editors: John Komar and
Ludovic Seifert
Received: 3 October 2023
Revised: 19 October 2023
Accepted: 20 October 2023
Published: 24 October 2023
Copyright:
© 2023 by the authors.
Licensee MDPI, Basel, Switzerland.
This article is an open access article
distributed
under
the
terms
and
conditions of the Creative Commons
Attribution (CC BY) license (https://
creativecommons.org/licenses/by/
4.0/).
sensors
Article
Physical Demands in the Worst-Case Scenarios of Elite Futsal
Referees Using a Local Positioning System
Gemma Martinez-Torremocha 1
, Javier Sanchez-Sanchez 2,*
, Antonio Alonso-Callejo 1,
Maria Luisa Martin-Sanchez 2, Carlos Serrano 2
, Leonor Gallardo 1
, Jorge Garcia-Unanue 1
and Jose Luis Felipe 1
1
IGOID Research Group, Physical Activity and Sport Sciences Department, University of Castilla-La Mancha,
45071 Toledo, Spain; [email protected] (G.M.-T.); [email protected] (A.A.-C.);
[email protected] (L.G.); [email protected] (J.G.-U.); [email protected] (J.L.F.)
2
School of Sport Sciences, Universidad Europea de Madrid, 28670 Villaviciosa de Odón, Spain;
[email protected] (M.L.M.-S.);
[email protected] (C.S.)
*
Correspondence: [email protected]
Abstract: The aim of this study is to analyze the worst-case scenarios of professional futsal referees
during the first and second half of official matches in the Spanish Futsal Cup using a Local Posi-
tioning System (LPS) for monitoring their movement patterns. Eight professional futsal referees
(40 ± 3.43 years; 1.80 ± 0.03 m; 72.84 ± 4.01 kg) participated in the study. The external load (total
distance, high-speed running distance and efforts, sprint distance and efforts, and accelerations
and decelerations distances) of the referees was monitored and collected using an LPS. The results
revealed significant differences in the worst-case scenarios of the futsal referees during the match
according to the time window analyzed (p < 0.05). The longest time windows (120 s, 180 s, and
300 s) showed lower relative total distances in the worst-case scenarios (p < 0.05). The high-speed
running distances were significatively higher in the first half for the 120 s (+2.65 m·min−1; ES: 1.25),
180 s (+1.55 m·min−1; ES: 1.28), and 300 s (+0.95 m·min−1; ES: 1.14) time windows (p < 0.05). No
differences were found between the first and second half for the high-intensity deceleration distance
(p > 0.05). These results will serve to prepare the referees in the best conditions for the competition
and adapt the training plans to the worst-case scenarios.
Keywords: team sport; competition; endurance; game analysis; physical performance
1. Introduction
Team sports have a referee who must regulate the rules so that matches can be carried
out correctly. Referees have an essential task since they must pay attention and perform
precise control over the game’s every single moment [1]. For that reason, the physical
demands of referees have been studied for some years thanks to Global Positioning Systems
(GPS) which are easy to transport and use [2].
Throughout the years, the physical demands in football referees have been stud-
ied [1,3,4]. However, football has different characteristics than futsal [5]. Futsal is a
high-intensity, intermittent team sport which is played indoors and involves short actions
of high intensity with a short recovery time between efforts such as changes of direction,
sprints, accelerations (Acc), and decelerations (Dec) [6]. It is a team sport with two times of
20 min each per game, at a standstill time. So, every time the ball comes out of the field,
time stops until the game resumes [7]. Moreover, in futsal there are two main referees on
the court, who have different functions and must have extraordinary positions on the futsal
field to observe the possible infractions [5].
Football referees cover between 10 km and 12 km per game and about 15% of the
total distance at a high speed (>18 km·h−1) [3,4]. Furthermore, they reach a maximum
Sensors 2023, 23, 8662. https://doi.org/10.3390/s23218662
https://www.mdpi.com/journal/sensors
Sensors 2023, 23, 8662
2 of 10
speed of 28.76 km·h−1 and cover 212.98 m at sprint speed [1]. Weston et al. [8] showed
that the distance covered at a high intensity and the total distance travelled in the first
15 min of the first half of the match are higher than the distance covered in the first 15 min
of the second half. This indicates that referees are subjected to high loads, which may cause
injuries if they do not train at optimal levels that are equivalent to the physical demands
that they have in official matches [9]. Nevertheless, in reference to futsal, Rebelo et al. [5]
demonstrated that referees perform intermittent endurance, of moderate–high intensity,
during a match, with several periods of running and sprints, while also performing long
recovery periods of low intensity.
There are several studies conducted on the profiles of futsal players [7,10], but there
is very little research about futsal referees [5,11,12]. It has been demonstrated that futsal
players from the 1st division of the Portuguese, Spanish, and Russian leagues perform
high-intensity efforts every 43 s, medium-intensity efforts every 37 s, and low-intensity
efforts every 14 s during futsal playoffs’ matches [13]. In addition, a recent study published
that futsal players perform around 70 Acc and Dec at a high intensity and, approximately,
170 changes of direction during official matches [14]. Also, players cover 3749 m in a
match, of which 134.9 m are carried out at a high speed (>18 km·h−1) [10]. Additionally,
Serrano et al. [12] showed that futsal referees cover 5719 m of long distance at slow and
moderate speeds, with a reduction in the second half of the games in the sprints’ distance
and high-speed running (HSR). Ahmed et al. [11] demonstrated that futsal referees have
150.9 beats per minute (bpm) and cover 161.1 m and 114.4 m at a high intensity in the first
and second half, respectively.
For the analysis of the physical demands of the matches of indoor and outdoor
sports, different methods have been used, but GPS is the most common for providing
measurements with validity and accuracy [2]. Additionally, the physical demands have
been analyzed in different ways, as follows: with video for futsal referees [5,11]; via GPS for
official football referees [1,3,4]; and through a Local Positioning System (LPS) with Ultra-
Wide Band (UWB) technology for futsal referees [12] and players [10,15,16]. Therefore,
new methods have been used to detect worst-case scenarios (WCSs). This is the case for
professional football players, as using averages may underestimate peak requirements [17].
The WCS is the most intense time of a game or training [17] (e.g., 30 s). WCSs have
recently started to be included in different studies [15,17] to examine the peak physical
demands at different playing times during matches [18] in several team sports such as
rugby [19], football [17], and futsal [15]. As a result, it would be interesting to use it for
referees. However, although there is some information about the physical demands of
professional futsal referees [11,12], much more evidence is needed to accurately establish
their activity. Furthermore, there is no prior information on WCSs of professional futsal
referees collected using tracking technology devices during official matches.
Therefore, the aim of this study is to analyze the WCSs of professional futsal referees
in the first and second half of official matches in the Spanish Futsal Cup using a LPS for
monitoring their movement patterns. The results of this study will facilitate the design of
training plans according to the physical demands of a match.
2. Materials and Methods
2.1. Experimental Approach to the Problem
The data from seven official Spanish Futsal Cup 2020 matches (first division teams of
the National Spanish Futsal League (LNFS)) were gathered using an UWB technology sys-
tem. This allowed for the quantification of the absolute and relative external training loads,
which were then divided into the first and second halves of the games. The competition
was organized over 4 days divided into four quarterfinals’ games, two semi-finals’ games,
and a final game.
Sensors 2023, 23, 8662
3 of 10
2.2. Participants
Eight professional Spanish futsal referees (age 40 ± 3.43 years; height 1.80 ± 0.03 m;
weight 72.84 ± 4.01 kg), with similar characteristics and the same number of training
sessions per week, were monitored during this study. They were studied over seven
games, which were spread across the quarterfinals, semifinals, and championship game
over four days. The referees were selected by the National Committee of Referees (CTA) for
participating in the Spanish Futsal Cup 2020. All of them had at least 6 years of experience
in the first division of the National Spanish Futsal League, and they must pass different
physical tests each year. Informed written consent was obtained from each participant
after being informed of the study’s requirements. The project was approved and followed
the guidelines established by the local institution—the Bioethics Committee for Clinical
Research of the Virgen de la Salud Hospital in Toledo (Ref.: 2551;17/02/2021).
2.3. Equipment
The referees’ movement patterns throughout each game were tracked using WIMU
PROTM LPS (RealTrack Systems SL, Almería, Spain) with an UWB technology. The reference
technology and the WIMU PROTM inertial device (which the referees transported) are the
two components that make up this technology. The WIMU PROTM has shown an accuracy
(bias: 0.57–5.85%), test–retest reliability (%TEM: 1.19), and inter-unit reliability (bias: 0.18)
in Bastida Castillo et al. [20], as well as a large ICC for the x-coordinate (0.65) and a
very large ICC for the y-coordinate (0.88), with a good 2%TEM [21]. For data recording,
storage, and uploading, each device has a dedicated internal microprocessor with a fast USB
interface [21]. The devices are made up of a variety of sensors, including an UWB chipset
with a signal frequency of 18 Hz, a gyroscope, four accelerometers, a magnetometer, and a
Global Navigation Satellite System (GNSS) [22]. Six antennas make up the reference system,
each of which can transmit and receive radio-frequency signals. The radio-frequency signal
works almost exactly like the GPS system, with the antennas (mainly the master antenna)
calculating the position of the devices in their range as they receive the calculation [21].
In terms of calculating the distance traveled, speed, mean velocity, ACC, and DEC for
intermittent activities, LPS has proven to be accurate and reliable [23,24].
2.4. Procedures
The LPS was installed on the futsal pitch where the games were played, and the
individual WIMU PROTM devices (RealTrack Systems SL, Almería, Spain) were used to
register the physical parameters’ data of external load. The LPS was activated after the
warmup of the referees with an autocalibration of the antennas lasting 5 min [21]. The
placement of the six antennas (Figure 1) was set 5 m apart, forming a hexagon, except for
those positioned at the field’s middle line, which were set 7 m from the perimeter. The
hexagonal shape of the antennas improved signal transmission and reception. The antennas
were then self-calibrated for five minutes, while the master antenna synchronized all the
antennas to a single clock after they had been installed. They were then switched on, one at
a time, with the master antenna being turned on last.
Sensors 2023, 23, 8662
4 of 10
Sensors 2023, 23, x FOR PEER REVIEW
4 of 10
Figure 1. Antenna distribution of the Local Positioning System and distance reference from the fut-
sal pitch. The arrows indicate the distance from the antennas to the court and the black lines indicate
the communication between the antennas [16].
2.5. Data Processing
The physical activity variables were considered in line with previous futsal studies
[12,16]. A specific software (SPROTM v.990) was used to analyze each match’s referees’
performance data. The WCSs were assessed using the WIMU SPROTM software version
990 (RealTrack Systems SL, Almería, Spain), with a rolling average method over each
physical variable selected using five different time windows (30, 60, 120, 180, and 300 s).
This method had been used in previous futsal investigation [15] and other team sports’
[17,25,26]. The physical variables examined were the following: the total distance covered
(TD); the HSR distance (distance covered above 15 km·h−1); the HSR efforts (number of
efforts above 15 km·h−1); the sprint distance (distance covered above 18 km·h−1); the sprint
efforts (number of efforts above 15 km·h−1); and the number (n) and distance (m) of high-
intensity ACC (>3 m·s−2) and DEC (<−3 m·s−2). All the speed variables and Acc and Dec
thresholds selected were in line with previous referee futsal research [12].
2.6. Statistical Analysis
The Shapiro–Wilk test was used to test for the normality of each variable and time
window, resulting in a non-normal distribution (p < 0.05). The non-parametric Wilcoxon
test for paired samples was run for each variable and time window. The same method was
used for comparing values between half-times. The confidence level was set to 95%, and
the p-values < 0.05 were considered significant. The standardized effect size was calculated
for each comparison and classified as negligible (Effect Size (ES) < 0.2), small (ES between
0.2 and 0.6), moderate (ES between 0.6 and 1.2), and large (ES > 1.2). The statistical analysis
was carried out and the figures were created using the RStudio software (R version 4.2.2,
RStudio 2022.12.0, © 2009–2022 Posit Software, PBC).
3. Results
The results revealed significant differences in the WCS of futsal referees during the
match according to the time window analyzed (p < 0.05). The longest time windows (120
s, 180 s, and 300 s) showed lower relative distances in the WCS in comparison to the short-
est intervals (Table 1; p < 0.05). According to the differences between the first and the sec-
ond half in the WCS (Table 2), the HSR distance was significatively higher in the first half
for the 120 s (+2.65 m·min−1; ES: 0.38), 180 s (+1.55 m·min−1; ES: 0.29), and 300 s (+0.95
m·min−1; ES: 0.27) time windows (p < 0.05). The WCS for the total distance was only
Figure 1. Antenna distribution of the Local Positioning System and distance reference from the futsal
pitch. The arrows indicate the distance from the antennas to the court and the black lines indicate the
communication between the antennas [16].
2.5. Data Processing
The physical activity variables were considered in line with previous futsal stud-
ies [12,16]. A specific software (SPROTM v.990) was used to analyze each match’s referees’
performance data. The WCSs were assessed using the WIMU SPROTM software version 990
(RealTrack Systems SL, Almería, Spain), with a rolling average method over each physical
variable selected using five different time windows (30, 60, 120, 180, and 300 s). This method
had been used in previous futsal investigation [15] and other team sports’ [17,25,26]. The
physical variables examined were the following: the total distance covered (TD); the HSR
distance (distance covered above 15 km·h−1); the HSR efforts (number of efforts above
15 km·h−1); the sprint distance (distance covered above 18 km·h−1); the sprint efforts
(number of efforts above 15 km·h−1); and the number (n) and distance (m) of high-intensity
ACC (>3 m·s−2) and DEC (<−3 m·s−2). All the speed variables and Acc and Dec thresholds
selected were in line with previous referee futsal research [12].
2.6. Statistical Analysis
The Shapiro–Wilk test was used to test for the normality of each variable and time
window, resulting in a non-normal distribution (p < 0.05). The non-parametric Wilcoxon
test for paired samples was run for each variable and time window. The same method was
used for comparing values between half-times. The confidence level was set to 95%, and
the p-values < 0.05 were considered significant. The standardized effect size was calculated
for each comparison and classified as negligible (Effect Size (ES) < 0.2), small (ES between
0.2 and 0.6), moderate (ES between 0.6 and 1.2), and large (ES > 1.2). The statistical analysis
was carried out and the figures were created using the RStudio software (R version 4.2.2,
RStudio 2022.12.0, © 2009–2022 Posit Software, PBC).
3. Results
The results revealed significant differences in the WCS of futsal referees during the
match according to the time window analyzed (p < 0.05). The longest time windows (120 s,
180 s, and 300 s) showed lower relative distances in the WCS in comparison to the shortest
intervals (Table 1; p < 0.05). According to the differences between the first and the second
half in the WCS (Table 2), the HSR distance was significatively higher in the first half for the
120 s (+2.65 m·min−1; ES: 0.38), 180 s (+1.55 m·min−1; ES: 0.29), and 300 s (+0.95 m·min−1;
ES: 0.27) time windows (p < 0.05). The WCS for the total distance was only significatively
Sensors 2023, 23, 8662
5 of 10
lower in the second half for the 120 s time window (−7.29 m·min−1; ES: 1.44). Finally,
the high intensity Acc distance was also higher in the first half, but only for the 120 s
(+3.71 m·min−1; ES: 0.87) and 180 s (+2.68 m·min−1; ES: 0.72) time windows (p < 0.05;
Figure 2). No differences were found between the first and second half for the high
intensity Dec distance (p > 0.05).
Sensors 2023, 23, x FOR PEER REVIEW
5 of 10
significatively lower in the second half for the 120 s time window (−7.29 m·min−1; ES: 1.44).
Finally, the high intensity Acc distance was also higher in the first half, but only for the
120 s (+3.71 m·min−1; ES: 0.87) and 180 s (+2.68 m·min−1; ES: 0.72) time windows (p < 0.05;
Figure 2). No differences were found between the first and second half for the high inten-
sity Dec distance (p > 0.05).
Figure 2. Worst-case scenarios of the elite futsal referees in the different time windows in accelera-
tions and decelerations. * Significant differences between halves. a,b,c Significant differences between
time windows [30 (a), 60 (b), 120 (c), 180 (d), 300 (e)].
Table 1. Worst-case scenarios of the elite futsal referees in the different time windows.
Time Window (s)
30 (a)
60 (b)
120 (c)
180 (d)
300 (e)
n = 28
Mean
SD
Mean
SD
Mean
SD
Mean
SD
Mean
SD
Total distance (m·min−1)
143.30
14.61
115.39 a
9.28
94.47 a,b
6.21
87.20 a,b,c
4.76
66.22 a,b,c,d
3.22
HSR Distance (m·min−1)
66.44
17.73
40.23 a
11.33
24.21 a,b
7.06
18.90 a,b,c
5.38
12.89 a,b,c,d
3.40
HSR count (n·min−1)
6.79
2.27
4.00 a
1.12
2.64 a,b
0.61
2.13 a,b,c
0.55
1.48 a,b,c,d
0.27
Sprint Distance (m·min−1)
51.21
15.81
28.38 a
10.90
17.11 a,b
6.67
12.78 a,b,c
4.80
8.39 a,b,c,d
3.06
Sprint count (n·min−1)
5.57
2.13
3.14 a
1.35
1.95 a,b
0.70
1.48 a,b,c
0.57
1.00 a,b,c,d
0.36
HI Acc Distance (m·min−1)
42.46
10.67
26.20 a
7.34
17.21 a,b
4.60
13.63 a,b,c
3.90
9.48 a,b,c,d
3.03
HI Dec Distance (m·min−1)
39.79
10.83
24.99 a
6.30
16.01 a,b
4.20
13.34 a,b,c
3.53
9.31 a,b,c,d
2.56
a,b,c,d Significant differences between time windows [30 (a), 60 (b), 120 (c), 180 (d), 300 (e)].
Figure 2. Worst-case scenarios of the elite futsal referees in the different time windows in accelerations
and decelerations. * Significant differences between halves. a,b,c Significant differences between time
windows [30 (a), 60 (b), 120 (c), 180 (d), 300 (e)].
Table 1. Worst-case scenarios of the elite futsal referees in the different time windows.
Time Window (s)
30 (a)
60 (b)
120 (c)
180 (d)
300 (e)
n = 28
Mean
SD
Mean
SD
Mean
SD
Mean
SD
Mean
SD
Total distance (m·min−1)
143.30
14.61
115.39 a
9.28
94.47 a,b
6.21
87.20 a,b,c
4.76
66.22 a,b,c,d
3.22
HSR Distance (m·min−1)
66.44
17.73
40.23 a
11.33
24.21 a,b
7.06
18.90 a,b,c
5.38
12.89 a,b,c,d
3.40
HSR count (n·min−1)
6.79
2.27
4.00 a
1.12
2.64 a,b
0.61
2.13 a,b,c
0.55
1.48 a,b,c,d
0.27
Sprint Distance (m·min−1) 51.21
15.81
28.38 a
10.90
17.11 a,b
6.67
12.78 a,b,c
4.80
8.39 a,b,c,d
3.06
Sprint count (n·min−1)
5.57
2.13
3.14 a
1.35
1.95 a,b
0.70
1.48 a,b,c
0.57
1.00 a,b,c,d
0.36
HI Acc Distance (m·min−1) 42.46
10.67
26.20 a
7.34
17.21 a,b
4.60
13.63 a,b,c
3.90
9.48 a,b,c,d
3.03
HI Dec Distance (m·min−1) 39.79
10.83
24.99 a
6.30
16.01 a,b
4.20
13.34 a,b,c
3.53
9.31 a,b,c,d
2.56
a,b,c,d Significant differences between time windows [30 (a), 60 (b), 120 (c), 180 (d), 300 (e)].
Sensors 2023, 23, 8662
6 of 10
Table 2. Worst-case scenarios of the elite futsal referees in the different time windows by half-time.
Time Window (s)
30 (a)
60 (b)
120 (c)
180 (d)
300 (e)
n = 14
Mean
SD
Mean
SD
Mean
SD
Mean
SD
Mean
SD
First Half
Total distance (m·min−1)
144.90
12.72
116.42
8.91
98.12 *,a
5.15
88.32 a,b
4.99
66.43 a,b,c
2.76
HSR Distance (m·min−1)
66.15
15.64
41.35
9.72
25.54 *,a
5.87
19.68 a,b
4.92
13.36 *,a,b,c 2.66
Sprint Distance (m·min−1)
51.93
15.01
29.08
10.27
17.59 a
6.10
13.38 a,b
4.45
8.71 a,b,c
2.63
HI Acc Distance (m·min−1) 45.44
11.99
28.24
8.49
19.06 *,a
5.30
14.97 *,a,b 4.59
10.39 a,b,c
3.72
HI Dec Distance (m·min−1) 42.32
11.88
27.43
6.33
17.38 a
4.07
14.03 a,b
3.62
10.06 a,b,c
2.72
Second Half
Total distance (m·min−1)
141.71
16.61
114.35
9.85
90.83 a
4.98
86.07 a,b
4.41
66.01 a,b,c
3.72
HSR Distance (m·min−1)
66.73
20.21
39.12
13.03
22.89 a
8.08
18.13 a,b
5.89
12.42 a,b,c
4.07
Sprint Distance (m·min−1)
50.49
17.11
27.68
11.85
16.62 a
7.40
12.18 a,b
5.21
8.07 a,b,c
3.51
HI Acc Distance (m·min−1) 39.48
8.59
24.17
5.57
15.35 a
2.90
12.29 a,b
2.58
8.58 a,b,c
1.88
HI Dec Distance (m·min−1) 37.25
9.40
22.55
5.44
14.62 a
4.00
12.64 a,b
3.43
8.56 a,b,c
2.23
* Significant differences between halves. a,b,c Significant differences between time windows [30 (a), 60 (b), 120 (c),
180 (d), 300 (e)].
4. Discussion
The findings of this study revealed interesting results which deserve further inves-
tigation. The aim of this research was to analyze the WCS in professional futsal referees
in the first and second half of the match in the Spanish Futsal Cup using an LPS. To our
knowledge, this is the first study that examined the WCS during official matches in the
Spanish Futsal Cup and futsal in general. The main findings of the study were that there
is an extreme influence on the time window analyzed, which confirmed that the longer
the time analyzed, the lower the physical demands’ requirement is during the match.
Additionally, the referees covered longer high-intensity distances and more high-intensity
Acc in the first half of the matches.
Previous studies have analyzed the physical demands of futsal referees [5,11,12,27]
and futsal players [15,16]. The results of this research determined the physical demands
without the time windows, except for the study of Illa et al. [15], which researched the
positional differences in the WCS of elite futsal players. Therefore, there are studies on the
physical demands of referees with similar results, although they do not analyze the WCS.
Our study highlights some interesting results. Curiously, all the variables obtained a
decreasing result among the different time windows in the first and the second half of the
match. These results have important practical applications for training design, as referees
might have shown fatigue during the match. Nevertheless, this fatigue may have been
due to the contextual variables of the match, which should also be studied in the future to
determine whether they really interfere with the results.
In comparison to previous research, similarities were found in the total distance, as it
decreases with time over the course of the match. This study, in contrast with Illa et al.’s [15],
showed that, during the WCS of the match play, the futsal players cover more distance than
the futsal referees for each time window (30 s: 37%, 60 s: 34%, 120 s and 180 s: 31%, and
300 s: 43%). An explanation for these differences might be the continuous displacement
of the players as they are moving constantly or even the type and categories of the match,
although, in this case, it was the same division in the same country. Another possible reason
for the differences is substitutions: while a player can rest during the match, the referee
cannot; however, they do not start under the same circumstances. On the other hand, we
agree with Illa et al. [15] that the WCS decreases as the time window analyzed increases;
this may be due to fatigue occurring during the match. Moreover, Ahmed et al. [11], in the
comparison of performance between halves of a match, determined a decline in the total
distance cover by the Iraq Futsal Premier League referees (3093 m vs. 2850 m), while the
findings in this study reported a similar decrease in the relative total distances covered in
both halves in the different time windows. The total distance is also analyzed by Serrano
et al. [12] (2888.39 ± 122.55 vs. 2831.51 ± 150.26 m), showing similarities to this study in
Sensors 2023, 23, 8662
7 of 10
the decrement of the variable and, also, with Ahmed et al. [11]. The physical demands
were monitored using different devices in Ahmed et al. [11], so this could explain the little
difference with the total results. Moreover, the competitions were different and, therefore,
the contextual demands were also dissimilar. In addition, there are studies that look at the
relative distances of elite futsal players in official matches, showing that they have similar
values between halves [16] or even experience an increase [10] compared to the results of
this research. The contextual variables during the matches might be the reason for these
differences, as well as the kind of competition, since the referees must cover the different
distances depending on the course of the match.
Furthermore, futsal referees cover more distance at sprint speed (>18 km·h−1) during
the WCS of the match play (30 s: +8%, +16%; 60 s: +5%, 15%; 120 s: +6%, +12%; 180 s:
+6%, +22%; and 300 s: +5%, +17%) than futsal defenders and pivots during the match in all
time windows, but cover less distance than wingers (30 s: −5%, 60 s: −13%, 120 s: −17%,
180 s: −17%, and 300 s: −31%) [15]. However, Illa et al. [15] only described the HSR as
>18 km·h−1, while in this study it is studied as >15 km·h−1 and the sprint as >18 km·h−1.
The differences in the distance covered at a high intensity by the referees and players may
be due to the specific nature of their individual roles during the matches. The referees must
always follow the game, as being too far away from fouls may result in incorrect decisions.
Even so, rugby union players [19] had lower values in the longest time window than the
futsal referees (−6%). The reason for the dissimilarity with rugby union players might be
due to the interruptions during the futsal matches (every action that stops the match clock),
in which the futsal referees and players continue moving even when the match is detained,
for example when there are outsides, corner-kicks, etc. [28]. Nevertheless, Serrano et al. [12]
did not study the WCS but showed that referees covered less distance at HSR (>15 km·h−1)
in the second half (235.06 ± 67.58 m vs. 207.78 ± 57.86 m, respectively). The same occurred
in this study, as the distance travelled at HSR decreased from the first to the second half
based on the time windows. The specific situations of futsal matches may slow down the
action in the second half, which could have an impact on the referees’ ability to handle
these demands. Moreover, the number of actions over 18 km·h−1 during matches is higher
in the referees, in the lower windows (30 s, 60 s, and 120 s), than in the players in the
different positions; however, in the longer windows (180 s and 300 s), the wingers have
a higher number of such actions [15]. Nevertheless, the course of the match and the type
of competition of futsal matches might also explain the differences in the results. Another
possible explanation for this difference could be that they have a wider range of motion to
achieve a greater speed. Additionally, the sprint distance of the present results showed a
lower sprint distance (>18 km·h−1) in the second half from every time window, as occurred
in Serrano et al.’s study [12], although they did not study the WCS.
On the other hand, the Acc and Dec variables were barely discussed in this study.
However, these results can be discussed using previous studies. Acc and Dec are an impor-
tant part of the physical demands of futsal referees since the number of high-intensities Acc
and Dec is very high during the matches [12]. Other studies have analyzed stops, sideways
running, and turns during matches [5,11] to report these important actions, as they are one
of the causes of sport injuries. While previous studies have examined the results as whole,
without taking the WCS into account [5,11,12], the result of the present study shows values
of high-intensity Acc and Dec distances of the different time windows. In addition, the
high-intensity Acc and Dec values show a decrease in the second half; these results are
similar to the results of a previous futsal referees’ study [12]. Therefore, high-intensity Acc
and Dec require a high eccentric force which may produce muscular fatigue, so monitoring
them could be useful for designing strength training and injury-prevention programs.
Player performance was not examined in the current study, but player activity and
game development may affect compliance with these requirements because referee activity
is influenced by the match activity [11]. Regardless, based on the data obtained in this study,
training programs should be adjusted to the characteristics of the specific competition to
improve referees’ physical performance and prevent injuries. Illa et al. [15] had studied
Sensors 2023, 23, 8662
8 of 10
the seasonal trend on all the dependent variables for the different positions, which may be
useful for developing training plans to prevent injuries. So, this could be a future, possible
focus for research on futsal referees.
This research had different limitations, as there was a low number of matches and
referees involved. Nevertheless, the study contains all the Spanish Futsal Cup matches.
Understanding the physical profile and performance of the referees may be aided by
the potential correlation between the physical requirements placed on the players during
competitions and the referees’ physical parameters [11,29].
5. Conclusions
Finally, the use of LPS to monitor physical performances provides knowledge of the
specific activity profiles of futsal referees. This information could be useful for making
more accurate training programs and even for developing new physical tests. With all of
this, referees will know their workload requirements for each match during the different
seasons and, also, it will serve them to be able to have different references so that they can
be in the best physical conditions to face the different matches.
One of the most significant findings to emerge from this study is the extreme effect
of the time window, which confirmed that the longer the time analyzed, the lower the
requirement is during the matches. Therefore, time windows can be used in different ways,
depending on their length. Shorter time windows (30 s, 60 s and 120 s) are useful for
designing high-intensity tasks in short periods of time, while longer time windows (180 s
and 300 s) are effective for tasks that are performed over larger spaces where extensive,
high-intensity actions are to be performed.
In addition, the present study also identified that referees have their best results in
the first half of the matches over the longer time windows. The decline in their physical
performance in the second half may be attributed to the referees’ need to keep up with
the players’ pace and the situations that arise during the futsal matches. This could lead
to a reduction in the intensity of the match in the second half, which would affect the
performance of the referees.
Finally, it should be considered that it could be helpful to study contextual variables
when carrying out future studies. Moreover, these results will serve to prepare referees
in the best conditions for the competition and allow to adapt the training plans to critical
match scenarios that may be accompanied by relevant decision making.
Author Contributions: Conceptualization, G.M.-T. and J.S.-S.; methodology, G.M.-T. and C.S.; soft-
ware, M.L.M.-S. and C.S.; validation, G.M.-T., J.S.-S. and J.L.F.; formal analysis, A.A.-C.; investigation,
G.M.-T.; resources, J.G.-U.; data curation, A.A.-C.; writing—original draft preparation, G.M.-T.;
writing—review and editing, G.M.-T., J.S.-S. and J.L.F.; visualization, G.M.-T.; supervision, L.G. and
J.S.-S.; project administration, J.L.F.; funding acquisition, J.G.-U. and L.G. All authors have read and
agreed to the published version of the manuscript.
Funding: This research received no external funding.
Institutional Review Board Statement: The study was conducted in accordance with the Declaration
of Helsinki and approved by the Bioethics Committee for Clinical Research of the Virgen de la Salud
Hospital in Toledo (Ref.: 2551;17/02/2021).
Informed Consent Statement: Informed consent was obtained from all the subjects involved in
the study.
Data Availability Statement: Due to the privacy terms, the data are not available.
Acknowledgments: The authors would like to thank all the officials and The National Commit-
tee of Referees from the Spanish Football Federation. G.M.-T. acknowledges the University of
Castilla-La Mancha and the “Fondo Social Europeo Plus (FSE+)” for funding the development of
her Ph.D. BDNS (identif.): 651201. (2022/9249). A.A.-C. acknowledges the Spanish Ministry of
Science, Innovation, and Universities for funding the development of his Ph.D (Grant Number: FPU
21/04332). This research has been developed with the help of Grant EQC2021-006804-P funded
Sensors 2023, 23, 8662
9 of 10
by MCIN/AEI/10.13039/501100011033 and by the “European Union NextGenerationEU/PRTR”,
Grant EQC2019-005843-P funded by MCIN/AEI/10.13039/50110001103 and ERDF ‘A way of making
Europe’.
Conflicts of Interest: The authors declare no conflict of interest.
References
1.
Martínez-Torremocha, G.; Martin-Sanchez, M.L.; Garcia-Unanue, J.; Felipe, J.L.; Moreno-Pérez, V.; Paredes-Hernández, V.;
Gallardo, L.; Sanchez, J. Physical demands on professional Spanish football referees during matches. Sci. Med. Footb. 2022, 11, 1–7.
[CrossRef]
2.
Scott, M.T.U.; Scott, T.J.; Kelly, V.G. The Validity and Reliability of Global Positioning Systems in Team Sport. J. Strength Cond. Res.
2016, 30, 1470–1490. [CrossRef] [PubMed]
3.
Castillo, D.; Weston, M.; McLaren, S.J.; Cámara, J.; Yanci, J. Relationships Between Internal and External Match-Load Indicators in
Soccer Match Officials. Int. J. Sports Physiol. Perform. 2017, 12, 922–927. [CrossRef] [PubMed]
4.
Castillo, D.; Cámara, J.; Lozano, D.; Berzosa, C.; Yanci, J. The association between physical performance and match-play activities
of field and assistants soccer referees. Res. Sports Med. 2019, 27, 283–297. [CrossRef] [PubMed]
5.
Rebelo, A.N.; Ascensão, A.A.; Magalhães, J.F.; Bischoff, R.; Bendiksen, M.; Krustrup, P. Elite Futsal Refereeing: Activity Profile
and Physiological Demands. J. Strength Cond. Res. 2011, 25, 980–987. [CrossRef]
6.
Yeemin, W.; Dias, C.S.; Fonseca, A.M. A Systematic Review of Psychological Studies Applied to Futsal. J. Hum. Kinet. 2016, 50,
247–257. [CrossRef]
7.
Barbero-Alvarez, J.C.; Soto, V.M.; Barbero-Alvarez, V.; Granda-Vera, J. Match analysis and heart rate of futsal players during
competition. J. Sports Sci. 2008, 26, 63–73. [CrossRef]
8.
Weston, M.; Castagna, C.; Impellizzeri, F.M.; Bizzini, M.; Williams, A.M.; Gregson, W. Science and Medicine Applied to Soccer
Refereeing. Sports Med. 2012, 42, 615–631. [CrossRef]
9.
Riiser, A.; Andersen, V.; Sæterbakken, A.; Ylvisaker, E.; Moe, V.F. Running Performance and Position is Not Related to Decision-
Making Accuracy in Referees. Sports Med. Int. Open 2019, 3, E66–E71. [CrossRef]
10.
Ribeiro, J.N.; Gonçalves, B.; Coutinho, D.; Brito, J.; Sampaio, J.; Travassos, B. Activity Profile and Physical Performance of Match
Play in Elite Futsal Players. Front. Psychol. 2020, 11, 1709. [CrossRef]
11.
Ahmed, H.; Davison, G.; Dixon, D. Analysis of activity patterns, physiological demands and decision-making performance of
elite Futsal referees during matches. Int. J. Perform. Anal. Sport. 2017, 17, 737–751. [CrossRef]
12.
Serrano, C.; Sánchez-Sánchez, J.; Felipe, J.L.; Hernando, E.; Gallardo, L.; Garcia-Unanue, J. Physical Demands in Elite Futsal
Referees During Spanish Futsal Cup. Front. Psychol. 2021, 12, 625154. [CrossRef]
13.
Méndez, C.; Gonçalves, B.; Santos, J.; Ribeiro, J.N.; Travassos, B. Attacking Profiles of the Best Ranked Teams From Elite Futsal
Leagues. Front. Psychol. 2019, 10, 1370. [CrossRef]
14.
Spyrou, K.T.; Freitas, T.; Marín-Cascales, E.; Herrero-Carrasco, R.E.; Alcaraz, P. External match load and the influence of contextual
factors in elite futsal. Biol. Sport. 2022, 39, 349–354. [CrossRef] [PubMed]
15.
Illa, J.; Fernandez, D.; Reche, X.; Serpiello, F.R. Positional Differences in the Most Demanding Scenarios of External Load Variables
in Elite Futsal Matches. Front. Psychol. 2021, 12, 625126. [CrossRef]
16.
Serrano, C.; Felipe, J.L.; Garcia-Unanue, J.; Ibañez, E.; Hernando, E.; Gallardo, L.; Sanchez, J. Local Positioning System Analysis of
Physical Demands during Official Matches in the Spanish Futsal League. Sensors 2020, 20, 4860. [CrossRef] [PubMed]
17.
Oliva-Lozano, J.M.; Rojas-Valverde, D.; Gómez-Carmona, C.D.; Fortes, V.; Pino-Ortega, J. Worst case scenario match analysis and
contextual variables in professional soccer players: A longitudinal study. Biol. Sport. 2020, 37, 429–436. [CrossRef] [PubMed]
18.
Menaspà, P. Are rolling averages a good way to assess training load for injury prevention? Br. J. Sports Med. 2017, 51, 618–619.
[CrossRef] [PubMed]
19.
Reardon, C.; Tobin, D.P.; Tierney, P.; Delahunt, E. The worst case scenario: Locomotor and collision demands of the longest
periods of gameplay in professional rugby union. PLoS ONE 2017, 12, e0177072. [CrossRef] [PubMed]
20.
Bastida Castillo, A.; Gómez Carmona, C.D.; De La Cruz Sánchez, E.; Pino Ortega, J. Accuracy.; intra- and inter-unit reliability, and
comparison between GPS and UWB-based position-tracking systems used for time–motion analyses in soccer. Eur. J. Sport Sci.
2018, 18, 450–457. [CrossRef]
21.
Bastida-Castillo, A.; Gómez-Carmona, C.; De la Cruz-Sánchez, E.; Reche-Royo, X.; Ibáñez, S.; Pino Ortega, J. Accuracy and
Inter-Unit Reliability of Ultra-Wide-Band Tracking System in Indoor Exercise. Appl. Sci. 2019, 9, 939. [CrossRef]
22.
Sczyslo, S.; Schroeder, J.; Galler, S.; Kaiser, T. Hybrid localization using UWB and inertial sensors. In Proceedings of the 2008 IEEE
International Conference on Ultra-Wideband, IEEE, Hannover, Germany, 10–12 September 2008; pp. 89–92.
23.
Serpiello, F.R.; Hopkins, W.G.; Barnes, S.; Tavrou, J.; Duthie, G.M.; Aughey, R.J.; Ball, K. Validity of an ultra-wideband local
positioning system to measure locomotion in indoor sports. J. Sports Sci. 2018, 36, 1727–1733. [CrossRef] [PubMed]
24.
Stevens, T.G.A.; de Ruiter, C.J.; van Niel, C.; van de Rhee, R.; Beek, P.J.; Savelsbergh, G.J.P. Measuring Acceleration and
Deceleration in Soccer-Specific Movements Using a Local Position Measurement (LPM) System. Int. J. Sports Physiol. Perform.
2014, 9, 446–456. [CrossRef] [PubMed]
Sensors 2023, 23, 8662
10 of 10
25.
Martin-Garcia, A.; Castellano, J.; Diaz, A.G.; Cos, F.; Casamichana, D. Positional demands for various-sided games with
goalkeepers according to the most demanding passages of match play in football. Biol. Sport 2019, 36, 171–180. [CrossRef]
26.
Vázquez-Guerrero, J.; Ayala, F.; Garcia, F.; Sampaio, J. The Most Demanding Scenarios of Play in Basketball Competition From
Elite Under-18 Teams. Front. Psychol. 2020, 11, 552. [CrossRef]
27.
Dixon, D. A Pilot Study of the Physiological Demands of Futsal Referees Engaged in International Friendly Matches. Am. J. Sports
Sci. Med. 2014, 2, 103–107. [CrossRef]
28.
Illa, J.; Fernandez, D.; Reche, X.; Carmona, G.; Tarragó, J.R. Quantification of an Elite Futsal Team’s Microcycle External Load by
Using the Repetition of High and Very High Demanding Scenarios. Front. Psychol. 2020, 11, 577624. [CrossRef]
29.
García-Santos, D.; Gómez-Ruano, M.A.; Vaquera, A.; Ibáñez, S.J. Systematic review of basketball referees’ performances. Int. J.
Perform. Anal. Sport 2020, 20, 495–533. [CrossRef]
Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual
author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to
people or property resulting from any ideas, methods, instructions or products referred to in the content.
| Physical Demands in the Worst-Case Scenarios of Elite Futsal Referees Using a Local Positioning System. | 10-24-2023 | Martinez-Torremocha, Gemma,Sanchez-Sanchez, Javier,Alonso-Callejo, Antonio,Martin-Sanchez, Maria Luisa,Serrano, Carlos,Gallardo, Leonor,Garcia-Unanue, Jorge,Felipe, Jose Luis | eng |
PMC4831173 | STUDY PROTOCOL
Open Access
Protocol for evaluating the effects of a
therapeutic foot exercise program on injury
incidence, foot functionality and
biomechanics in long-distance runners:
a randomized controlled trial
Alessandra B. Matias1, Ulisses T. Taddei1, Marcos Duarte2 and Isabel C. N. Sacco1*
Abstract
Background: Overall performance, particularly in a very popular sports activity such as running, is typically
influenced by the status of the musculoskeletal system and the level of training and conditioning of the biological
structures. Any change in the musculoskeletal system’s biomechanics, especially in the feet and ankles, will strongly
influence the biomechanics of runners, possibly predisposing them to injuries. A thorough understanding of the
effects of a therapeutic approach focused on feet biomechanics, on strength and functionality of lower limb
muscles will contribute to the adoption of more effective therapeutic and preventive strategies for runners.
Methods/Design: A randomized, prospective controlled and parallel trial with blind assessment is designed to
study the effects of a "ground-up" therapeutic approach focused on the foot-ankle complex as it relates to the
incidence of running-related injuries in the lower limbs. One hundred and eleven (111) healthy long-distance
runners will be randomly assigned to either a control (CG) or intervention (IG) group. IG runners will participate in a
therapeutic exercise protocol for the foot-ankle for 8 weeks, with 1 directly supervised session and 3 remotely
supervised sessions per week. After the 8-week period, IG runners will keep exercising for the remaining 10 months
of the study, supervised only by web-enabled software three times a week. At baseline, 2 months, 4 months and
12 months, all runners will be assessed for running-related injuries (primary outcome), time for the occurrence of
the first injury, foot health and functionality, muscle trophism, intrinsic foot muscle strength, dynamic foot arch
strain and lower-limb biomechanics during walking and running (secondary outcomes).
Discussion: This is the first randomized clinical trial protocol to assess the effect of an exercise protocol that was
designed specifically for the foot-and-ankle complex on running-related injuries to the lower limbs of long-distance
runners. We intend to show that the proposed protocol is an innovative and effective approach to decreasing the
incidence of injuries. We also expect a lengthening in the time of occurrence of the first injury, an improvement in
foot function, an increase in foot muscle mass and strength and beneficial biomechanical changes while running
and walking after a year of exercising.
(Continued on next page)
* Correspondence: [email protected]
1Department of Physical Therapy, Speech, and Occupational Therapy, School
of Medicine, University of São Paulo, Rua Cipotânea, 51 - Cidade Universitária,
05360-160 São Paulo, São Paulo, Brazil
Full list of author information is available at the end of the article
© 2016 Matias et al. Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0
International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and
reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to
the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver
(http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.
Matias et al. BMC Musculoskeletal Disorders (2016) 17:160
DOI 10.1186/s12891-016-1016-9
(Continued from previous page)
Trial registration: Clinicaltrials.gov Identifier NCT02306148 (November 28, 2014) under the name “Effects of Foot
Strengthening on the Prevalence of Injuries in Long Distance Runners”. Committee of Ethics in Research of the School
of Medicine of the University of Sao Paulo (18/03/2015, Protocol # 031/15).
Keywords: Running, Sports injuries, Exercise therapy, Foot, Biomechanics
Background
Human performance, particularly in one of the most
popular sports activities such as running, is typically influ-
enced by the state of the musculoskeletal system, either by
the level of training and conditioning of the biological
structures, or by the aging process. Although popular
worldwide due to its low cost, versatility, convenience [1],
and health benefits to people of all ages [2], running is as-
sociated with a high prevalence of lower extremity injuries
(between 19.4 and 79.3 %) [3]. The occurrence of injuries
limits the intended benefits by inducing changes in prac-
tice habits [4] or temporary or even permanent cessation
of running. In addition, injuries lead to increased costs
due to medical treatment and/or work absence [5].
The understanding of risk factors associated with these
injuries, particularly the intrinsic factors, can provide
important benefits for runners. Among these intrinsic
factors, those that are noteworthy include biomechanical
factors and muscle functionality of the lower extremities,
particularly the feet. A systematic review by van der
Worp et al. [5] included 11 high-quality longitudinal
studies and concluded that alterations in the biomechan-
ical force distribution patterns, amount of training, his-
tory of previous injuries, increased index of the navicular
drop, and the misalignment of the ankle, knee, and hip
are among the main intrinsic risk factors for running-
related injuries. In addition, extrinsic factors such as the
training surface and the type of footwear are also relevant
risk factors [5]. It is noteworthy that out of these seven di-
verse risk factors, two are related to the foot-ankle com-
plex, demonstrating the importance of maintaining the
health and functionality of its musculoskeletal structures
to prevent injuries. It is also believed that any biomechan-
ical alteration in the musculoskeletal system, in particular
the foot-ankle complex, broadly influences a runner’s
functionality, predisposing him/her to a lesser or greater
extent to injuries, in addition to the possibility of com-
promising his/her quality of life [2, 6].
The foot has a complex structure that can perform a
broad variety of functions in different postural and dy-
namic tasks [7, 8]. This versatility can only be achieved
through its unique arch-shaped architecture and its
powerful intrinsic and extrinsic muscular activity, which
is responsible for the maintenance and control of foot
arches, postural corrections during disturbances, and
torque generation during body displacement [9, 10].
Even with this unique and specialized structure, a high
prevalence of injuries associated with running practices
occurs in this complex. Among the most common hy-
potheses used to explain this high prevalence are factors
such as the excessive ankle/foot pronation in the stance
phase of running [11], the lowering of the medial longi-
tudinal arch due to navicular drop [12, 13], the alteration
of rigidity of the plantar arches [14], and the increase in
impact and acceleration of the tibia during running [15].
Evidence suggests the importance of the intrinsic foot
musculature, showing that fatigue can cause a significant
increase in pronation, which is evaluated by the navicu-
lar drop [12]. In addition, weakness may be a risk factor
for falls in the elderly population [16]. Therefore, it is
understandable that the specific training of foot [13, 17]
and ankle muscles [18–20] is an important tool that im-
proves functions and functionalities of the lower extrem-
ities, as has been shown in recent studies [13, 19–21].
In one of those studies, the unsupervised practice of a
single exercise for the feet (short-foot exercise) four
times a week promoted a decrease in the navicular drop,
an increase in the medial longitudinal arch index, and an
increase in the functionality quality of the intrinsic foot
muscles in asymptomatic individuals [13]. These results
were maintained 1 month after the training had been
completed. Although the results of Mulligan and Cook
[13] are promising, they only measured the foot function
in static conditions and the unsupervised practice of an
isolated exercise for 4 weeks may not have been suffi-
cient to cause a transfer of the static gains for a more
dynamic task where the foot would be more robustly
utilized, according to the star excursion balance test. In
contrast, one study compared two groups: one group
performed a 4-week period of short-foot exercises, in-
cluding 100 repetitions for five seconds each, and the
second group performed a 4-week period of towel-curl
exercises with the same amount of exercise [20]. This
controlled study showed that both groups exhibited de-
creased displacement of the centre of pressure during
the modified star excursion test. Therefore, a load in-
crease in the same exercises used by Mulligan and Cook
[13] resulted in positive effects for postural control.
The same short-foot exercise was practiced by individ-
uals with flat feet in a randomized controlled trial to in-
vestigate its effect on the use of foot orthoses [17]. The
protocol consisted of three to five sets of exercises with
Matias et al. BMC Musculoskeletal Disorders (2016) 17:160
Page 2 of 11
five repetitions each, twice a day, for 8 weeks. In both
study groups, the isometric force and the transversal sec-
tion area of the abductor hallucis muscle were increased
after the interventions, with a significant increase in the
group that used orthoses during exercises. These results
demonstrated that even in structurally unfavourable
conditions, exercise for the foot muscles leads to import-
ant strength gains. It is noteworthy that even with a well-
planned intervention, the lack of a control group and the
evaluation of the muscle strength alone limit the study
conclusions. In addition, the study did not take into ac-
count the potential clinical and functional changes of the
plantar arches, as performed by Goldmann et al. [19]. This
group of researchers investigated the effects of the hallucis
flexors strengthening in the kinetic and kinematic of foot
and ankle during walking, running, and vertical jumping
among university athletes. Training of the experimental
group consisted of isometric contractions of the hallucis
flexors at 90 % of the maximum voluntary contraction
using a dynamometer four times a week for 7 weeks. The
authors observed a significant increase in the performance
of vertical jumping and extensor and flexor momentum of
the metatarsal-phalangeal joint and a gain of 60 to 70 % in
the strength of the hallucis flexors. This study shows that
the flexor muscles of the foot respond in a quick and in-
tense manner to training; even for simple training, the
strengthening of the muscles in question results in global
kinematic and kinetic alterations. It would still be interest-
ing to determine how long these gains would last after the
completion of the intervention and whether more elabor-
ate training, involving more muscles and different pos-
tures and loads, would alter the study outcome, especially
with regard to foot biomechanics during locomotor tasks.
The understanding of the effects of a therapeutic ap-
proach focused on the foot biomechanics of walking and
running, on the strength and functionality of lower ex-
tremity muscles will contribute to the adoption of more
effective therapeutic and preventive strategies for runners.
However, no evidence exists that supports the efficacy of
the therapeutic exercises already used and recommended
for the health of the feet [7, 17, 19, 20, 22] with regard to
preventing recurrent injuries in long-distance runners.
However, one research protocol aims to assess the effects
of ankle and hip muscle strengthening and functional bal-
ance training on running mechanics, postural control, and
injury incidence in novice runners with less than 1 year of
running experience but without focusing on the interven-
tion of intrinsic and extrinsic muscles of the feet [23].
Therefore, a controlled and randomized clinical trial
would determine whether these interventions are effi-
cacious by using the incidence of running-related in-
juries as the primary outcome and following both
intervention and control subjects during a period of
time equal to or greater than 1 year (the period during
which the incidence and prevalence of these injuries
are reported) [4, 16, 24–27].
It is important to highlight that rehabilitation programs
rarely include the intrinsic muscles of the feet in their
therapeutic protocols. The present proposal uses a new
paradigm in which the focus of training and preventive in-
terventions in runners is a “ground-up” approach rather
than the traditional "top-down" approach, which focuses
on the hip strengthening. This new approach, advocated
by Baltich et al. [23], will seek to improve the function of
the ankle-foot complex, which is directly associated with
the absortion and transmission of body forces to the
ground and vice-versa during running.
Hypotheses
Our hypotheses are that the therapeutic exercise proto-
col for the foot-ankle as practiced by long-distance rec-
reational runners for 1 year will:
H 1. Reduce the incidence of running-related injury in
the lower limbs,
H 2. Lengthening the time for the occurrence of the
first running-related injury in the lower limbs,
H 3. Increase intrinsic foot muscle strength,
H 4. Increase foot muscle cross-sectional area and
volume,
H 5. Improve foot health and functionality status,
H 6. Reduce dynamic strain on the foot’s longitudinal
arch during running and walking, and
H 7. Produce beneficial biomechanical changes during
running that denote an improvement in the mechanical
efficiency of absorbing loads and propelling the body
while walking and running. Such changes would
include an increase in the ankle range of motion in the
sagittal plane and increases in 1) ankle extensor
moment and power and 2) knee extensor moment and
power during the second half of the stance phase.
Our aim is therefore to investigate the effects of a
"ground-up" therapeutic approach focused on the foot-
ankle for 1 year as they relate to 1) the incidence of
running-related injuries in the lower limbs of long-distance
runners, 2) time of occurrence of the first injury, 3) foot
health and functionality, 4) strength of the intrinsic foot
muscles; 5) foot muscle trophism, 6) dynamic foot arch
strain and 7) lower-limb biomechanics during walking and
running.
Methods/Design
Overview of the research design
A randomized, prospective controlled and parallel trial
with blind assessment is designed to study the effects of a
"ground-up" therapeutic approach focused on the foot-
ankle concerning the incidence of running-related injuries
Matias et al. BMC Musculoskeletal Disorders (2016) 17:160
Page 3 of 11
to the lower limbs of long-distance runners. This trial has
an allocation ratio of 1:1. Its framework is exploratory to
gather preliminary information on the intervention of
conducting a full scale trial. The trial follows all recom-
mendations established by SPIRIT [28].
Long-distance recreational runners are recruited from
the vicinity of the city of São Paulo and referred to a
physical therapist, who performs the group allocation.
The participants are then referred to another physical
therapist, who performs the initial blind assessment. All
runners allocated to the intervention group (IG) partici-
pate in a protocol of therapeutic exercises for the foot-
ankle complex for 8 weeks, with one session per week
supervised by a physical therapist and three sessions per
week remotely supervised by web-enabled software [29].
They receive access to the web software on the first day
and use it for 8 weeks. After the 8-week period, the IG
runners will continue exercising for 10 more months,
supervised only by the web software three times a week.
The runners allocated to the control group (CG) do not
receive any intervention training, but receive a placebo
stretching exercise program.
All runners will be assessed at baseline and 2 months
(end of intervention). They are then assessed twice more
for follow-up purposes, at 4 and 12 months after the
baseline. Assessments will concern the incidence of
running-related injuries (primary outcome), and all other
secondary outcomes.
The design and flowchart of the protocol are pre-
sented in Fig. 1. The assessments are performed at the
Laboratory of Biomechanics of Human Movement and
Posture (LaBiMPH) at the Physical Therapy, Speech and
Occupational Therapy department of the School of
Medicine of the University of São Paulo, São Paulo,
Brazil.
Participants and recruitment
This study is currently recruiting patients (study start
date: April 2015)
The eligibility criteria for the volunteer runners are:
– aged between 18 and 55 years old
– at least 1 year of running experience
– a weekly training distance greater than 20 km an
less than 100 km as their main physical activity
– within 2 months prior to baseline assessment, lack
of any lower limb musculoskeletal injury or pain
that might lead to stopping running practice
– no prior experience within the last year of isolated
foot and ankle strength training
– not receiving any physical therapy intervention
– no history of using minimalist shoes for running
practice
– no prior experience of barefoot running
Runners are not selected if they have other neuro-
logical or orthopedic impairments (such as congenital
foot malformations, stroke, cerebral palsy, poliomyelitis,
rheumatoid arthritis, prosthesis or moderate or severe
osteoarthritis), major vascular complications (venous or
arterial ulcers), diabetes mellitus, sequelae from poorly
healed fractures or prior lower-limb surgeries.
These runners may use the running technique of
fore-, mid- or rear-foot ground contact, which will be
classified by the strike index, according to Cavanagh
and Lafortune [30].
One hundred and eleven (111) runners will be re-
cruited by radio advertisements, print media and run-
ning association groups at their site of practice around
the city of São Paulo. The potential subjects will be
interviewed by telephone and, when selected, assessed in
the laboratory to confirm all the eligibility criteria. This
first laboratory assessment represents the baseline con-
dition (blind assessment).
The runners allocated to the IG will be treated during
their locally supervised session at the Physical Therapy
Department in an ambulatory setting that assists all the
physical therapy treatments of the Department, providing
a reliable therapeutic environment for the intervention.
Randomization, allocation and blinding
The randomization schedule was prepared using Clinstat
software [31] by an independent researcher (Researcher
1) who was not aware of the numeric code for the CG
and IG groups. A numeric block randomization se-
quence will be kept in opaque envelopes.
After the runners’ agreement to participate and assign-
ment in the research, the allocation into the groups will
be made by another independent researcher (Researcher
2), who also will be unaware of the codes. Only the
physiotherapist (Researcher 3) responsible for the locally
supervised training knows who is receiving the interven-
tion. Researcher 3 will also be responsible for the remote
monitoring of the training by web software [29] and tele-
phone. One physiotherapist (Researcher 4), who will also
be blind to the treatment allocation, will be responsible
for all clinical, functional and biomechanical assess-
ments. Both physiotherapists (researchers 3 and 4) will
be blind to the block size used in the randomization
procedure.
To guarantee the blindness of researcher 4, before
each evaluation, runners will be instructed not to reveal
whether they are in the CG or IG; their questions should
be asked only to the physiotherapist in charge of web
software [29] and local training (Researcher 3).
The trial statistician will also be blind to treatment al-
location until the main treatment analysis has been
completed.
Matias et al. BMC Musculoskeletal Disorders (2016) 17:160
Page 4 of 11
Treatment arms
The CG runners will receive a 5-min placebo routine of
warm-up and muscle-stretching exercises to be per-
formed immediately before every running practice dur-
ing their 8-week study (Additional file 1: Table S3).
The IG runners will receive a therapeutic foot-ankle
exercise protocol for strengthening and improving func-
tionality under the supervision of a physiotherapist
(Researcher 4) once a week for 8 weeks, and a series of
foot-ankle exercises to be performed under remote
supervision through web software [29] three times a
week for the full 1-year length of the study (1 year). Both
locally (Additional file 1: Table S1) and remotely super-
vised therapeutic routines (Additional file 1: Table S2)
will take from 20 to 30 min. In particular, the remotely
supervised practice will be preferentially performed at
Fig. 1 Flow chart of the study’s design
Matias et al. BMC Musculoskeletal Disorders (2016) 17:160
Page 5 of 11
home; the web software includes written descriptions,
photos and videos of each exercise.
Each week, IG runners will be requested to evaluate
the subjective effort of each exercise’s performance using
a score of 0 to 10 either with the web software [29] or to
the physiotherapist during locally supervised practice. If
the effort score ranges from 0 to 5 and the runner’s per-
formance of each exercise is found adequate during the
supervised session by the physiotherapist, the exercises
will increase in difficulty according to the progression
chart in Additional file 1: Table S1 and Table S2. If the
effort score ranges from 6 to 7, the exercise will not in-
crease in difficulty and no progression would be done on
that exercise. Thus, the runner remains in the same ex-
ercise progression until he/she scores 0 to 5 in each par-
ticular exercise. Finally, if an IG runner reports a score
from 8 to 10, the exercise will decrease in difficulty, if
possible, until the subject is able to perform it without
pain or discomfort.
Assessments
A physiotherapist (Researcher 3) who is blind to group
allocation will perform all assessments. Each assessment
will consist of taking a clinical history of personal details,
anthropometry, running practice details (years of prac-
tice, weekly frequency and volume, usual shoe and train-
ing surface, number of races and whether the runner
trains with a running coach), previous orthopedic sur-
gery, other physical activity practiced regularly (previous
to running practice or simultaneously with running) and
an injury history concerning the most important risk
factors previously published [3, 32, 33].
A foot-health status questionnaire [34] will be used to
characterize foot health and functionality. We will use a
Brazilian-Portuguese version (FHSQ-BR) translated and
validated by Ferreira et al. [35]. This instrument is di-
vided into three sections. Section I evaluates foot health
in four domains: foot pain, foot function, footwear and
general foot health. Section II evaluates general health in
four domains: general health, physical activity, social
capacity and vigour. Sections I and II are composed of
questions with answer options presented in affirmative
sentences and corresponding numbers. Section III col-
lects general demographic data of the individuals [36].
We will not use the scores from Section III. Each do-
main scores from 0 to 100 points, where 100 is the best
condition and 0 the worst.
We will access variations in foot posture of the run-
ners using the Foot Posture Index (FPI) [36]. The FPI is
a six component measures that allows multiple segment
evaluation of foot posture on a static measurement and
requires that subjects stand in their relaxed stance pos-
ition looking straight ahead while the assessment is in
process. The assessment consists on the (1) palpation of
the talar head, (2) observation of supra and infra malleo-
lar curvature, (3) observation of the calcaneal frontal
plane position, (4) observation of the bulging in the re-
gion of the talo-navicular joint, (5) observation of the
height and congruence of the medial longitudinal arch
and (6) presence of abduction or adduction of the fore-
foot. Scores reaching from -12 to +12 and normative
values are presented on the literature.
Subjects will then be assessed for intrinsic foot muscles
strength, lower-limb running kinematics and kinetics, and
dynamic foot-arch strain. The feet of 30 % of the partici-
pants in each group (41 participants) will be imaged by
magnetic resonance imaging (MRI) to assess trophism
and strength of the foot intrinsic muscles; this will be
scheduled for the same week of each subject’s baseline
measurements.
After baseline assessment, all subjects will be sched-
uled for two follow-ups assessments, one at 8 weeks and
the other at 16 weeks. They will maintain contact with
the Researcher 3 through the follow-up period by the
web software [29], e-mail and telephone.
Running-related injuries
Running-related injuries will be assessed initially at the
baseline and will be assessed continually throughout the
study by the web software [29]. The definition of
running-related injury was set according to the study of
Macera et al. [4]. They stated that any musculoskeletal
pain or injury that was caused by running practice and
that induces changes in the form, duration intensity or
frequency of training for at least 1 week will be consid-
ered a running-related injury. Only lower-limb injuries
will be accounted during the 12-month period after the
baseline assessment; both the incidence and time of oc-
currence of the first injury will be analyzed.
If any subject presents a new injury during his or her
participation in the study, the injury will be accounted
for and the intervention or placebo intervention will be
discontinued, even though all subjects will still keep be-
ing followed for the completion of the study.
Isometric intrinsic foot muscles strength
Strength of the foot’s intrinsic muscles will be assessed
in trials using a pressure platform (EMED: Novel,
Germany) on which the subjects will place their domin-
ant foot while standing with knees extended. They will
push down as hard as possible using only their hallux
and toes, particularly the metatarsophalangeal joints and
not the hallux interphalangeal joint. A physiotherapist
will determine whether the subject lifted the heel, and
inspect fluctuations in the line of gravity and trunk pos-
ture during each trial. If any changes are observed in the
line of gravity or positioning of the heel or trunk, the
trial will be excluded. Three trials will be completed on
Matias et al. BMC Musculoskeletal Disorders (2016) 17:160
Page 6 of 11
each foot (left and right) according to Mickle et al.
(2006) [37]. Maximum force will be normalized by body
weight and analyzed for hallux and toes areas separately.
Foot muscle trophism and strength
One indirect method of measuring foot strength is
through MRI, which, combined with other techniques,
offers good reliability and a way to follow changes in
muscular volume [38]. In addition, MRI can facilitate
understanding the etiology of running-related injuries
and rehabilitation of the foot-ankle complex [39].
The MRI of the foot will be performed with a 1.5 T sys-
tem. Foot images will be acquired by the same technician
using a coil of four channels positioned in the magnetic
centre. Participants will be placed in supine position with
the ankle at 45° of plantar flexion inside the coil. Images
will be acquired in the frontal, sagittal and transverse
planes to confirm the position of the feet, and the subject
will be repositioned if necessary. T1-weighted images of
the entire foot length will be acquired perpendicular to
the plantar aspect of the foot using a spin-echo sequence
(repetition time = 500 ms, echo time = 16 ms, averages = 3,
slice thickness = 4 mm, gap between slices = 0 mm, field of
view = 120 × 120 mm, flip angle = 90°, matrix = 512 × 512)
[39]. The set of images will cover the distance between the
most proximal and most distal images in which every in-
trinsic foot muscle is visible.
To assess changes in the cross-sectional area (CSA)
and volume of the intrinsic foot muscles, 30 % of the
subjects from each group will have MRI of the foot at
three times: baseline, 8 weeks and 16 weeks.
The CSA will be measured by ImageJ planimeter soft-
ware [40]. Following, Miller et al. [14] for each muscle at
each slice and muscle volume will be calculated by
multiplying the CSA of all slices for a muscle by their
linear distance (4 mm) and adding these volumes.
Walking and running biomechanics
To ensure maximum reliability, all biomechanical testing
sessions will be completed by the same researcher.
Gait and running kinematics will be acquired using
three-dimensional displacements of passive reflective
markers (10 mm in diameter) tracked by nine infrared
cameras at 100 Hz (OptiTrack FLEX: V100, Natural
Point, Corvallis, OR, USA) [41, 42]. Some 14 markers
will be placed on the right subject’s foot according to
Leardini’s protocol [43]. Extra markers will be placed at
the medial knee joint line, lateral knee joint line and bi-
laterally at the iliac spine antero-superior, superior as-
pect of the greater trochanter, and sacrum. These
markers will be used to determine relative joint centres
of rotation for the longitudinal axis of the foot, ankle
and knee. The extra markers from the medial aspect of
the knee joint line will be removed during the dynamic
trial. In addition, three non-collinear reflective markers
will be fixed at two technique clusters. One of the clus-
ters will be placed in the lateral thigh and the other over
the shank.
The laboratory coordinate system will be established at
one corner of the force plate and all initial calculations
will be based on this coordinate system. Each lower-limb
segment (shank and thigh), will be modelled based on
surface markers as a rigid body with a local coordinate
system that coincides with the anatomical axes. Transla-
tions and rotations of each segment will be reported
relative to the neutral positions defined during the initial
static standing trial. All joints will be considered to be
spherical (i.e., with three rotational degrees of freedom).
The foot will be modeled according to Leardini et al.
[43]. That is, the calcaneus, mid-foot and metatarsus are
considered rigid bodies and the longitudinal axis of the
first, second and fifth metatarsal bones and proximal
phalanx of the hallux will be tracked independently.
Ground reaction forces will be acquired by a force
plate (AMTI OR-6-1000, Watertown, MA, USA) with a
sampling frequency of 1 kHz embedded in the centre of
the walkway. Force and kinematic data acquisition will
be synchronized and sampled by an A/D card (AMTI,
DT 3002, 12 bits).
The subjects will go through a habituation period before
the data acquisition to establish confidence and comfort
in the laboratory environment, and to ensure appropriate
movement velocity. To assess lower-extremity running
mechanics, subjects will perform 10 valid over-ground
walking trials and 10 valid over-ground running trials at a
constant velocity (9.5 km/h to 10.5 km/s); these will be
monitored by two photoelectrical sensors (Speed Test Fit
Model, Nova Odessa, Brazil).
The automatic digitizing process, 3D reconstruction of
the markers’ positions and filtering of kinematic data will
be performed using AMASS software (C-motion, Kingston,
ON, Canada). Kinematic data will be processed using a
zero-lag second-order low-pass filter with cutoff frequen-
cies of 6Hz for walking and 12 Hz for running. Ground
reaction force data will be processed using a zero-lag low-
pass Butterworth fourth-order filter with cutoff frequencies
of 50Hz for walking and 200 Hz for running.
A bottom-up inverse dynamics method will be used to
calculate the net moments in the sagittal and frontal planes
of the ankle and knee joints using Visual3D software
(C-motion, Kingston, ON, Canada). The human body will
be
modeled
by
three
linked
segments
(foot,
shank
and thigh) and the inertial properties will be based on
Dempster’s standard regression equations. The moment of
inertia and location of center of mass will be computed
assuming the thigh and shank segments as cylinders.
Calculation of all variables will be performed using a
custom-written MATLAB function (MathWorks, Natick,
Matias et al. BMC Musculoskeletal Disorders (2016) 17:160
Page 7 of 11
MA, USA). Data of only one lower limb (randomly
chosen) per subject will be analyzed and compared.
The following ankle kinematic variables will be analysed:
maximum dorsiflexion at foot contact, maximum plantar-
flexion, maximum dorsiflexion at the toe-off and dorsiflex-
ion range of motion (ROM) in the sagittal plane during the
stance phase. The knee kinematic variables are: maximum
flexion at foot contact, maximum extension, maximum
flexion in the stance phase, ROM on sagittal plane, max-
imum abduction and adduction in the stance phase. The
foot kinematic variables are: elevation/drop of the longitu-
dinal arch angle and of the first, second and fifth metatarsal
bones; rearfoot to forefoot rotation; transverse plane angle
between first and second metatarsal bones and between
second and fifth metatarsal bones; and maximum inversion
and eversion of the calcaneus (frontal plane).
The ankle and knee kinetic variables to be analysed are
net ankle and knee moments normalized by body weight
times height and power normalized by body weight in the
sagittal plane. The ground reaction force variables will be
normalized by body weight and are as followings: first peak
force (body weight – BW), second peak (BW), loading rate
80 [N/ms], defined as the force rate between 20 and 80 %
of the contact of the foot with the ground during the first
peak; loading rate 100 [N/ms], as determined by the force
rate between 0 and 100 % of the first peak and push-off rate
[N/ms], as defined as the rate of the second peak force, be-
tween the minimal values until the second peak.
Dynamic longitudinal foot arch strain
The dynamic longitudinal foot arch strain will be measured
according to Liebermann et al. [44]. The measurement in-
volves navicular height (NH), which is the minimum distance
from the navicular tuberosity relative to the line formed by
the first metatarsal head and the medial process of the calca-
neus. These three landmarks form a plane and NH is inde-
pendent of rear-foot inversion or eversion. Arch strain can
also be quantified by fitting a parabola to markers (with the
navicular head as the vertex) and then measuring the average
curvature at 100 points evenly spaced along the curve.
Outcome measurements
The primary outcome measurement will be the inci-
dence of running-related injuries in the lower limbs
accounted at the end of 12 months of study.
The secondary outcomes will be: 1) the time of the
occurrence of the first injury along the study period (time
to event); 2) foot health and functionality (change from
baseline); 3) foot, ankle and knee kinematics, ankle and
knee joint moments, and knee and ankle power during
walking and running (change from baseline); 4) strength
of the intrinsic foot muscles (change from baseline);
5) foot muscle trophism (change from baseline); and
6) dynamic foot arch strain (change from baseline).
Interventions
Runners allocated to the IG will receive a foot-ankle thera-
peutic exercise protocol for strengthening and improving
functionality. Part of the exercise protocol (12 exercises) is
to be performed once a week under the supervision of a
physiotherapist for 8 weeks (Additional file 1: Table S1).
And a series of eight foot-ankle exercises is also to be
performed three times a week remotely supervised by
web software [29] (Additional file 1: Table S2) for the full
1-year completion time of the study. Each session,
whether supervised locally or remotely, lasts 20 to 30 min.
The therapeutic exercise protocol is described in details in
Additional file: 1 Table S1 and S2.
Gradual and progressive difficulty will be offered to
the runner, respecting any limitation due to pain, fatigue
and/or decrease in performance during execution. The
runners in the IG will access the web software [29] daily,
entering their data regarding performance of the foot
exercise training and ranking their level of difficulty in
each exercise from 0 to 10.
During the locally supervised sessions, the physiotherapist
will focus on proper alignment of the foot-ankle segments,
especially if the runner has difficulty in maintaining it, in a
way that allows no movement compensations.
Runners allocated to the CG will receive a 5-min
placebo warm-up and muscle stretching exercise routine
(Additional file 1: Table S3) that they are to perform for
8 weeks immediately before each running practice. This
placebo training can also be assessed and followed
through the web software [29]. The stretching exercises
are described in detail in Additional file: 1 Table S3. We
hypothesized that a warm-up combined with muscle
stretching exercises would not have any effect on foot
muscular strength and functionality, lower extremity
biomechanics or injury prevention.
Both groups will access the web software [29] daily,
entering their running practice data (daily training
duration and volume) and information concerning the
occurrence of any injury event.
The discontinuation criteria for the exercises during
any session includes cramps, moderate to intense pain,
fatigue or any other condition that exposes the runner
to any discomfort.
The discontinuation criteria for the training includes an
occurrence of a running-related injury in the lower limbs.
If any subject fails to access the web software [29]
for three consecutive weeks without explanation, or
fails to attend the locally supervised training three
consecutive times, that subject will be terminated
from the study.
To improve adherence, several actions will be per-
formed by the researchers in the web software [29]. Data
regarding the subjects running practice, such as training
volume, time of practice and occurrence of injuries, will
Matias et al. BMC Musculoskeletal Disorders (2016) 17:160
Page 8 of 11
be reported to the web software, which will summarize
it and make it viewable in the users’ area. In addition,
for the duration of the study, runners' responses in the
web software concerning their foot-ankle exercise prac-
tice and running training will be stored and be accessible
to the researchers and subjects at any time. If any sub-
ject fails to log in to the web software for more than five
consecutive days, an e-mail will be automatically be sent,
asking the subject to log in to his or her account and re-
port data on the training (or lack of it) for the past week.
The physiotherapist responsible for the
therapeutic
protocol will make phone contact with subjects who fail
to attend to any of the weekly locally supervised ses-
sions. They will also make phone contact with subjects
who do not respond to e-mail reminders from the web
software. Subjects will also be contacted by personal
phone calls if data they reported on the web software is
found to be inconsistent [45].
After the period of intervention and after 12 weeks of
follow up all runners will be questioned about their sat-
isfaction to the training protocol with one question (Did
you enjoy doing the exercises?) with three answer possi-
bilities (No; A Little; A lot). To avoid evaluation bias,
runners will answer this question secretly through an
online-unidentified form sent to their e-mail. Runners
will be informed about the anonymity and this form will
only be accessed after completion of the study.
For the duration of the trial, subjects will be advised not
to engage in any new physical activity or preventive train-
ing protocols for the foot and ankle. If any subject cannot
avoid such behavior, he or she must report this situation
during web software [29] access. Concomitant care, such
as physical therapy, acupuncture or other conventional
medical care, will not be permitted except for runners
who are injured during the study. At the end of 12 months,
CG participants that are interested will receive access to
the software for the foot exercise protocol.
Sample size and statistical analysis
The sample size calculation was made using an effect size
of 0.28 (proportion), considering the categorical primary
outcome variable, which is the incidence of running-
related lower-limb injuries [33]. A sample size of 101 run-
ners is needed to provide 80 % power to detect a moderate
effect difference between the highest and lowest group
injury incidence medians, assuming an alpha of 0.05 and a
χ2 (chi-squared test) statistical design – contingency tables
(df = 1) [46]. Assuming a 10 % dropout rate during the
study, a sample size of 111 runners is needed.
The statistical analysis will be based on intention-to-
treat analysis, and mixed general linear models of analysis
of variance for repeated measure will be used to detect
treatment-time interactions (α = 5 %). The outcome mea-
sures will be compared at baseline, 2, 4 and 12 months.
Effect sizes (Cohen´s d coefficient) will also be provided
between baseline and 2 months and between 2 months
and follow-up (4 and 12 months), if the intervention
shows any treatment effect. The missing data will be
treated by imputation methods depending on the type of
the missing data we will face: missing completely at ran-
dom, missing at random, or missing not at random [47].
Discussion
This clinical trial will provide important data on foot-
training effectiveness, its influence on the incidence of
injuries and its efficacy on strengthening the muscles of the
foot-ankle complex. It will also facilitate the identification
of risk factors and biomechanical mechanisms involved in
injury processes and prevention. We also intend to contrib-
ute new evidence that could be used as a guide for further
studies on biomechanical changes in dynamic tasks result-
ing from the strengthening of the foot-ankle complex.
The few existing clinical trials that have proposed exercise
protocols to reduce the incidence of runners’ injuries have
not included the incidence of injury as a primary outcome.
They also have had short follow-up periods and usually
failed to follow the subjects’ adherence to the program and
the correctness of exercise performance throughout the
study [13, 17, 19, 20]. In contrast, this trial has the incidence
of running-related injuries as a primary outcome, will have a
long period of follow-up (12 months), proposes an interven-
tion training protocol with several exercises that are easy to
perform with short durations for each session (20–30 min)
and does not require subjects to be continuously supervised
by a health professional. In addition, it utilizes open-access
web software [29] that will support adherence control.
We understand that the number of MRIs that we are
performing (on 30 % of the subjects) is limited and might
prevent a broad conclusion about changes in intrinsic foot
muscle cross-sectional area (CSA) and volume.
Running-related injuries in this population cause inter-
ruptions and abandonment of physical activity. They also
could lead to the development of chronic injury that
would prevent the practice of other sports and hence
frustrate the individual’s pursuit of a healthy lifestyle.
Runners are constantly looking for ways to remain free
from injury and the information they receive from coa-
ches or media is often conflicting and varied [48]. Our
protocol has the potential to change the course of this
vicious cycle experienced by long-distance runners.
If our hypothesis that such an exercise protocol reduces
the incidence of running-related injuries to long-distance
runners is confirmed, it could be easily incorporated into
their warm-up routine prior to running practice.
Ethics approval and consent to participate
This trial was approved by the Ethics Committee of the
School of Medicine of the University of São Paulo (Protocol
Matias et al. BMC Musculoskeletal Disorders (2016) 17:160
Page 9 of 11
number n°031/15). Additionally, this trial is registered in
ClinicalTrials.gov (a service of U.S. National Institutes of
Health) Identifier NCT02306148 (November 28, 2014)
under the name “Effects of Foot Strengthening on the Preva-
lence of Injuries in Long Distance Runners”. All runners will
be asked for written informed consent according to the
standard forms and the researcher 4 will obtain them.
Consent to publish
Written informed consent for publication of all images
was obtained from the models.
Availability of data and materials
All personal data from potential or enrolled runners will be
maintained confidential before, during and after the trial by
encoding participant’s name. All data access and storage are
in keeping with National Health and Medical Research
Council guidelines, as approved. All files will be available
from the database published at figshare.com. All important
protocol amendments will be reported to investigators, re-
view boards and trial registration by the Researcher 3.
Additional file
Additional file 1: Table S1. Exercises included in the supervised
sessions by a physiotherapist. Table S2. Exercises included in the
remotely supervised sessions in the web software. Table S3. Warm up
and stretching exercises - Control group. (DOCX 2425 kb)
Abbreviations
CG: Control group; CSA: Cross-sectional area; FHSQ-BR: Foot-health status
questionnaire - Brazilian-Portuguese version; FPI: Foot Posture Index;
IG: Intervention group; MRI: Magnetic resonance imaging; NH: Navicular height.
Competing interests
The authors affirm that this study has not received any funding/assistance
from a commercial organization which may lead to a conflict of interests.
Authors’ contributions
All authors have made substantial contributions to all three of sections (1),
(2) and (3): (1) The conception and design of the study, or acquisition of
data, or analysis and interpretation of data (2) drafting the article or revising
it critically for important intellectual content (3) final approval of the version
to be submitted. And in the protocol the following roles will be played by
the authors: UTT is responsible for the study design, intervention,
interpretation of the data, writing the report and submission of the
manuscript. ABM is responsible for the study design, data collection,
management, analysis, and interpretation, writing the report and submission
of the manuscript. ICNS is responsible for the study design, interpretation of
the data, writing the report and submission of the manuscript.
Acknowledgements
The authors are grateful to the State of São Paulo Research Foundation (FAPESP
2014/27311-9; 2015/14810-0), and the Agency Coordination of Improvement of
Higher Education Personnel (CAPES) for the funding granted to this study. The
funders do not have any role in the study and do not have any authority over
any study activity or in the decision to submit the report for publication. The
authors acknowledge Oliveira CC, Soares L, Amorim LG and Vilas Boas C for the
help with the web-software’s construction.
Author details
1Department of Physical Therapy, Speech, and Occupational Therapy, School
of Medicine, University of São Paulo, Rua Cipotânea, 51 - Cidade Universitária,
05360-160 São Paulo, São Paulo, Brazil. 2Federal University of ABC, Biomedical
Engineering, São Bernardo, São Paulo, Brazil.
Received: 2 March 2016 Accepted: 7 April 2016
References
1.
Paluska SA. An overview of hip injuries in running. Sports Med. 2005;35:991–1014.
2.
Haskell WL, Lee I-M, Pate RR, Powell KE, Blair SN, Franklin BA, Macera CA,
Heath GW, Thompson PD, Bauman A. Physical activity and public health:
updated recommendation for adults from the American College of Sports
Medicine and the American Heart Association. Med Sci Sports Exerc. 2007;
39:1423–34.
3.
van Gent RN, Siem D, van Middelkoop M, van Os AG, Bierma-Zeinstra SM,
Koes BW. Incidence and determinants of lower extremity running injuries in
long distance runners: a systematic review. Br J Sport Med. 2007;41:469–80.
discussion 480.
4.
Macera CA, Pate RR, Powell KE, Jackson KL, Kendrick JS, Craven TE.
Predicting lower-extremity injuries among habitual runners. Arch Intern
Med. 1989;149:2565–8.
5.
Van Der Worp MP, Ten Haaf DSM, Van Cingel R, De Wijer A, Nijhuis-Van Der
Sanden MWG, Bart Staal J. Injuries in runners; a systematic review on risk
factors and sex differences. PLoS One. 2015;10:1–18.
6.
Hespanhol Junior LC, van Mechelen W, Postuma E, Verhagen E. Health and
economic burden of running-related injuries in runners training for an
event: A prospective cohort study. Scand J Med Sci Sports 2015:1–9. http://
onlinelibrary.wiley.com/doi/10.1111/sms.12541/abstract.
7.
McKeon PO, Hertel J, Bramble D, Davis I. The foot core system: a new paradigm
for understanding intrinsic foot muscle function. Br J Sports Med. 2014; 0:1–9.
8.
Saltzman CL, Nawoczenski DA. Complexities of foot architecture as a base
of support. J Orthop Sports Phys Ther. 1995;21:354–60.
9.
Dubin A. Gait: the role of the ankle and foot in walking. Med Clin North
Am. 2014;98:205–11.
10.
Dugan SA, Bhat KP. Biomechanics and analysis of running gait. Phys Med
Rehabil Clin N Am. 2005;16:603–21.
11.
Jam B. Evaluation and retraining of the intrinsic foot muscles for pain
syndromes related to abnormal control of pronation. Available at: http://www.
aptei.com/articles/pdf/IntrinsicMuscles.pdf . Accessed 10 November 2015.
12.
Headlee DL, Leonard JL, Hart JM, Ingersoll CD, Hertel J. Fatigue of the
plantar intrinsic foot muscles increases navicular drop. J Electromyogr
Kinesiol. 2008;18:420–5.
13.
Mulligan EP, Cook PG. Effect of plantar intrinsic muscle training on medial
longitudinal arch morphology and dynamic function. Man Ther. 2013;18:425–30.
14.
Miller EE, Whitcome KK, Lieberman DE, Norton HL, Dyer RE. The effect of
minimal shoes on arch structure and intrinsic foot muscle strength. J Sport
Heal Sci. 2014;3:74–85.
15.
Crowell HP, Davis IS. Gait retraining to reduce lower extremity loading in
runners. Clin Biomech. 2011;26:78–83.
16.
Mickle KJ, Munro BJ, Lord SR, Menz HB, Steele JR. ISB Clinical Biomechanics
Award 2009. Toe weakness and deformity increase the risk of falls in older
people. Clin Biomech. 2009;24:787–91.
17.
Jung DY, Kim MH, Koh EK, Kwon OY, Cynn HS, Lee WH. A comparison in the
muscle activity of the abductor hallucis and the medial longitudinal arch angle
during toe curl and short foot exercises. Phys Ther Sport. 2011;12:30–5.
18.
Green SM, Briggs PJ. Flexion strength of the toes in the normal foot. An
evaluation using magnetic resonance imaging. Foot. 2013;23:115–9.
19.
Goldmann J-P, Brüggemann G-P. The potential of human toe flexor muscles
to produce force. J Anat. 2012;221:187–94.
20.
Lynn SK, Padilla RA, Tsang KKW. Differences in static- and dynamic-balance
task performance after 4 weeks of intrinsic-foot-muscle training: the short-
foot exercise versus the towel-curl exercise. J Sport Rehabil. 2012;21:327–33.
21.
Menz HB, Morris ME, Lord SR. Foot and ankle risk factors for falls in older
people: a prospective study. J Gerontol A Biol Sci Med Sci. 2006;61:866–70.
22.
Sherman KP. The foot in sport. Br J Sports Med. 1999;33:6–13.
23.
Baltich J, Emery CA, Stefanyshyn D, Nigg BM. The effects of isolated ankle
strengthening and functional balance training on strength, running
mechanics, postural control and injury prevention in novice runners: design
of a randomized controlled trial. BMC Musculoskelet Disord. 2014;15:407.
24.
Walter SD, Hart LE, McIntosh JM, Sutton JR. The Ontario cohort study of
running-related injuries. Arch Intern Med. 1989;149:2561–4.
Matias et al. BMC Musculoskeletal Disorders (2016) 17:160
Page 10 of 11
25.
Bennell KL, Malcolm SA, Thomas SA, Wark JD, Brukner PD. The incidence
and distribution of stress fractures in competitive track and field athletes. A
twelve-month prospective study. Am J Sports Med. 1996;24:211–7.
26.
Bovens AM, Janssen GM, Vermeer HG, Hoeberigs JH, Janssen MP,
Verstappen FT. Occurrence of running injuries in adults following a
supervised training program. Int J Sports Med. 1989;10 Suppl 3:S186–90.
27.
Lysholm J, Wiklander J. Injuries in runners. Am J Sports Med. 1987;15:168–71.
28.
Chan A, Tetzlaff JM, Altman DG, Laupacis A, Gøtzsche PC, Hro A, Mann H,
Dickersin K, Berlin JA, Dore CJ, Parulekar WR, Summerskill WSM, Groves T,
Schulz KF, Sox HC, Rockhold FW, Rennie D, Moher D. Research and
Reporting Methods Annals of Internal Medicine SPIRIT 2013 Statement :
Defining Standard Protocol Items for Clinical Trials. Ann Intern Med. 2013;
158:200–7.
29.
SAEC - Software for home-based foot and ankle exercises for runners.
[http://biton.uspnet.usp.br/labimph/?page_id=1820]. Accessed 20 Feb 2015.
30.
Cavanagh PR, Lafortune MA. Ground reaction forces in distance running.
J Biomech. 1980;13:397–406.
31.
Bland M. Estimating Mean and Standard Deviation from the Sample Size,
Three Quartiles, Minimum, and Maximum. Int J Stat Med Res. 2014;4:57–64.
32.
Saragiotto BT, Yamato TP, Hespanhol Junior LC, Rainbow MJ, Davis IS, Lopes
AD. What are the main risk factors for running-related injuries? Sport Med.
2014;44:1153–63.
33.
Taunton JE, Ryan MB, Clement DB, McKenzie DC, Lloyd-Smith DR, Zumbo BD.
A prospective study of running injuries: the Vancouver Sun Run “In Training”
clinics. Br J Sports Med. 2003;37:239–44.
34.
Bennett PJ, Patterson C, Wearing S, Baglioni T. Development and validation of a
questionnaire designed to measure foot-health status. J Am Podiatr Med Assoc.
1998;88:419–28.
35.
Ferreira AFB, Laurindo IMM, Rodrigues PT, Ferraz MB, Kowalski SC, Tanaka C.
Brazilian version of the foot health status questionnaire (FHSQ-BR): cross-cultural
adaptation and evaluation of measurement properties. Clinics (Sao Paulo). 2008;
63:595–600.
36.
Redmond AC, Crosbie J, Ouvrier RA. Development and validation of a novel
rating system for scoring standing foot posture: the Foot Posture Index. Clin
Biomech. 2006;21:89–98.
37.
Mickle KJ, Chambers S, Steele JR, Munro BJ. A novel and reliable method to
measure toe flxor strength. Clin Biomech. 2008;23:683.
38.
Soysa A, Hiller C, Refshauge K, Burns J. Importance and challenges of
measuring intrinsic foot muscle strength. J Foot Ankle Res. 2012;5:29.
39.
Chang R, Kent-Braun JA, Hamill J. Use of MRI for volume estimation of
tibialis posterior and plantar intrinsic foot muscles in healthy and chronic
plantar fasciitis limbs. Clin Biomech. 2012;27:500–5.
40.
Schneider CA, Rasband WS, Eliceiri KW. NIH Image to ImageJ: 25 years of
image analysis. Nat Methods. 2012;9:671–5.
41.
Trombini-Souza F, Matias A, Yokota M, Schainberg C, Fuller R, Sacco IC. Low
cost minimalist shoe as a mechanical treatment for algo-functional aspects
and analgesic medicine intake in elderly women with knee osteoarthritis.
Osteoarthr Cartil. 2016;22:S195.
42.
Trombini-Souza F, Fuller R, Matias AB, Yokota M, Butugan MK, Goldenstein-
Schainberg C, Sacco IC. Effectiveness of a long-term use of a minimalist footwear
versus habitual shoe on pain, function and mechanical loads in knee osteoarthritis:
a randomized controlled trial. BMC Musculoskelet Disord. 2012;13:121.
43.
Leardini A, Benedetti MG, Berti L, Bettinelli D, Nativo R, Giannini S. Rear-foot,
mid-foot and fore-foot motion during the stance phase of gait. Gait
Posture. 2007;25:453–62.
44.
Perl DP, Daoud AI, Lieberman DE. Effects of footwear and strike type on
running economy. Med Sci Sports Exerc. 2012;44:1335–43.
45.
Malisoux L, Ramesh J, Mann R, Seil R, Urhausen A, Theisen D. Can parallel
use of different running shoes decrease running-related injury risk? Scand J
Med Sci Sport. 2015;25:110–5.
46.
Faul F, Erdfelder E, Lang A-G, Buchner A. G*Power 3: a flexible statistical
power analysis program for the social, behavioral, and biomedical sciences.
Behav Res Methods. 2007;39:175–91.
47.
Haukoos JS, Newgard CD. Advanced Statistics: Missing Data in Clinical
Research-Part 1: An Introduction and Conceptual Framework. Acad Emerg
Med. 2007;14:662–8.
48.
Heiderscheit B. Always on the run. J Orthop Sports Phys Ther. 2014;44:724–6.
• We accept pre-submission inquiries
• Our selector tool helps you to find the most relevant journal
• We provide round the clock customer support
• Convenient online submission
• Thorough peer review
• Inclusion in PubMed and all major indexing services
• Maximum visibility for your research
Submit your manuscript at
www.biomedcentral.com/submit
Submit your next manuscript to BioMed Central
and we will help you at every step:
Matias et al. BMC Musculoskeletal Disorders (2016) 17:160
Page 11 of 11
| Protocol for evaluating the effects of a therapeutic foot exercise program on injury incidence, foot functionality and biomechanics in long-distance runners: a randomized controlled trial. | 04-14-2016 | Matias, Alessandra B,Taddei, Ulisses T,Duarte, Marcos,Sacco, Isabel C N | eng |
PMC9941527 | Estimated power output for a
distance run and maximal oxygen
uptake in young adults
Gen-Min Lin1,2*, Kun-Zhe Tsai 1,2,3, Xuemei Sui4 and Carl J. Lavie5*
1Department of Internal Medicine, Hualien Armed Forces General Hospital, Hualien City, Taiwan,
2Department of Medicine, Tri-Service General Hospital, National Defense Medical Center, Taipei, Taiwan,
3Department of Stomatology of Periodontology, Mackay Memorial Hospital, Taipei, Taiwan, 4Department of
Exercise Science, Arnold School of Public Health, University of South Carolina, Columbia, SC, United States,
5Ochsner Clinical School, John Ochsner Heart and Vascular Institute, The University of Queensland School of
Medicine, New Orleans, LA, United States
Background: Both cardiopulmonary exercise testing (CPET) and run field tests are
recommended by the American Heart Association for assessing the maximal oxygen
uptake (VO2 max) of youth. Power output was highly correlated with VO2 max in
CPET. However, it is unclear regarding the correlations of time and estimated power
output (EPO) for a run field test with VO2 max obtained from CPET in young adults.
Methods: This study included 45 participants, aged 20–40 years, from a sample of
1,120 military personnel who completed a 3,000-m run field test in Taiwan in 2020.
The participants subsequently received CPET using the Bruce protocol to assess VO2
max in the same year. According to the physics rule, EPO (watts) for the run field test
was defined as the product of half body mass (kg) and [distance (3000-m)/time (s) for
a run field test]. Pearson product–moment correlation analyses were performed.
Results: The Pearson correlation coefficient (r) of time against EPO for the run field
test was estimated to be 0.708 (p <0.001). The correlation coefficient between the
time for the run field test and VO2 max (L/min) in CPET was estimated to be 0.462 (p =
0.001). In contrast, the correlation coefficient between time for the run field test and
VO2 max scaled to body mass in CPET was estimated to be 0.729 (p <0.001). The
correlation coefficient of EPO for the run field test against VO2 max in CPET was
estimated to be 0.813 (p <0.001).
Conclusion: In young adults, although the time for a run field test was a reliable
estimate of VO2 max scaled to body mass, EPO proportional to the mean square
velocity was found as a superior estimate of VO2 max.
KEYWORDS
cardiopulmonary exercise testing, maximal oxygen uptake, estimated power output, run
field test, velocity
Introduction
Cardiorespiratory fitness (CRF) is inversely associated with the risk of metabolic syndrome,
diabetes mellitus, cardiovascular diseases (CVD), and mortality in the general population
(Carnethon et al., 2009; Mehta et al., 2020; Wang et al., 2021; Lin et al., 2022; Sui et al., 2022).
Obtaining greater CRF levels for sedentary individuals has been proposed as one of the major
preventive measures to reduce the severity and burden of atherosclerosis in developed countries
(Lavie et al., 2019; Sanchis-Gomar et al., 2022). The gold standard for CRF assessment is
maximal oxygen uptake (VO2 max) with or without an adjustment for body mass from
cardiopulmonary exercise testing (CPET). The main advantages of CPET include a strictly
OPEN ACCESS
EDITED BY
Elisabetta Salvioni,
Monzino Cardiology Center (IRCCS), Italy
REVIEWED BY
Maurizio Bussotti,
Scientific Clinical Institute Maugeri (ICS
Maugeri), Italy
Yaoshan Dun,
Xiangya Hospital, Central South University,
China
*CORRESPONDENCE
Gen-Min Lin,
[email protected]
Carl J. Lavie,
[email protected]
SPECIALTY SECTION
This article was submitted to Exercise
Physiology,
a section of the journal
Frontiers in Physiology
RECEIVED 05 December 2022
ACCEPTED 20 January 2023
PUBLISHED 07 February 2023
CITATION
Lin G-M, Tsai K-Z, Sui X and Lavie CJ
(2023), Estimated power output for a
distance run and maximal oxygen uptake
in young adults.
Front. Physiol. 14:1110802.
doi: 10.3389/fphys.2023.1110802
COPYRIGHT
© 2023 Lin, Tsai, Sui and Lavie. This is an
open-access article distributed under the
terms of the Creative Commons
Attribution License (CC BY). The use,
distribution or reproduction in other
forums is permitted, provided the original
author(s) and the copyright owner(s) are
credited and that the original publication in
this journal is cited, in accordance with
accepted academic practice. No use,
distribution or reproduction is permitted
which does not comply with these terms.
Frontiers in Physiology
frontiersin.org
01
TYPE Original Research
PUBLISHED 07 February 2023
DOI 10.3389/fphys.2023.1110802
controlled environment, e.g., a maintained indoor temperature,
atmospheric pressure, and calibration gas. However, the limitation
to CPET is facility-dependent, and examinees require a tolerance to
wear a mask during the graded exercise testing. Therefore, CPET may
not be feasible for some specific populations, such as children and the
elderly, and might not be practical at a large population level.
Based on the findings of previous studies, the American Heart
Association (AHA) has recommended VO2 max and some alternative
measures for the CRF levels of youth (Mayorga-Vega et al., 2015;
Raghuveer et al., 2020). For instance, the performance of a field-based
20-m shuttle run test or distance run test has a moderate-to-high
correlation with VO2 max obtained from CPET, whereas the
performance of a field-based step test or 6-minute walk test has
only a low-to-moderate correlation with VO2 max in children or
adolescents (Raghuveer et al., 2020).
In addition, previous studies also revealed that the peak power output
of the heart or body measured in CPET has a high correlation with VO2
max in patients with recovering heart failure and athletes (Hawley and
Noakes, 1992; Jakovljevic et al., 2011). Physiologically, most oxygen
uptake is translated to energy output during peak exercise. However,
to the best of our knowledge, there have been no studies with regard to the
correlation between estimated power output (EPO) for a run field test and
VO2 max in CPET. The correlation between a run test performance and
VO2 max in CPET was not clarified in adults and varied by study (Cooper,
1968; O’Gorman et al., 2000; Casajus and Castagna, 2007). The aim of this
study was to investigate the correlations of time and EPO for a distance
run field test with VO2 max measured from CPET in a group of young
military adults.
Materials and methods
Study population
The study included 45 participants, aged 20–40 years, without any
medication use, randomly selected from the cardiorespiratory fitness and
health in eastern armed forces (CHIEF) study participants (N = 1,120)
(Lin et al., 2021; Liu et al., 2021) in 2020, which aimed to carry out a
preliminary study on the correlation of time for a 3000-m run field test
with VO2 max from CPET. All participants received exercise training
daily, e.g., a 3,000-m run within 25 min in the morning at the military base
for over 6 months. A history of moderate physical activity, such as a
limited-time 3,000-m run in the morning per week in the past half year,
was obtained from each participant. Each participant underwent the
2020 annual health examinations for physical examinations (Hsu et al.,
2022) in the Hualien Armed Forces General Hospital of Taiwan. Each
participant also underwent the 2020 annual military exercise test for a
3,000-m run field test to assess endurance capacity. Within 2 weeks of the
3,000-run field test, the 45 volunteers were scheduled for CPET to
objectively assess VO2 max.
Anthropometric and blood pressure (BP)
measurements
Anthropometric parameters, i.e., body height and weight, were
measured in a standing position by a medical staff member in the
annual military exercise test. Body mass index was defined as the body
weight divided by body height squared (kg/m2). Overweight or obesity
was defined as a body mass index ≥27.5 kg/2 for Asians, according to
the recommendations of the World Health Organization (Jih et al.,
2014). The BP and pulse rate of each participant were automatically
measured using the same device (FT201, Parama-Tech Co., Ltd.,
Fukuoka, Japan), which utilized the oscillometric method (Lin
et al., 2016; Lin et al., 2020a; Lin et al., 2020b). The BP was
measured once over the right arm in a sitting position after resting
for longer than 15 min and was recorded by a medical staff member. If
the pulse rate was ≥90 beats per minute, systolic BP ≥140 mmHg, or
diastolic BP ≥90 mmHg was found, the participant would undergo
another two rounds of hemodynamic parameter measurements, which
were averaged as the final result.
The 3000-m run field test
The 3,000-m run field test was performed outdoors on a flat
playground of the Military Physical Training and Testing Center in
Hualien, Taiwan, at 16:00. Each participant wore sweat suits and did
not carry any additional objects. The whole running process of each
participant was video recorded and supervised by eight military sports
officers. The 3,000-m run field test was carried out if there was no heavy
rain, and the coefficient of the heat stroke risk formula, the product of
relative humidity (%) and outdoor temperature (Celsius scale) x 0.1, was
less than 40. The time for the 3,000-m run field test was utilized to evaluate
the endurance capacity of each participant. EPO for the run field test was
defined as “1/2 x body mass (kg) x square of mean velocity (m/s),” on the
basis of the physics rule of the kinetic energy theorem (Serway and Jewett,
2004). The mean velocity was calculated by the formula “distance (3,000-
m) divided by time (s) for the run field test.”
The performance of CPET
CPET was performed on a Trackmaster TMX-428 stress treadmill
(SCHILLER, Baar, Switzerland) using the standard Bruce protocol.
The same supervisor conducted all of the CPETs throughout the study.
All participants were asked not to consume caffeine or alcohol for 12 h
or longer before the CPET and exercised after a 2-h postprandial
period. The room for the CPET used an air conditioning system to
maintain a constant temperature of approximately 22 degrees Celsius.
Throughout the CPET, electrocardiography and BP were monitored.
The rates of oxygen uptake (VO2), production of carbon dioxide
(VCO2), tidal volume (Vt), end-tidal partial pressure of carbon dioxide
(PETCO2), and respiratory rate were recorded breath by breath using a
Cardiovit CS-200 Excellence Ergo-Spiro analytic system (SCHILLER,
Baar, Switzerland). VO2 max was defined as the average of VO2 during
the last minute of maximal exercise.
Statistical analysis
Characteristics of overall participants for a 3,000-m run field test and
those for both a 3,000-m run field test and CPET were presented as
numbers (%) for categorical variables and mean ± standard deviation for
continuous variables, respectively. Continuous variables were compared
using analysis of variance if the Kolmogorov–Smirnov test for the normal
distribution was met; otherwise, the Wilcoxon signed-rank test was used.
Categorical variables were compared using Fisher’s exact test. The Pearson
Frontiers in Physiology
frontiersin.org
02
Lin et al.
10.3389/fphys.2023.1110802
product–moment correlation coefficient was used to determine the
association strength of time and EPO for a 3,000-m run field test with
VO2 max scaled to body mass or not in CPET. The correlations of time
and EPO for a 3,000-m run field test were performed for a comparison
between selected participants for both a 3,000-m run field test and CPET
and a sample of age-, sex-, body mass index-, and BP-matched
participants from the overall study participants. Scatter plots between
time and EPO for the 3,000-m run field test and VO2 max scaled for body
mass or not in CPET were obtained. Internal validation was performed for
those whose body mass index <27.5 kg/m2. A value of p <0.05 was
considered significant. All analyses were carried out using SPSS version
25.0 for Windows (IBM Corp., Armonk, NY, United States). This study
has been approved by the Institutional Review Board of the Clinical Ethics
Committee of the Mennonite Christian Hospital (No. 16-05-008),
Hualien City, Eastern Taiwan, R.O.C., and written informed consent
was obtained from all participants.
Results
Clinical characteristics of the participants
Table 1 shows the clinical characteristics of the participants for a
3,000-m run field test (N = 1,120) and those for both a 3,000-m run field
test and CPET (N = 45). The characteristics of the selected participants for
CPET were similar to the original overall sample, except greater age, pulse
rate, and BP levels were observed in participants for CPET. The mean age
of the participants for CPET was approximately 30 years old, and over
90% of the participants were males. Since only one woman was included
for the CPET, the characteristics of men were also compared between the
original group (N = 911) and the selected group (N = 44), and the results
are provided in Supplementary Table S1, which show consistent findings.
Correlations between time and EPO for a
3000-m run field test
The correlation coefficient (r) of time against EPO for a 3,000-m
run field test was estimated to be 0.708 (p <0.001) in participants for
both a 3,000-m run field test and CPET (Figure 1A), which was close to
the correlation coefficient (r = 0.703 and p <0.001) in the age-, sex-,
body mass index-, and BP-matched samples of 707 participants for a
3000-m run test (Figure 1B). The characteristics of the variable-
matched population are provided in Supplementary Table S2. The
correlation coefficients for men only were in line with the main
findings, and the results are provided in Supplementary Figure S1.
Correlations of time for a 3000-m run field
test against VO2 max in CPET
The correlation coefficient of time for a 3000-m run field test
against VO2 max (L/min) in CPET was estimated to be 0.462 (p =
0.001) (Figure 2A). In contrast, the correlation coefficient between
time for a 3000-m run field test and VO2 max scaled to body mass (kg)
in CPET was estimated to be 0.729 (p <0.001) (Figure 2B). The
correlation coefficients for men only were in line with the main
findings, and the results are provided in Supplementary Figure S2.
Correlations of EPO for a 3000-m run
field test against VO2 max in CPET
The correlation coefficient of EPO for a 3,000-m run field test against
VO2 max (L/min) in CPET was estimated to be 0.813 (p <0.001)
(Figure 3A). However, the correlation coefficient between EPO for a
3,000-m run field test and VO2 max scaled to body mass (kg) in CPET
was estimated to be only 0.364 (p <0.001) (Figure 3B). The correlation
coefficients for men only were in line with the main findings, and the
results are provided in Supplementary Figure S3.
Internal validation for non-obese participants
The results of interval validation for participants with body mass
index <27.5 kg/m2 (N = 35) are revealed in Figure 4. The correlation
coefficient of time for a 3,000-m run field test against VO2 max (L/min)
was 0.453 (p = 0.006) (Figure 4A), and the correlation coefficient of time
for a 3,000-m run field test with VO2 max scaled to body mass (kg) was
0.485 (p = 0.003) (Figure 4B). The correlation coefficient of EPO for a
3,000-m run field test with VO2 max (L/min) was 0.757 (p <0.001)
(Figure 4C), and the correlation coefficient of EPO for a 3,000-m run field
FIGURE 1
(A) Correlation coefficient (r) of time against EPO for a 3,000-m run field test, estimated to be 0.708 (p <0.001) in participants for both a 3,000-m run field
test and CPET. (B) Correlation coefficient estimated to be 0.703 (p <0.001) in the variable-matched participants for a 3,000-m run field test.
Frontiers in Physiology
frontiersin.org
03
Lin et al.
10.3389/fphys.2023.1110802
TABLE 1 Clinical Characteristics of the Overall Participants for a Run Field Test and the Selected Participants for a Cardiopulmonary Exercise Testing.
N = 45
N = 1120
p-value
Sex, males (%)
44 (97.8)
911 (81.3)
0.07
Age, years
29.93 ± 7.05
27.61 ± 5.87
0.01
Body height, cm
170.74 ± 6.47
170.80 ± 6.67
0.93
Body mass, kg
72.74 ± 11.81
71.93 ± 12.21
0.66
Body mass index, kg/m2
24.95 ± 3.86
24.58 ± 3.54
0.49
Pulse rate, beats per min
77.50 ± 11.18
67.09 ± 10.95
<0.001
Systolic blood pressure, mmHg
127.68 ± 11.83
116.97 ± 12.83
<0.001
Diastolic blood pressure, mmHg
80.32 ± 10.39
69.09 ± 9.88
<0.001
Time for a 3000-m run, secs
876.62 ± 94.34
893.55 ± 106.19
0.29
EPO for a 3000-m run, watts
437.44 ± 105.22
421.12 ± 114.18
0.34
Moderate activity per week
100-150 minutes
8 (17.8)
224 (20.0)
0.92
150-300 minutes
18 (40.0)
423 (37.8)
>300 minutes
19 (42.2)
473 (42.2)
Abbreviation: EPO, estimated power output.
EPO was defined as 1/2 x body mass (kg) x (3000-m/time for a 3000-m run test)2.
FIGURE 2
In 45 participants, for both a 3,000-m run field test and CPET: (A) correlation coefficient of time for a 3,000-m run field test against VO2 max (L/min) was
0.462 (p = 0.001); (B) correlation coefficient of time for a 3,000-m run field test with VO2 max scaled to body mass (kg) was 0.729 (p <0.001).
FIGURE 3
In 45 participants, for both a 3,000-m run field test and CPET: (A) correlation coefficient of EPO for a 3000-m run field test against VO2 max (L/min) was
0.813 (p <0.001); (B) correlation coefficient between EPO for a 3,000-m run field test and VO2 max scaled to body mass (kg) was estimated to be 0.364
(p <0.001).
Frontiers in Physiology
frontiersin.org
04
Lin et al.
10.3389/fphys.2023.1110802
test against VO2 max scaled to body mass (kg) was 0.349 (p = 0.04)
(Figure 4D). Although the correlation coefficients in the sample for
interval validation were all lower than that in the overall sample (N =
45) receiving the CPET, all of the associations of time and EPO for a
3,000-m run with VO2 max in a CPET were significant, and the EPO
association remained with the greatest strength.
Estimation of VO2 max in CPET from time and
EPO for a run field test
Based on the regression line in Figure 2B, formula 1, with regard to
time for a run field test to estimate VO2 max scaled to body mass, can
be derived as follows:
Formula 1
Y = −0.0386X + 67.151
X: Time for a 3,000-m field run (s)
Y: VO2 max scaled to body mass (mL/min/kg)
Based on the regression line in Figure 3A, formula 2, with regard to
EPO for a run field test to estimate VO2 max, can be derived as follows:
Formula 2
Y = 3.7943X + 757.6
X: EPO for a 3000-m field run (watts)
Y: VO2 max (mL/min)
Discussion
The principal finding of this study was that in young adults, the time
for a run field test can be an acceptable estimate of VO2 max scaled to
body mass obtained in CPET. In addition, the EPO for a run field test,
calculated according to the kinetic energy theorem, can be a more precise
estimate of VO2 max (L/min) than the time for a run field test.
Numerous studies have investigated the correlation between distance-
or time-based run field test performance and VO2 max in CPET in adults
(Mayorga-Vega et al., 2016). However, the correlation coefficients were
distributed widely, ranging from 0.60 to 0.90, among various run field
tests (Cooper, 1968). In addition, no consensus has been reached in
previous studies to unify the VO2 max unit, with or without an
adjustment
for
body
mass,
when
analyzing
the
correlation.
Furthermore, the correlation coefficients might vary even in the same
run field test among different studies (Cooper, 1968; O’Gorman et al.,
2000; Casajus and Castagna, 2007). For instance, Cooper found a high
correlation between running distance and VO2 max scaled to body mass
in a 12-min run field test in a sample of military males, in whom the
correlation coefficient was estimated to be 0.897 (Cooper, 1968). In the
O’Gorman et al. (2000) study, there was a moderate correlation in a 12-
min run field test and in a 3,000-m run field test using a sample of
physically fit young males, in whom the correlation coefficient was
estimated equally to be 0.67 and −0.67. In contrast, the Casajus and
Castagna study revealed a relatively lower correlation in a 12-min run field
test in a sample of elite soccer players, in whom the correlation coefficient
was estimated to be only 0.46 (Casajus and Castagna, 2007). In the present
study, we demonstrated a moderate correlation of time for a 3,000-m run
field test with VO2 max scaled to body mass, whereas the results showed a
relatively lower correlation with VO2 max, which was not scaled to body
mass. These findings are in line with previous studies and a meta-analysis
(Serway and Jewett, 2004) made by Mayorga-Vega et al., which showed
the correlation coefficient between time for a 3,000-m run field test and
VO2 max scaled to a body mass of 12 studies, including 951 young adults,
was estimated to be 0.70. We further highlighted the importance of the
VO2 max unit for examining the correlation. It is reasonable that the
FIGURE 4
In 35 participants with body mass index <27.5 kg/m2, for both a 3000-m run field test and CPET: (A) correlation coefficient of time for a 3000-m run field
test against VO2 max (L/min) was estimated to be 0.453 (p = 0.006); (B) correlation coefficient of time for a 3,000-m run field test against VO2 max scaled to
body mass (kg) was estimated to be 0.485 (p = 0.003); (C) correlation coefficient of EPO for a 3,000-m run field test with VO2 max (L/min) was 0.757
(p <0.001). (D) correlation coefficient between EPO for a 3,000-m run field test and VO2 max (L/min) scaled to body mass was 0.349 (p = 0.04).
Frontiers in Physiology
frontiersin.org
05
Lin et al.
10.3389/fphys.2023.1110802
examinees’ running velocity in run field tests was inversely related to their
body mass. The correlation between the running velocity in a run field test
and VO2 max scaled to body mass in CPET would theoretically result in
the optimal result, except that the examinees had a similar level of body
mass at baseline.
Some reports have shown a moderate-to-high correlation between
peak cardiac power output (Watts) and VO2 max (L/min) not scaled
to body mass in CPET in patients with heart failure, in whom the
greatest correlation coefficient was 0.85, observed in those with
recovering heart function (Jakovljevic et al., 2011), and a high
correlation of peak power output with VO2 max in cycling athletes
(correlation coefficient greater than 0.90) (Hawley and Noakes, 1992).
The present study is the first report using EPO for a run field test to
estimate VO2 max in CPET in young adults. Oxygen consumption
generates energy, heat, and waste. Accordingly, it is apt to use peak
exercise power output to assess VO2 max not adjusted for body mass
in CPET for adults. In the present study, EPO was calculated according
to the kinetic energy theorem and proportional to the square of the
mean velocity in a 3,000-m run field test. The use of the square of the
mean velocity was better than the mean velocity in the run field test to
correlate with VO2 max in CPET in adults. This finding is consistent
with a previous study on CPET, where peak power output was superior
to the cycling speed to estimate VO2 max in athletes (Hawley and
Noakes, 1992).
The present study, however, has a few limitations. First, the limited
number of enrolled subjects is the major limitation to this preliminary
study, and further study is required to expand the sample size to obtain
greater power. Second, this study included only one woman and lacked
multi-ethnic/racial diversity, making generalization of the results
difficult. Third, although the annual military exercise test was held
with some restrictions of the weather, there could have been a bias for
differences in outdoor temperature and humidity, which may affect
the running performance and EPO estimation. Fourth, as the study
included merely healthy subjects, the results may not be appropriately
applied to those with a mismatch between heart and lung functions.
External validation should be performed in further study. Finally,
since we chose mean running velocity in the kinetic formula, the
maximum actual power output during the run test might be
underestimated, possibly leading to a bias for the correlation with
VO2 max.
Conclusion
There have been no recommendations from the AHA regarding the
role of time and EPO for a run field test to evaluate VO2 max in adults,
and the VO2 max unit was not emphasized. Our findings suggest that in
young adults, although the time for a distance run field test was an
acceptable estimate of VO2 max scaled to body mass, EPO proportional to
the square of the mean velocity in a run field test was found as a superior
estimate of VO2 max than the time for a run field test in this population.
Further studies are needed involving young women.
Data availability statement
The raw data supporting the conclusions of this article will be
made available by the authors, without undue reservation.
Ethics statement
The studies involving human participants were reviewed and
approved by the Institutional Review Board of the Clinical Ethics
Committee of the Mennonite Christian Hospital (No 16-05-008)
Hualien City, Eastern Taiwan. The patients/participants provided
their written informed consent to participate in this study.
Author contributions
GL collected and interpreted the data and wrote the manuscript;
KT analyzed the data; XS and CL raised critical comments and edited
the text; GL was the principal investigator for the administration of the
study.
Funding
The present study was supported by grants from the Hualien Armed
Forces General Hospital (HAFGH-D-112004) and the Medical Affairs
Bureau, Ministry of National Defense (MND-MAB-D-112182).
Conflict of interest
The authors declare that the research was conducted in the
absence of any commercial or financial relationships that could be
construed as a potential conflict of interest.
Publisher’s note
All claims expressed in this article are solely those of the
authors and do not necessarily represent those of their affiliated
organizations, or those of the publisher, the editors, and the
reviewers. Any product that may be evaluated in this article, or
claim that may be made by its manufacturer, is not guaranteed or
endorsed by the publisher.
Supplementary material
The Supplementary Material for this article can be found online at:
https://www.frontiersin.org/articles/10.3389/fphys.2023.1110802/
full#supplementary-material
Frontiers in Physiology
frontiersin.org
06
Lin et al.
10.3389/fphys.2023.1110802
References
Carnethon, M. R., Sternfeld, B., Schreiner, P. J., Jacobs, D. R., Jr, Lewis, C. E., Liu, K., et al.
(2009). Association of 20-year changes in cardiorespiratory fitness with incident type
2 diabetes: The coronary artery risk development in young adults (CARDIA) fitness study.
Diabetes Care 32 (7), 1284–1288. doi:10.2337/dc08-1971
Casajus, J. A., and Castagna, C. (2007). Aerobic fitness and field test performance in elite
Spanish soccer referees of different ages. J. Sci. Med. Sport 10 (6), 382–389. doi:10.1016/j.
jsams.2006.08.004
Cooper, K. H. (1968). A means of assessing maximal oxygen intake. Correlation between field
and treadmill testing. JAMA 203 (3), 201–204. doi:10.1001/jama.1968.03140030033008
Hawley, J. A., and Noakes, T. D. (1992). Peak power output predicts maximal oxygen
uptake and performance time in trained cyclists. Eur. J. Appl. Physiol. Occup. Physiol. 65
(1), 79–83. doi:10.1007/BF01466278
Hsu, C. Y., Liu, P. Y., Liu, S. H., Kwon, Y., Lavie, C. J., and Lin, G. M. (2022). Machine learning
for electrocardiographic features to identify left atrial enlargement in young adults: CHIEF heart
study. Front. Cardiovasc Med. 9, 840585. doi:10.3389/fcvm.2022.840585
Jakovljevic, D. G., Birks, E. J., George, R. S., Trenell, M. I., Seferovic, P. M., Yacoub, M.
H., et al. (2011). Relationship between peak cardiac pumping capability and selected
exercise-derived prognostic indicators in patients treated with left ventricular assist
devices. Eur. J. Heart Fail 13 (9), 992–999. doi:10.1093/eurjhf/hfr069
Jih, J., Mukherjea, A., Vittinghoff, E., Nguyen, T. T., Tsoh, J. Y., Fukuoka, Y., et al. (2014).
Using appropriate body mass index cut points for overweight and obesity among Asian
Americans. Prev. Med. 65, 1–6. doi:10.1016/j.ypmed.2014.04.010
Lavie, C. J., Ozemek, C., Carbone, S., Katzmarzyk, P. T., and Blair, S. N. (2019).
Sedentary behavior, exercise, and cardiovascular health. Circ. Res. 124 (5), 799–815. doi:10.
1161/CIRCRESAHA.118.312669
Lin, G. M., Li, Y. H., Lee, C. J., Shiang, J. C., Lin, K. H., Chen, K. W., et al. (2016).
Rationale and design of the cardiorespiratory fitness and hospitalization events in armed
forces study in Eastern Taiwan. World J. Cardiol. 8 (8), 464–471. doi:10.4330/wjc.v8.i8.464
Lin, G. M., Liu, P. Y., Tsai, K. Z., Lin, Y. K., Huang, W. C., and Lavie, C. J. (2022).
Cardiorespiratory fitness and carotid intima-media thickness in physically active young
adults: CHIEF atherosclerosis study. J. Clin. Med. 11 (13), 3653. doi:10.3390/jcm11133653
Lin, G. M., Tsai, K. Z., Lin, C. S., and Han, C. L. (2020). Physical fitness and long-term
blood pressure variability in young male military personnel. Curr. Hypertens. Rev. 16 (2),
156–160. doi:10.2174/1573402115666191023111351
Lin, Y. K., Tsai, K. Z., Han, C. L., Lin, Y. P., Lee, J. T., and Lin, G. M. (2021). Obesity
phenotypes and electrocardiographic characteristics in physically active males: CHIEF
study. Front. Cardiovasc Med. 8, 738575. doi:10.3389/fcvm.2021.738575
Lin, Y. P., Fan, C. H., Tsai, K. Z., Lin, K. H., Han, C. L., and Lin, G. M. (2020).
Psychological stress and long-term blood pressure variability of military young males: The
cardiorespiratory fitness and hospitalization events in armed forces study. World
J. Cardiol. 12 (12), 626–633. doi:10.4330/wjc.v12.i12.626
Liu, P. Y., Tsai, K. Z., Lima, J. A. C., Lavie, C. J., and Lin, G. M. (2021). Athlete’s heart in
asian military males: The CHIEF heart study. Front. Cardiovasc Med. 8, 725852. doi:10.
3389/fcvm.2021.725852
Mayorga-Vega, D., Aguilar-Soto, P., and Viciana, J. (2015). Criterion-related validity of
the 20-M shuttle run test for estimating cardiorespiratory fitness: A meta-analysis. J. Sports
Sci. Med. 14 (3), 536–547.
Mayorga-Vega, D., Bocanegra-Parrilla, R., Ornelas, M., and Viciana, J. (2016). Criterion-
related validity of the distance- and time-based walk/run field tests for estimating
cardiorespiratory fitness: A systematic review and meta-analysis. PLoS One 11,
e0151671. doi:10.1371/journal.pone.0151671
Mehta, A., Kondamudi, N., Laukkanen, J. A., Wisloff, U., Franklin, B. A., Arena, R.,
et al. (2020). Running away from cardiovascular disease at the right speed: The impact of
aerobic physical activity and cardiorespiratory fitness on cardiovascular disease risk and
associated subclinical phenotypes. Prog. Cardiovasc Dis. 63 (6), 762–774. doi:10.1016/j.pcad.
2020.11.004
O’Gorman, D., Hunter, A., MacDonnacha, C., and Kirwan, J. P. (2000). Validity of field
tests for evaluating endurance capacity in competitive and international-level sports
participants. J. Strength Cond. Res. 14 (1), 62–67. doi:10.1519/00124278-200002000-00011
Raghuveer, G., Hartz, J., Lubans, D. R., Takken, T., Wiltz, J. L., Mietus-Snyder, M., et al.
(2020). Cardiorespiratory fitness in youth: An important marker of health: A scientific
statement from the American heart association. Circulation 142 (7), e101–e118. doi:10.
1161/CIR.0000000000000866
Sanchis-Gomar, F., Lavie, C. J., Marín, J., Perez-Quilis, C., Eijsvogels, T. M. H., O’Keefe,
J. H., et al. (2022). Exercise effects on cardiovascular disease: From basic aspects to clinical
evidence. Cardiovasc Res. 118 (10), 2253–2266. doi:10.1093/cvr/cvab272
Serway, R., and Jewett, J. W. (2004). Physics for scientists and engineers, 6th edition
Books/Cole. 0-534-40842-7.
Sui, X., Sarzynski, M. A., Gribben, N., Zhang, J., and Lavie, C. J. (2022).
Cardiorespiratory fitness and the risk of all-cause, cardiovascular and cancer mortality
in men with hypercholesterolemia. J. Clin. Med. 11 (17), 5211. doi:10.3390/jcm11175211
Wang, S. H., Chung, P. S., Lin, Y. P., Tsai, K. Z., Lin, S. C., Fan, C. H., et al. (2021).
Metabolically healthy obesity and physical fitness in military males in the CHIEF study.
Sci. Rep. 11 (1), 9088. doi:10.1038/s41598-021-88728-0
Frontiers in Physiology
frontiersin.org
07
Lin et al.
10.3389/fphys.2023.1110802
| Estimated power output for a distance run and maximal oxygen uptake in young adults. | 02-07-2023 | Lin, Gen-Min,Tsai, Kun-Zhe,Sui, Xuemei,Lavie, Carl J | eng |
PMC5133111 | nutrients
Article
Cardiorespiratory Fitness and Peak Torque
Differences between Vegetarian and Omnivore
Endurance Athletes: A Cross-Sectional Study
Heidi M. Lynch *, Christopher M. Wharton and Carol S. Johnston
Arizona State University, School of Nutrition and Health Promotion, Phoenix, AZ 85004, USA;
[email protected] (C.M.W.); [email protected] (C.S.J.)
* Correspondence: [email protected]; Tel.: +1-847-828-1332
Received: 1 September 2016; Accepted: 10 November 2016; Published: 15 November 2016
Abstract: In spite of well-documented health benefits of vegetarian diets, less is known regarding
the effects of these diets on athletic performance. In this cross-sectional study, we compared elite
vegetarian and omnivore adult endurance athletes for maximal oxygen uptake (VO2 max) and
strength. Twenty-seven vegetarian (VEG) and 43 omnivore (OMN) athletes were evaluated using
VO2 max testing on the treadmill, and strength assessment using a dynamometer to determine peak
torque for leg extensions. Dietary data were assessed using detailed seven-day food logs. Although
total protein intake was lower among vegetarians in comparison to omnivores, protein intake as
a function of body mass did not differ by group (1.2 ± 0.3 and 1.4 ± 0.5 g/kg body mass for VEG
and OMN respectively, p = 0.220). VO2 max differed for females by diet group (53.0 ± 6.9 and
47.1 ± 8.6 mL/kg/min for VEG and OMN respectively, p < 0.05) but not for males (62.6 ± 15.4 and
55.7 ± 8.4 mL/kg/min respectively). Peak torque did not differ significantly between diet groups.
Results from this study indicate that vegetarian endurance athletes’ cardiorespiratory fitness was
greater than that for their omnivorous counterparts, but that peak torque did not differ between diet
groups. These data suggest that vegetarian diets do not compromise performance outcomes and may
facilitate aerobic capacity in athletes.
Keywords: vegetarian; endurance; VO2 max; dynamometer; protein; sustainability; torque; body
composition; Dual X-ray Absorptiometry (DXA)
1. Introduction
Vegetarian diets are increasingly being adopted for a variety of reasons including health,
sustainability, and ethics-related concerns. Adherence to a vegetarian diet has been associated with a
reduced risk of developing coronary heart disease [1], breast cancer [2], colorectal cancers [3], prostate
cancer [4], type 2 diabetes [5], insulin resistance [6], hypertension [7], cataracts [8] and dementia [9].
Vegetarians also typically have a lower body mass index (BMI) [10] and an improved lipid profile [11].
In addition to promoting physical health, reducing or eliminating meat from the diet is environmentally
advantageous since producing meat requires more land, water, and energy resources than growing
plants for food [12], and producing meat creates more greenhouse gases compared to a plant-based
diet [13,14].
In spite of the many health aspects of vegetarian diets some concern has been raised pertaining to
the nutrient adequacy of vegetarian diets for supporting athletic performance. Vegetarian diets are
typically lower in vitamin B12, protein, creatine, and carnitine [15,16], and iron and zinc from plant
sources are less bioavailable than from meat sources [17]. However, vegetarian diets are typically
higher in carbohydrate and antioxidants [18,19], which may be advantageous for athletic performance,
particularly for endurance activities [20].
Nutrients 2016, 8, 726; doi:10.3390/nu8110726
www.mdpi.com/journal/nutrients
Nutrients 2016, 8, 726
2 of 11
Despite these issues, little research directly examining vegetarian diets and athletic performance is
available. There have been mixed results regarding hypertrophic potential when comparing vegetarian
diets with omnivore diets during resistive exercise training; however, in all cases these differences
did not translate to differential strength gains at the completion of the trials [21–24]. Adoption of
a lacto-ovo vegetarian (LOV) diet for six weeks did not significantly affect endurance performance
among a group of trained, male endurance athletes, in spite of a decrease in total testosterone while on
the vegetarian diet [25]. There were also no group differences between 20 participants adopting an
LOV diet compared to maintaining their usual omnivorous diet in terms of muscle buffering capacity
in conjunction with sprint training for five weeks [26]. These studies provide some insight into the
effect of a vegetarian diet on athletic performance. However, a considerable limitation in many of
these studies is the inclusion of participants who typically consume meat but subsequently adopt
a vegetarian diet only for the duration of the study rather than comparing participants who have
adhered to a vegetarian or meat-containing diet long-term.
In a 1986 observational trial, Hanne and colleagues compared athletes who had maintained either
an LOV or omnivore diet for at least two years and found no group differences for aerobic or anaerobic
capacity [27]. However, aerobic capacity was estimated using cycle ergometry and predicted VO2
max, and strength or torque were not measured. Moreover, body adiposity was estimated using
skinfold thickness. Given the current interest in vegetarian diets, in terms of both long-term health
and environmental benefits, it is important to reaffirm, using leading-edge technology, that high-level
athletic performance is supported by these diets.
The purpose of the present cross-sectional study was to examine body composition and
performance measures in vegetarian and omnivore adult endurance athletes who had adhered to
their respective diet plans for at least three months. Body composition, including visceral adiposity,
was measured using dual-energy X-ray absorptiometry (DXA), leg strength was measured using a
dynamometer, and aerobic capacity was determined using the Bruce protocol treadmill test. It was
hypothesized that there would be no differences between groups on any parameters.
2. Materials and Methods
2.1. Participant Recruitment
Healthy men and women,
both vegetarians and omnivores,
were recruited through
advertisements on Stevebay.org (a popular website for endurance athletes), Facebook, and through
word of mouth. Participants were either on a competitive club sports team at a National Collegiate
Athletic Association (NCAA) Division 1 university or training for a major endurance race (such as
a marathon, triathlon, cycling race, or other ultra-endurance event). An equal number of omnivore
and vegetarian athletes were enrolled in the study between the ages of 21–58 years (35 per group);
however, answers to diet questions indicated that eight of the vegetarians ate meat on occasion, and
these subjects were reclassified as omnivores. Participants completed a health history questionnaire
and were excluded if they had any chronic disease. All participants had the study verbally explained to
them and provided their written consent; this study was approved by the Institutional Review Board
at Arizona State University, number HS1211008557. Study recruitment and all study measurements
took place between August and November 2015.
2.2. Experimental Approach
In this cross-sectional investigation participants completed all study measurements in a single
visit. Prior to the visit, participants completed a seven-day food log. Fifty-seven out of seventy
participants returned completed food logs, all of which were used in dietary analysis using Food
Processor SQL Nutrition and Fitness Software by ESHA Research, Inc. (version 10.11.0, Salem, OR,
USA). Height and body mass were measured using a SECA directprint 284 digital measuring station
when participants were wearing light clothing and no shoes. Participants also completed a full-body
Nutrients 2016, 8, 726
3 of 11
DXA scan (Lunar iDXA, General Electric Company, East Cleavland, OH, USA), which was conducted
by a certified radiology technologist.
Maximal oxygen uptake was determined by following the Bruce protocol [28] on a Trackmaster
TMX425C treadmill using the Parvo Medics TrueOne 2400 (Sandy, UT, USA) metabolic measurement
system. Prior to beginning the test, participants were instructed how to report their fatigue level
using the Borg rating of perceived exertion (RPE) scale [29]. When asked by a research assistant, they
reported their RPE at the end of each minute of the test by pointing to a printed Borg RPE chart being
held by a research assistant. Participants were verbally encouraged by the research team to push as
long as they could and to try to reach a true maximal effort. Handrail support was not allowed during
the test. Maximal respiratory exchange ratio (RER) was recorded to help determine whether subjects
had reached a “true” maximal effort during the test. Maximal RER values of ≥1.1 were considered
indicative of true maximal oxygen uptake [30,31]. Peak oxygen uptake reported is the highest oxygen
uptake measured during the test.
Finally, participants completed a series of leg extensions and flexions on the HumacNorm
isokinetic dynamometer (Computer Sports Medicine Inc. (CSMi, Stoughton, MA, USA) at 60 degrees
per second (d/s), 180 d/s, and 240 d/s. Participants were familiarized with the protocol and conducted
one practice repetition at each speed prior to performing three maximal effort repetitions at each speed.
All sets, including practice repetitions, were performed on both legs, and self-reported dominant side
was recorded. Participants moved from the VO2 max test immediately into the dynamometer testing,
and there were 30 s of rest between sets on the dynamometer.
2.3. Statistical Analyses
Based on the data of Hanne et al. [27], at 80% power and an alpha level of 5%, 15 participants per
group would be needed to detect a 10% difference in strength and 80 participants per group would be
needed to detect a 10% change in aerobic capacity between groups. Data were analyzed for normality
and log transformed if necessary, and outliers (values > 3 standard deviations (SD) from the mean)
were removed prior to data analyses. Data reported are the mean ± SD, and participant characteristics
are displayed by gender and diet group. A 2-way analysis of variance (ANOVA) analysis was used to
determine differences between diet groups for participant characteristics followed by an independent
t-test for post-hoc examination by diet within gender if indicated. Dietary data are reported by group,
and a general linear model analysis was used to examine differences between groups controlling
for gender. Data were analyzed using the Statistical Package for Social Sciences (SPSS) 23.0 for Mac
(SPSS, Inc., Chicago, IL, USA).
3. Results
In the vegetarian group, 24 of the 27 participants (89%) had adhered to a vegetarian diet
for >2 years. Of the remaining three participants, the diet had been followed for three, six, or eleven
months. Fifteen of the vegetarians were vegans (nine men and six women), and twelve were lacto-ovo
vegetarians (five men and seven women).
There were no significant age or gender differences between groups (Table 1).
Significant
differences were noted between diet groups for body mass and for lean body mass (LBM): female
vegetarians tended to have a lower total body mass and LBM compared to the female omnivores
(−11% and −7% respectively). Adiposity, however, did not differ between diet groups. Physical
activity levels, recorded as kcal·kg−1·week−1, were 20% higher for vegetarians compared to omnivores
(p = 0.018) (Table 1). Maximal oxygen uptake (mL/kg/min) differed significantly between diet groups,
and post-hoc analyses revealed a significantly greater aerobic capacity in the female vegetarians in
comparison to the female omnivores (+13%, p < 0.05) (Table 1); however, absolute maximal oxygen
uptake (L/min) did not differ between diet groups. Peak torque when doing leg extensions was
not different between diet groups. The 7-day diet records revealed several differences in nutrient
intake between diet groups. Although total energy intakes were similar between the diet groups,
Nutrients 2016, 8, 726
4 of 11
the vegetarians consumed more carbohydrate, fiber, and iron daily compared to omnivores (Table 2).
However, daily intakes for protein, saturated fat, cholesterol, vitamin B12, and selenium were lower
among the vegetarians in comparison to the omnivores.
Table 1. Participant characteristics by diet group (vegetarian, VEG; omnivorous, OMN) 1.
VEG
OMN
p
Measure
Male (14)
Female (13)
Male (26)
Female (17)
Age, year
36.1 ± 10.2
36.7 ± 7.7
38.0 ± 10.0
37.1 ± 8.7
0.608
Body mass, kg
73.3 ± 14.8
58.3 ± 7.6 **
78.0 ± 11.0
65.4 ± 11.6
0.043
BMI, kg/m2
24.0 ± 4.4
21.8 ± 2.5
24.8 ± 2.6
23.5 ± 3.8
0.123
Lean mass, kg
56.3 ± 7.4
42.0 ± 4.9 **
60.2 ± 7.3
45.4 ± 5.1
0.026
Waist, cm
81.6 ± 10.7
69.0 ± 14.8
85.2 ± 7.4
73.8 ± 8.2
0.093
Body fat, %
19.2 ± 6.5
25.5 ± 4.2
19.2 ± 6.4
26.9 ± 8.1
0.659
Visceral fat, cm3
447.4 ± 419.8
110.4 ± 123.0
538.5 ± 404.3
206.4 ± 254.6
0.656
METS, kcal·kg−1·week−1
108.8 ± 32.9
106.1 ± 36.6 **
91.7 ± 33.2
85.6 ± 20.8
0.018
VO2 max, mL/kg/min
62.6 ± 15.4
53.0 ± 6.9 *
55.7 ± 8.4
47.1 ± 8.6
0.011
VO2 max, L/min
4.44 ± 0.81
3.21 ± 0.67
4.29 ± 0.59
3.03 ± 0.49
0.295
Peak torque, ft-lbs
114.4 ± 26.2
65.5 ± 12.8
124.2 ± 24.5
73.6 ± 18.6
0.104
1 Data are the mean ± SD; n in parentheses; gender distribution did not differ by diet group (p = 0.460;
Chi Square analysis). p for 2-way ANOVA analyses by diet (non-normal data transformed prior to analysis
(visceral fat)). The single asterisk (*) indicates significant difference within gender by diet group (p < 0.05);
the double asterisk (**) indicates a trend for difference within gender by diet group (0.05 < p < 0.10).
Table 2. Nutrient differences by diet group (vegetarian, VEG; omnivorous, OMN) 1.
VEG (22)
OMN (35)
p
Reference Range 2
Total kilocalories (kcal)
2443 ± 535
2266 ± 612
0.072
-
Carbohydrate (CHO) (g)
328 ± 70
248 ± 101
0.001
-
CHO (% energy)
53 ± 6
48 ± 7
0.010
45%–65%
Fiber (g)
38 ± 13
24 ± 9
<0.001
38/25 g [M/F]
Protein (g)
78 ± 19
101 ± 35
0.006
-
Protein (% energy)
12 ± 2
17 ± 4
<0.001
10%–35%
Protein (g/kg body mass)
1.2 ± 0.3
1.4 ± 0.5
0.220
0.8 g/kg
Fat (g)
90 ± 26
83 ± 33
0.901
-
Fat (% energy)
32 ± 5
32 ± 6
0.952
20%–35%
Saturated fat (g)
22.8 ± 11.2
25.7 ± 10.1
0.207
-
Saturated fat (% energy)
8.3 ± 3.1
11.6 ± 6.3
0.002
<10%
Cholesterol (mg)
102.8 ± 119.5
301.2 ± 165.6
<0.001
-
Vitamin C (mg)
117.0 ± 64.0
83.0 ± 46.5
0.076
90/75 mg [M/F]
Vitamin D (IU)
115.4 ± 111.4
129.0 ± 115.5
0.201
600 IU
Vitamin B12 (mcg)
3.0 ± 3
4.8 ± 4.6
0.006
2.4 mcg
Selenium (mcg)
41.8 ± 36.0
62.6 ± 33.6
0.002
55 mcg
Sodium (mg)
2931.2 ± 783.1
2972.8 ± 887.5
0.794
<2300 mg
Iron (mg)
19.4 ± 7.8
15.4 ± 5.4
0.017
8/18 mg [M/F]
Zinc (mg)
8.5 ± 9.1
8.9 ± 4.9
0.149
11/8 mg [M/F]
Calcium (mg)
971.0 ± 401.6
878.1 ± 314.9
0.378
1000 mg
Phosphorus (mg)
782.0 ± 378.0
831.2 ± 336.4
0.507
700 mg
Omega-3 fatty acid (g)
1.6 ± 2.5
0.9 ± 0.7
0.326
-
Omega-3 fatty acid (% energy)
0.004 ± 0.005
0.004 ± 0.003
0.613
0.6%–1.2%
Omega-6 fatty acid (g)
7.7 ± 5.4
6.1 ± 4.4
0.145
-
Omega-6 fatty acid (% energy)
2.8 ± 1.6
2.4 ± 1.3
0.358
5%–10%
1 Data are the mean ± SD; sample size in parentheses. p for general linear model analyses (non-normal data
transformed prior to analysis (all variables except carbohydrate variables and fat percentage) and 2 outliers
(VEG group) removed prior to analysis for saturated fat); 2 Reference ranges are the Recommended Dietary
Allowance or the Acceptable Macronutrient Distribution Range; note the American College of Sports Medicine
recommends that athletes consume 1.2–2.0 g protein/kg body mass.
4. Discussion
Results from this study indicate that compared to their omnivore counterparts, vegetarian
endurance athletes have comparable strength as indicated by leg extension peak torque, and possibly
a greater degree of aerobic capacity, particularly in females, as indicted by a progressive maximal
Nutrients 2016, 8, 726
5 of 11
treadmill test to exhaustion. Dietary intake on several key nutrients differed considerably between
groups. Some, but not all, results are consistent with previous reports.
Our study is significant for its increased rigor in measurement assessments compared to previous
comparisons of vegetarian and omnivore athletes. We determined maximal oxygen uptake by a
graded test to exhaustion on a treadmill instead of predicting VO2 max using a cycle ergometer,
as recommended by Shepard and colleagues [32]. Additionally, we measured body composition
using a DXA scan, currently regarded as the clinical gold standard for body composition assessment,
instead of skinfolds [33]. Finally, we assessed both athletic performance and nutrient intake differences
between vegetarians and omnivores, whereas most previously published studies focus exclusively on
one of these areas.
4.1. Body Mass and BMI
Like other studies of vegetarians in the general population, vegetarian participants in the present
study had significantly lower body mass compared to omnivores [10,34]. This is in spite of the fact
that our study included participants engaged in considerable endurance activities, which could be
very different in multiple ways from the general population. One prior study in athletes, conducted
by Hanne et al. compared vegetarians and omnivores anthropometrically and found no significant
differences between groups for weight [27]. It is noteworthy that the athletes in the Hanne et al. study
included football, basketball, and water polo players in addition to endurance athletes.
4.2. Lean Body Mass
LBM was significantly lower for the vegetarian athletes compared to their omnivore counterparts,
a difference which was most prominent among the female participants with female vegetarian athletes
possessing 7% less LBM as compared to the female omnivore athletes. In spite of this, there were no
significant differences in body fat percentage or BMI between groups. To our knowledge, this is the
first study to examine lean body mass differences between vegetarian and omnivore athletes. It is
important to note, however, that this difference in lean body mass did not translate into differential
peak torque on the leg extension.
Although other studies have not assessed lean body mass of vegetarian athletes specifically,
Campbell and colleagues compared resistance-training induced changes in lean body mass and
strength between groups assigned to either an omnivorous diet or a lacto-ovo-vegetarian diet for the
duration of the study and found that, in spite of differential lean body mass gains, the two groups
increased strength similarly [21]. Conversely, a 12-week training study by Haub and colleagues showed
no significant differences in strength, body composition, or muscle cross-sectional area between groups
assigned to either a lacto-ovo-vegetarian or beef-containing diet.
4.3. Body Fat Percent and Visceral Adipose Tissue (VAT)
Contrary to the female vegetarian athletes in Hanne’s group, no significant differences in body
fat percentage were found between vegetarian and omnivore athletes in this study. Additionally,
there were no significant differences between groups for visceral adipose tissue (VAT). Participants
in the present study had VAT values above those reported for similar aged healthy lean sedentary
adults (~250 cm3), both omnivores and vegetarians [35,36], but lower than those noted for older
adults (1000–1560 cm3) [37]. Although there are no standard reference ranges for VAT, values near
1000 cm3 were associated with BMI values near 25 kg/m2 and values > 300 cm3 have been suggested
as predictive of risk for metabolic syndrome in young adults [36,37]. As technology permitting
quantification of visceral adipose tissue is relatively new for research purposes, this study contributes
to the emerging literature by providing VAT values for athletes. VAT and BMI is strongly correlated
in this study (p = 0.742), a factor that may be important for estimating VAT inexpensively without a
DXA scan.
Nutrients 2016, 8, 726
6 of 11
4.4. VO2 Max
Unlike athletes in Hanne’s study, vegetarians in the present study had significantly higher
maximal oxygen uptake than their omnivore counterparts [27]. This difference was most predominant
in the female participants with a 13% greater VO2 max score for the female vegetarians as compared
to the female omnivores, but this difference was not observed for absolute VO2 max (L/min),
which suggests that body weight factored into this difference. This gender difference is intriguing and
merits further investigation in future studies. One potential reason that athletes in the present study
had higher VO2 max values than those in Hanne’s study may be due to the difference between cycle
ergometry and treadmill testing methods. However, it is possible that the athletes in our study simply
were more trained and that diet effects on differences in VO2 potential emerge only at higher levels
of fitness.
Other work that contributes to our understanding of aerobic and anaerobic performance
differences by diet include the study of Hietavala et al. that found no significant difference in time
to exhaustion (albeit a higher oxygen uptake at a given percent of maximal oxygen consumption)
between participants following a low-protein vegetarian diet compared to a mixed diet [38]. Subjects
in this study adhered to the low protein vegetarian diet (0.80 ± 0.11 g of protein per kilogram of body
mass (g/kg) vs. 1.59 ± 0.28 g/kg on their normal diet) for four days before being tested on a cycle
ergometer. As this study did not use participants who practiced vegetarianism outside of the study,
and the amount of protein that subjects were allowed to consume on the vegetarian diet was restricted,
true differences between vegetarians and omnivores may not be evident. Baguet et al. found no
differences in repeated sprint ability between participants following a vegetarian or mixed diet for five
weeks; again, these subjects were not following a vegetarian diet long-term [26]. Raben et al. found no
differences in maximal oxygen uptake among subjects after adoption of a lacto-ovo vegetarian diet for
six weeks [25]. However, the major disadvantage of interpreting results of these studies for vegetarian
athletes is that participants in these studies only adhered to a vegetarian diet briefly for the duration of
the study.
4.5. Peak Torque
Similar to the Hanne et al. study that compared the power output of vegetarian and omnivore
athletes [27], we found no significant differences by diet in terms of peak torque using leg extensions.
Other studies in untrained older men that have examined strength development over time in response
to a training program have found mixed results when comparing participants following a vegetarian
or mixed diet [21,24]. This is noteworthy, particularly since strength and lean body mass were strongly
correlated (r = 0.764) in the present study, as well as the fact that omnivores had significantly more
lean body mass vs. the vegetarians. A nonsignificant trend for omnivores to produce higher peak
torque is observed, however. It is conceivable that the omnivore diet pattern may be preferred for
sports that rely on greater lean mass, and subsequently peak torque. To further investigate this, future
work ought to examine if strength can be increased similarly by vegetarian and omnivore athletes
engaged in strength training (not just by participants following a vegetarian diet for a few weeks).
4.6. Nutrient Intake
Nutrient intake was calculated from food and beverage intakes only and did not include any
supplements. There were no significant differences in caloric intake or total fat intake between
vegetarians and omnivores. However, vegetarians reported significantly more dietary carbohydrate
(both in terms of absolute intake and as a percent of daily calories), fiber, and iron intake. Omnivores
consumed more dietary protein (both in terms of absolute intake and as a percent of daily calories),
saturated fat, cholesterol, and vitamin B12. However, when expressed relative to body mass, there
were no differences in dietary protein intake.
Nutrients 2016, 8, 726
7 of 11
That vegetarians and omnivores in the present study did not differ in terms of caloric intake is
consistent with findings by Janelle and Barr from their comparison of 45 vegetarian and omnivore
women [16], yet it is in contrast to results from Calkins and colleagues who compared 50 vegetarian,
vegan, and omnivores. They found vegetarians consumed about 200 fewer kcal than omnivores [19].
These studies were both in the general population, not specifically with athletes. Calkins et al. also
reported that omnivores consumed more fat than vegetarians, a fact that partially contributed to the
higher caloric intake. This too is in contrast to the findings in the present study which found no
significant difference either in grams of fat consumed or the percent contribution of fat to the daily
calorie intake, even though saturated fat was significantly higher in omnivorous diets. Other studies
involving the general population have also reported omnivores eating more energy and total fat than
vegetarians [10,39–41].
Higher carbohydrate (when expressed either as an absolute amount or as a percent of total daily
calories) and fiber intake among vegetarians in comparison to omnivores in the present study is
consistent with findings in other studies [10,39,41–44]. As these studies have been conducted in the
general population, the present study contributes to the literature by demonstrating that this dietary
pattern can be extended to endurance athletes as well. One study by Janelle and Barr stands in contrast
to these findings, as they did not find significant differences in carbohydrate or fiber intake between
vegetarian and omnivore women; those participants were not athletes [16]. That vegetarians in the
present study consumed more carbohydrates than omnivores is notable since they are all athletes,
and the importance of carbohydrates for exercise is well-established [45–47].
Like the present study, other studies have also reported that vegetarians consume less protein
(both absolute intake and as a percent of the daily calories) [10,16,39,42] and vitamin B12 [40,48]
than omnivores. Our study contributes to the literature since other reports have been in the general
population instead of within athletic groups. Of note, though, differences in dietary protein intake
are not significant when expressed relative to body mass, which is typically the preferred method for
recommending protein for athletes [47]. Nonetheless, dietary protein intake was weakly correlated
with peak torque (r = 0.359, p = 0.006) in the present study, and dietary protein intake was moderately
correlated with lean body mass (r = 0.415, p = 0.001). Expectantly, lean body mass was strongly
correlated with peak torque (r = 0.764, p < 0.001). Hence, it is conceivable that protein intake could
influence strength if intakes had been inadequate. In the present evaluation, protein intakes in the
vegetarian participants averaged 1.2 g/kg body mass, which falls in the recommended range for
athletes [47,49].
There are conflicting findings in the rest of the literature regarding whether omnivores or
vegetarians consume more iron. The Wilson et al. study of vegetarian men found that vegetarians
consumed more iron [41], but Ball and Bartlett reported no difference in dietary iron intake
between female vegetarian and omnivores [50]. Clary et al. compared 1475 vegans, vegetarians,
semi-vegetarians, pescetarians, and omnivores and also showed that vegetarians consume more iron
than omnivores [39]. Although vegetarians consumed more iron than omnivores in the present study,
iron bioavailability was likely reduced as has been shown in other trials [17]. Dietary intakes of zinc
did not vary by diet group herein, but generally the literature suggests that vegetarians consume
somewhat less dietary zinc than omnivores [16,51–53]. The lower intakes of selenium by vegetarians
in comparison to omnivores has also been reported by others and reflects the low levels of selenium in
plant foods relative to flesh foods [54,55].
4.7. Limitations
In addition to the small sample size, limitations to the study include the variable level of experience
of the athletes for their respective sports, and related fitness levels. Although most participants were
training for and competing in races such as marathons, Ironman-distance triathlons, and competitive
cycling, there were a few participants who were training for shorter distance races.
However,
this variation makes results more generalizable to athletes of various fitness levels.
Nutrients 2016, 8, 726
8 of 11
4.8. Future Directions
Future work is needed to compare vegetarian and omnivore endurance athletes’ performance on
events more similar to actual sporting events (such as time trials or peak power on a cycle ergometer)
and probe differences by type of vegetarian diet (lacto-ovo vegetarian or vegan). Additional work is
needed to explore the adequacy of long-term adherence to vegetarian and vegan diets for supporting
development of lean body mass.
5. Conclusions
Our cross-sectional comparison of vegetarian and omnivore adult endurance athletes shows
higher maximal oxygen uptake values among vegetarians and comparable strength, in spite of
anthropometric and dietary differences.
This study suggests that following a vegetarian diet
may adequately support strength and cardiorespiratory fitness development, and may even be
advantageous for supporting cardiorespiratory fitness. Certainly many factors affect an athlete’s
sports performance, and there is no dietary substitute for quality training. However, our study
contributes to the literature about cardiorespiratory and strength comparisons between vegetarian
and omnivore endurance athletes, and may provide a rationale about the adequacy of vegetarian
diets for sport performance. As this was a small cross-sectional study using endurance athletes,
larger intervention trials are necessary to bolster conclusions about adequacy of vegetarian diets to
support performance in strength and power-focused sports.
Acknowledgments: This study was supported by a grant through the Graduate and Professional Student
Association (GPSA) at Arizona State University.
Author Contributions: H.M.L. and C.S.J. conceived and designed the experiments; H.M.L. performed the
experiments; H.M.L. and C.S.J. analyzed the data; H.M.L., C.S.J., and C.M.W. wrote the paper.
Conflicts of Interest: The authors declare no conflict of interest. The funding sponsor had no role in the design of
the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, and in the decision
to publish the results.
References
1.
Fraser, G. A comparison of first event coronary heart disease rates in two contrasting California populations.
J. Nutr. Health Aging 2004, 9, 53–58.
2.
Catsburg, C.; Kim, R.S.; Kirsh, V.A.; Soskolne, C.L.; Kreiger, N.; Rohan, T.E. Dietary patterns and breast
cancer risk: A study in 2 cohorts. Am. J. Clin. Nutr. 2015, 101, 817–823. [CrossRef] [PubMed]
3.
Orlich, M.J.; Singh, P.N.; Sabaté, J.; Fan, J.; Sveen, L.; Bennett, H.; Knutsen, S.F.; Beeson, W.L.; Jaceldo-Siegl, K.;
Butler, T.L.; et al. Vegetarian dietary patterns and the risk of colorectal cancers. JAMA Int. Med. 2015, 175,
767–776. [CrossRef] [PubMed]
4.
Tantamango-Bartley, Y.; Knutsen, S.F.; Knutsen, R.; Jacobsen, B.K.; Fan, J.; Beeson, W.L.; Sabate, J.; Hadley, D.;
Jaceldo-Siegl, K.; Penniecook, J.; et al. Are strict vegetarians protected against prostate cancer? Am. J.
Clin. Nutr. 2016, 103, 153–160. [CrossRef] [PubMed]
5.
Kahleova, H.; Pelikanova, T. Vegetarian Diets in the Prevention and Treatment of Type 2 Diabetes. J. Am.
Coll. Nutr. 2015, 34, 448–458. [CrossRef] [PubMed]
6.
Kim, M.-H.; Bae, Y.-J. Comparative Study of Serum Leptin and Insulin Resistance Levels Between Korean
Postmenopausal Vegetarian and Non-vegetarian Women. Clin. Nutr. Res. 2015, 4, 175–181. [CrossRef]
[PubMed]
7.
Yokoyama, Y.; Nishimura, K.; Barnard, N.D.; Takegami, M.; Watanabe, M.; Sekikawa, A.; Okamura, T.;
Miyamoto, Y. Vegetarian diets and blood pressure: A meta-analysis. JAMA Int. Med. 2014, 174, 577–587.
[CrossRef] [PubMed]
8.
Appleby, P.N.; Allen, N.E.; Key, T.J. Diet, vegetarianism, and cataract risk. Am. J. Clin. Nutr. 2011, 93,
1128–1135. [CrossRef] [PubMed]
9.
Giem, P.; Beeson, W.L.; Fraser, G.E. The incidence of dementia and intake of animal products: Preliminary
findings from the Adventist Health Study. Neuroepidemiology 1993, 12, 28–36. [CrossRef] [PubMed]
Nutrients 2016, 8, 726
9 of 11
10.
Spencer, E.A.; Appleby, P.N.; Davey, G.K.; Key, T.J. Diet and body mass index in 38000 EPIC-Oxford
meat-eaters, fish-eaters, vegetarians and vegans. Int. J. Obes. 2003, 27, 728–734. [CrossRef] [PubMed]
11.
Quiles, L.; Portolés, O.; Sorlí, J.V.; Corella, D. Short Term Effects on Lipid Profile and Glycaemia of a Low-Fat
Vegetarian Diet. Nutr. Hosp. 2014, 32, 156–164.
12.
Pimentel, D.; Pimentel, M. Sustainability of meat-based and plant-based diets and the environment. Am. J.
Clin. Nutr. 2003, 78, 660S–663S. [PubMed]
13.
Monsivais, P.; Scarborough, P.; Lloyd, T.; Mizdrak, A.; Luben, R.; Mulligan, A.A.; Wareham, N.J.; Woodcock, J.
Greater accordance with the Dietary Approaches to Stop Hypertension dietary pattern is associated with
lower diet-related greenhouse gas production but higher dietary costs in the United Kingdom. Am. J.
Clin. Nutr. 2015, 102, 138–145. [CrossRef] [PubMed]
14.
Masset, G.; Vieux, F.; Verger, E.O.; Soler, L.-G.; Touazi, D.; Darmon, N. Reducing energy intake and energy
density for a sustainable diet: A study based on self-selected diets in French adults. Am. J. Clin. Nutr. 2014,
99, 1460–1469. [CrossRef] [PubMed]
15.
Delanghe, J.; De Slypere, J.P.; De Buyzere, M.; Robbrecht, J.; Wieme, R.; Vermeulen, A. Normal reference
values for creatine, creatinine, and carnitine are lower in vegetarians. Clin. Chem. 1989, 35, 1802–1803.
[PubMed]
16.
Janelle, K.C.; Barr, S.I. Nutrient intakes and eating behavior see of vegetarian and nonvegetarian women.
J. Am. Diet. Assoc. 1995, 95, 180–189. [CrossRef]
17.
Hunt, J.R. Bioavailability of iron, zinc, and other trace minerals from vegetarian diets. Am. J. Clin. Nutr. 2003,
78, 633S–639S. [PubMed]
18.
Rauma, A.-L.; Mykkänen, H. Antioxidant status in vegetarians versus omnivores. Nutrition 2000, 16, 111–119.
[CrossRef]
19.
Calkins, B.M.; Whittaker, D.J.; Nair, P.P.; Rider, A.A.; Turjman, N. Diet, nutrition intake, and metabolism
in populations at high and low risk for colon cancer. Nutrient intake. Am. J. Clin. Nutr. 1984, 40, 896–905.
[PubMed]
20.
Nieman, D. Vegetarian dietary practices and endurance performance. Am. J. Clin. Nutr. 1988, 48, 754–761.
[PubMed]
21.
Campbell, W.W.; Barton, M.L., Jr.; Cyr-Campbell, D.; Davey, S.L.; Beard, J.L.; Parise, G.; Evans, W.J. Effects
of an omnivorous diet compared with a lactoovovegetarian diet on resistance-training-induced changes in
body composition and skeletal muscle in older men. Am. J. Clin. Nutr. 1999, 70, 1032–1039. [PubMed]
22.
Haub, M.D.;
Wells, A.M.;
Tarnopolsky, M.A.;
Campbell, W.W. Effect of protein source on
resistive-training-induced changes in body composition and muscle size in older men. Am. J. Clin. Nutr.
2002, 76, 511–517. [PubMed]
23.
Wells, A.M.; Haub, M.D.; Fluckey, J.; Williams, D.K.; Chernoff, R.; Campbell, W.W. Comparisons of vegetarian
and beef-containing diets on hematological indexes and iron stores during a period of resistive training in
older men. J. Am. Diet. Assoc. 2003, 103, 594–601. [CrossRef] [PubMed]
24.
Haub, M.D.; Wells, A.M.; Campbell, W.W. Beef and soy-based food supplements differentially affect serum
lipoprotein-lipid profiles because of changes in carbohydrate intake and novel nutrient intake ratios in older
men who resistive-train. Metabolism 2005, 54, 769–774. [CrossRef] [PubMed]
25.
Raben, A.; Kiens, B.; Richter, E.A.; Rasmussen, L.B.; Svenstrup, B.; Micic, S.; Bennett, P. Serum sex hormones
and endurance performance after a lacto-ovo vegetarian and a mixed diet. Med. Sci. Sports Exerc. 1992, 24,
1290–1297. [CrossRef] [PubMed]
26.
Baguet, A.; Everaert, I.; De Naeyer, H.; Reyngoudt, H.; Stegen, S.; Beeckman, S.; Achten, E.; Vanhee, L.;
Volkaert, A.; Petrovic, M.; et al. Effects of sprint training combined with vegetarian or mixed diet on muscle
carnosine content and buffering capacity. Eur. J. Appl. Physiol. 2011, 111, 2571–2580. [CrossRef] [PubMed]
27.
Hanne, N.; Dlin, R.; Nrotstein, A. Physical fitness, anthropometric and metabolic parameters in vegetarian
athletes. J. Sports Med. Phys. Fit. 1986, 26, 180–185.
28.
Bruce, R.A.; Blackmon, J.R.; Jones, J.W.; Strait, G. Exercising testing in adult normal subjects and cardiac
patients. Pediatrics 1963, 32, 742–756. [CrossRef] [PubMed]
29.
Borg, G. Borg’s Perceived Exertion and Pain Scales; Human Kinetics: Champaign, IL, USA, 1998.
30.
Wier, L.T.; Jackson, A.S.; Ayers, G.W.; Arenare, B. Nonexercise models for estimating VO2 max with waist
girth, percent fat, or BMI. Med. Sci. Sports Exerc. 2006, 38, 555–561. [CrossRef] [PubMed]
Nutrients 2016, 8, 726
10 of 11
31.
Astorino, T.A.; Robergs, R.A.; Ghiasvand, F.; Marks, D.; Burns, S. Incidence of the oxygen plateau at VO2
max during exercise testing to volitional fatigue. Methods 2000, 3, 1–12.
32.
Shephard, R.J.; Allen, C.; Benade, A.J.S.; Davies, C.T.M.; di Prampero, P.E.; Hedman, R.; Merriman, J.E.;
Myhre, K.;
Simmons, R. The maximum oxygen intake:
An international reference standard of
cardio-respiratory fitness. Bull. World Health Organ. 1968, 38, 757. [PubMed]
33.
Andreoli, A.; Garaci, F.; Cafarelli, F.P.; Guglielmi, G. Body composition in clinical practice. Eur. J. Radiol.
2016, 85, 1461–1468. [CrossRef] [PubMed]
34.
Berkow, S.E.; Barnard, N. Vegetarian diets and weight status. Nutr. Rev. 2006, 64, 175–188. [CrossRef]
[PubMed]
35.
Knurick, J.R.; Johnston, C.S.; Wherry, S.J.; Aguayo, I. Comparison of correlates of bone mineral density in
individuals adhering to lacto-ovo, vegan, or omnivore diets: A cross-sectional investigation. Nutrients 2015,
7, 3416–3426. [CrossRef] [PubMed]
36.
Miazgowski, T.; Krzy˙zanowska-´Swiniarska, B.; Dziwura-Ogonowska, J.; Widecka, K. The associations
between cardiometabolic risk factors and visceral fat measured by a new dual-energy X-ray
absorptiometry-derived method in lean healthy Caucasian women. Endocrine 2014, 47, 500–505. [CrossRef]
[PubMed]
37.
Lin, H.; Yan, H.; Rao, S.; Xia, M.; Zhou, Q.; Xu, H.; Rothney, M.P.; Xia, Y.; Wacker, W.K.; Ergun, D.L.; et al.
Quantification of visceral adipose tissue using lunar dual-energy X-ray absorptiometry in Asian Chinese.
Obesity 2013, 21, 2112–2117. [CrossRef] [PubMed]
38.
Hietavala, E.-M.; Puurtinen, R.; Kainulainen, H.; Mero, A.A. Low-protein vegetarian diet does not have
a short-term effect on blood acid–base status but raises oxygen consumption during submaximal cycling.
J. Int. Soc. Sports Nutr. 2012, 9, 50. [CrossRef] [PubMed]
39.
Clarys, P.; Deliens, T.; Huybrechts, I.; Deriemaeker, P.; Vanaelst, B.; De Keyzer, W.; Hebbelinck, M.;
Mullie, P. Comparison of nutritional quality of the vegan, vegetarian, semi-vegetarian, pesco-vegetarian and
omnivorous diet. Nutrients 2014, 6, 1318–1332. [CrossRef] [PubMed]
40.
Alexander, D.; Ball, M.; Mann, J. Nutrient intake and haematological status of vegetarians and age-sex
matched omnivores. Eur. J. Clin. Nutr. 1994, 48, 538–546. [PubMed]
41.
Wilson, A.; Ball, M. Nutrient intake and iron status of Australian male vegetarians. Eur. J. Clin. Nutr. 1999,
53, 189–194. [CrossRef] [PubMed]
42.
Key, T.J.; Davey, G.K.; Appleby, P.N. Health benefits of a vegetarian diet. Proc. Nutr. Soc. 1999, 58, 271–275.
[CrossRef] [PubMed]
43.
Kennedy, E.T.; Bowman, S.A.; Spence, J.T.; Freedman, M.; King, J. Popular diets: Correlation to health,
nutrition, and obesity. J. Acad. Nutr. Diet. 2001, 101, 411. [CrossRef]
44.
Hardinge, M.G.; Chambers, A.C.; Crooks, H.; Stare, F.J. Nutritional studies of vegetarians III. Dietary levels
of fiber. Am. J. Clin. Nutr. 1958, 6, 523–525. [PubMed]
45.
Costill, D.; Miller, J. Nutrition for endurance sport: Carbohydrate and fluid balance. Int. J. Sports Med. 1980,
1, 2–14. [CrossRef]
46.
Rodriguez, N.R.; DiMarco, N.M.; Langley, S. Position of the American dietetic association, dietitians of
Canada, and the American college of sports medicine: Nutrition and athletic performance. J. Am. Diet. Assoc.
2009, 109, 509–527. [PubMed]
47.
Thomas, D.T.; Erdman, K.A.; Burke, L.M. Position of the academy of nutrition and dietetics, dietitians of
canada, and the american college of sports medicine: Nutrition and athletic performance. J. Acad. Nutr. Diet.
2016, 116, 501–528. [CrossRef] [PubMed]
48.
Antony, A.C. Vegetarianism and vitamin B-12 (cobalamin) deficiency. Am. J. Clin. Nutr. 2003, 78, 3–6.
[PubMed]
49.
Phillips, S.M.; Van Loon, L.J. Dietary protein for athletes: From requirements to optimum adaptation.
J. Sports Sci. 2011, 29 (Suppl. 1), S29–S38. [CrossRef] [PubMed]
50.
Ball, M.J.; Bartlett, M.A. Dietary intake and iron status of Australian vegetarian women. Am. J. Clin. Nutr.
1999, 70, 353–358. [PubMed]
51.
Freeland-Graves, J.H.; Bodzy, P.W.; Eppright, M.A. Zinc status of vegetarians. J. Am. Diet. Assoc. 1980, 77,
655–661. [PubMed]
52.
Anderson, B.M.; Gibson, R.S.; Sabry, J.H. The iron and zinc status of long-term vegetarian women. Am. J.
Clin. Nutr. 1981, 34, 1042–1048. [PubMed]
Nutrients 2016, 8, 726
11 of 11
53.
Gibson, R.S. Content and bioavailability of trace elements in vegetarian diets. Am. J. Clin. Nutr. 1994, 59,
1223S–1232S. [PubMed]
54.
Larsson, C.L.; Johansson, G.K. Dietary intake and nutritional status of young vegans and omnivores in
Sweden. Am. J. Clin. Nutr. 2002, 76, 100–106. [PubMed]
55.
Letsiou, S.; Nomikos, T.; Panagiotakos, D.; Pergantis, S.A.; Fragopoulou, E.; Antonopoulou, S.; Pitsavos, C.;
Stefanadis, C. Dietary habits of Greek adults and serum total selenium concentration: The ATTICA study.
Eur. J. Nutr. 2010, 49, 465–472. [CrossRef] [PubMed]
© 2016 by the authors; licensee MDPI, Basel, Switzerland. This article is an open access
article distributed under the terms and conditions of the Creative Commons Attribution
(CC-BY) license (http://creativecommons.org/licenses/by/4.0/).
| Cardiorespiratory Fitness and Peak Torque Differences between Vegetarian and Omnivore Endurance Athletes: A Cross-Sectional Study. | 11-15-2016 | Lynch, Heidi M,Wharton, Christopher M,Johnston, Carol S | eng |
PMC7365446 | See responses in green.
Reviewer #1: General Comments:
The reviewer would like to commend the authors for undertaking an important and interesting
topic. Determining the shoe recommendations for different running levels is an important topic,
that can aid clinicians and running coaches in choosing the right foot wear for different runners
of different abilities.
Overall this is a well written manuscript, with good methodology. There are some specific
comments which are written below.
>>Thank you for your compliments and suggestions. They have improved our manuscript.
Abstract:
General comment: For an abstract, the background should be brief. Suggest only have 2
sentence for the background. I do not think you need to describe why a Delphi study is powerful
within the abstract. I think the first 3 sentences would suffice, and potentially reduce these three
sentences into 2.
Within the abstract methods, a little bit more information is needed. For example, how many
questions did the study begin with, and how were they whittled down through the three rounds,
and how was data tallied. Further, within the results, you describe that there were originally 20
proposed variables. This is an example of something that needs to be in the methods.
>>We have updated the abstract as suggested with the following:
“Providing runners with footwear that match their functional needs has the potential to improve
footwear comfort, enhance running performance, and reduce the risk of overuse injuries. It is
currently not known how footwear research experts make decisions about different shoe
features and their properties for runners of different levels. We performed a Delphi study in
order to understand: 1) definitions of different runner levels, 2) which footwear features are
considered important, and 3) how these features should be prescribed for runners of different
levels. Experienced academics, journalists, coaches, bloggers and physicians that examine the
effects of footwear on running were recruited to participate in three rounds of a Delphi study.
Three runner level definitions were refined throughout this study based on expert feedback.
Experts were also provided a list of 20 different footwear features. They were asked which
features were important and what the properties of those features should be.” (line 26-35)
Need key words at the end of the abstract.
>>Thank you for the reminder. We have included the following key words: Individualized
footwear, running biomechanics, runner abilities, footwear experts, midsole hardness
Introduction:
Line 54: Delete the parenthetical citation fully written citation, should just be a reference
number.
>>This citation has been replaced with the appropriate number.
Line 67: Same here, please deleted written citation, should just be a reference number.
>>This citation has been replaced with the appropriate number.
Line 69-70: Reword this to not be a numbered list. Within the intro, it should just be written
sentences.
>>We have removed the numbers from the sentence and updated the text to the following:
“On the other hand, there has been little scientific attention on footwear features such as outsole
traction or forefoot flares which could indicate: the prescription of these features to different
runner levels is trivial, or that these features are not considered important by footwear
professionals, or little is known on how to prescribe these features.” (line 78-81)
Line 71: You state, “it is close to impossible for running footwear professionals to provide
evidence-based recommendations for footwear properties for runners of different levels.” But
then you go on to say you are performing a Delphi to find the best recommendations from the
experts. I think this is contradictory. I think you should focus more on how there is not clarity on
professional recommendations for footwear for different running skills or groups.
>>We have removed the last comment from the introduction to here as we addressed these two
comments together. The corresponding statements now read:
“In summary, there is a need to better understand how footwear research experts make
decisions about different footwear features and their properties” (Lines 81-83)
I think the third to last and second to last paragraphs can be amalgamated into one paragraph.
Further, the second to last paragraph ends abruptly and a better conclusion is need to set up
the purpose paragraph.
>>We have combined the two paragraphs and updated the phrasing so that it is more focused
on “how there is not clarity in professional recommendations”:
“Modern running shoes are complex systems. They incorporate many different features (e.g.,
crash-pads, heel counters, flares, midsole hardness) and each of these features can be
included, excluded, and/or tuned individually to modify the characteristics of the final running
shoe system (e.g., cushioning, stability, heel-to-toe transition, energy return). Some of these
shoe features have been studied more extensively, such as rearfoot midsole hardness, while
others have received little attention, such as upper breathability (12). Nevertheless, a strong
research focus on certain footwear features (e.g., midsole hardness) does not necessarily
translate into agreement on how modifying these features may affect the running mechanics,
performance, injury risk, or footwear comfort in runners of different levels. For example, a recent
review found inconclusive evidence regarding the biomechanical effects of different midsole
hardness – one of the most studied footwear features (12). On the other hand, there has been
little scientific attention on footwear features such as outsole traction or forefoot flares which
could indicate: the prescription of these features to different runner levels is trivial, or that these
features are not considered important by footwear professionals, or little is known on how to
prescribe these features. In summary, there is a need to better understand how footwear
research experts make decisions about different footwear features and their properties. A
powerful way to examine these decisions is to gather and summarize opinions of experts in the
field of running biomechanics and footwear using a Delphi study. The Delphi method has been
utilized for gathering and summarizing opinions via survey-based responses of an expert panel
in order to obtain consensus on complex topics. For example, this technique has been
successfully applied to establish the now frequently reported “Minimalist Index” of running shoes
(13). Such an understanding can target future systematic investigations around the presumed
optimal property of important footwear features.” (Lines 69-88)
Line 73-76: Why are aims here and the purpose in the final introduction paragraph? This is
confusing for the reader. Suggest only having the purpose at the last intro paragraph and
deleting the aims.
>>We have deleted the aims as suggested.
Methods:
General comments: An overall study design sub section is needed at the beginning of the
methods. This should give the 10,000 foot view of the study.
>>We have included an overview at the beginning of the Methods section:
“Footwear research experts were asked to complete three rounds of a Delphi study, with each
successive round building on the results gathered from the previous round. Three runner level
definitions were refined throughout the three rounds of the Delphi study through expert
feedback. Experts were also provided a list of 20 different footwear features. Through the three
rounds of the study, experts provided opinions on which features are important and what the
properties should be for the footwear features for the three different running levels.” (lines 96-
101)
You need to give inclusion/exclusion criteria for who was considered an expert for this study.
>>We have added the following exclusion criteria in the methods:
“Participants were excluded if they had under two years of research experience related to
running footwear.” (line 113-114)
Lines 103-117: I see that 142 experts were contacted. How many responded and were included.
A flow chart might help the reader to understand this process.
>>We have included a consort diagram (new Figure 1) to show the number of experts in each
round.
Line 122: Need to cite the running lit used. Further, it is confusing with the parenthetical
statement “as detailed below. Suggest deleting this.
>>We have deleted the parenthetical statement and replaced it with the citations used in the
subsequent paragraph (line 128).
Novice versus Recreational runner definition: In the novice group, you state that they run no
more than 20km/week, but in the recreational group, they run 10-50 km/week? How do
delineated between someone that runs 15-20 km/week? Is this based off of times per week (0-3
v 1-5)? Please clarify. High Caliber runners: I see the same thing here, they run 30km+/week.
Please clarify
>>Thank you for clarifying this. We have included the following description to clarify the
overlapping mileage:
“The proposed characteristics provide guidelines for runner classification. As such, there is
overlap in the running distance per week between the different running levels in order to
accommodate runners that train less and have a better running performance.” (line 128-130)
Line 211: Can you explain further why the ‘don’t know’ questions were not included in round 3?
>>We have expanded upon our explanation with the following:
“These questions were only included in the second-round as we received feedback from the
experts that the questionnaire was time consuming which may have increased the drop out rate
if these questions were asked in the third-round again.” (line 231-233)
Line 218-219: Please add the software program used to calculate these statistics.
>>We added the software program that we used to calculate the statistics in the “Analysis and
Visualization” section: “All statistical analyses were performed in MATLAB (MathWorks, Natick,
MA, USA)”.(line 240)
Results:
General comment: It is not recommended to use bullet points within the results. Please edit
accordingly
>>We have eliminated the bullet points and updated the text to the following:
“The respondents’ rating of the running level definitions improved as the Delphi study
progressed. The median score given to the running level definitions increased each round and
the interquartile range decreased as 88% of respondents rated the running level definitions
between 7 and 10 in the third-round as opposed to 69% in the first-round (see Fig. 3). The
changes to the running level definitions for the second-round were: increased “novice” running
experience to one year (from six months) and increased “recreational” running experience to
greater than one year (from six months), increased “high caliber” running habits to >4
sessions/week (from >3 sessions/week) and >50 km/week (from >30 km/week), specified the
running performance as males between the ages of 18 to 34, replaced “stress management”
with enjoyment for running motivation for all levels, re-order the “high caliber” running motivation
from 1) Improve general health, 2) Stress management, 3) Competition to: 1) Competition, 2)
Improve general heath, and 3) Enjoyment, and re-order the priorities for footwear design for
“High caliber” from: 1) Improve performance, 2) Improve comfort, 3) Reduce injury risk, to: 1)
Improve performance, 2) Reduce injury risk, 3) Improve comfort. Subsequent changes to the
running level definitions were to ensure that the high caliber and recreational runner 5km and 10
km time were indicative of the respective marathon times. These updates resulted in the final
updated runner level definitions in Table 3.” (line 251-266)
Overall the results are well written.
>>Thank you!
Discussion:
Line 307-10: These are more article strengths and should be moved to the strength and
limitations section, not the summary discussion paragraph.
>>We have eliminated the sentences in question.
Line 315-17: This is a future research, implications, and/or conclusion sentence and should be
moved
>>I understand that this concluding sentence pertains to future research, in our opinion it is a
major point of discussion. We have kept this sentence as it wraps together the discussion
summary paragraph.
Line 369: Suggest deleting the term ground truth and just state that this should serve as
valuable information, etc.
>>We have eliminated “ground truth” and updated the sentence to the following:
“As such, the findings from this study can serve as a valuable starting point for future systematic
biomechanical investigations examining the influence of footwear features on runners with
different training/performance levels.” (line 412-414)
Line 373-381: suggest that this paragraph be moved before the limitations paragraph.
>>We have moved the paragraph before the Limitation paragraph as suggested. (line 379-387)
Line 384-386: Delete the first sentence, this can be said in the strengths paragraph but the
conclusion paragraph should focus on the findings and future directions.
>>We have removed the strengths portion of the sentence and updated it to:
“Footwear research experts provided feedback on the effects of different footwear features on
running biomechanics across three running levels as well as provided a consensus on the
characteristics of runners in these different running levels.” (line 418-427)
Reviewer #2: Overall this manuscript fills an obvious void in the literature and aims to assist
researchers, clinicians, coaches, and running enthusiasts with shoe prescriptions, while also
informing future running shoe research. This work is generally well written and free from
fundamental flaws; however, several minor revisions to the proposed article will undoubtedly
improve this already great work.
>>Thank you for your kind comments. Your suggestions have improved the manuscript.
1. The words "the participants" are over utilized throughout the manuscript. Varied diction will
help to maintain reader interest and attention.
>>We have updated the manuscript so that there is more varied diction.
2. As this is a study employing Delphi techniques no statistical analyses are necessary and
furthermore, no analyses were actually conducted. The "Statistical Analysis" section is therefore
unnecessary and the subsequent descriptive statistics can simply be presented in the "Results"
as well as Fig 3.
>> Together with your suggestion and the Reviewer #1’s comment about what program the
statistics were performed in and your #11 comment, we have updated the “Statistical Analysis”
section to “Analysis and Visualization” section.
3. A more clear and consistent distinction between footwear properties and features throughout
the manuscript would improve readability.
>>We have checked the entire manuscript and ensured that “property” and “feature” were used
correctly.
4. "Appendix A" utilizes the term "categories" as opposed to "properties" further illustrating the
previous point.
>>We have updated “categories” to “Property categories” throughout the Appendix and updated
the file name of Appendix A to: “S1 Appendix A – Shoe Feature Descriptions and Properties”
5. Additional headings for the "Footwear Properties" in the "Methods" and "Results" sections
would assist readers navigating between parts of the manuscript.
>>We have added sections titled: “Footwear Feature Properties” in both the Methods and
Results sections.
6. The described methods for determining footwear features and feature properties importance
is challenging to read at times (particularly lines 169-179; lines 188-192); please try to concisely
and succinctly explain these steps.
>>We have updated the mentioned sections with the following:
“The importance of the footwear features was assessed in the first-round and verified in the
second-round. In the first-round, participants were asked if footwear features are important
when designing footwear for different running levels. The experts could choose between the
following for each footwear feature: (a) is important, (b) is not important or (c) they do not know
if it is important. The footwear features were important if over 75% of the first-round participants
selected option (a). Prior Delphi studies have defined consensus between 51% (21) and 80%
(22) of respondents. The important features were then presented to the second-round
participants. The participants were asked if they agreed with each of the features selected as
important/non important on a 10-point scale where “1” indicated that the list of important/non
important features were “Not at all appropriate” and “10” indicated “Most Appropriate”. The list of
important features was verified if over 75% of the second-round participants answered with a
seven or higher on the 10 point-scale. The second- and third-rounds of the Delphi study were
then limited to the important footwear features. In each round, the experts were asked if other
footwear features should be included in the Delphi study. If there were at least five suggestions
to add a certain feature, this new footwear feature was added to the subsequent round and the
participants were asked if the newly added footwear features were important.” (lines 187-201)
and
“The experts were asked to recommend footwear feature properties for the different running
levels in each round of the study from a multiple-choice selection (see Appendix A for the lists of
footwear feature properties). Most footwear feature properties were obtained through literature;
however, if there was no related literature (e.g., upper elasticity), properties were provided
based on commercially available shoes. In rounds 2 and 3, the results from the previous round
were presented to the participants. If at least 51% of the participants agreed on a footwear
feature property for a specific running level (e.g., high breathability for novice runners), the
participants would be asked if they agreed with the consensus the next round. If at least 51% of
the participants verified the consensus, the experts were not asked again to recommend a
footwear feature property for that running level (see Fig. 2). In comparison to the consensus for
the importance of shoe features (agreement of 75% of respondents), the threshold for
consensus was set lower for agreement on footwear feature properties (51%) because of the
greater number of available response options.” (lines 204-215)
7. Lines 181-184 seem somewhat redundant.
>>We have removed the sentence.
8. The reference to Fig 2 in line 182 seems somewhat premature. Describing the general flow of
these methods prior to interpreting Fig 2 made this section easier for this reviewer to
understand.
>>We have moved the reference to Fig. 2 to near the end of the paragraph after the explanation
of the methods.
9. It is not clear how the Likert scale used to rate footwear features (as described in the
"Methods" section) is actually used in this study.
>>We have clarified the use of the Likert scale by including the following in our methods:
“The list of important features was verified if over 75% of the second-round participants
answered with a seven or higher on the 10 point-scale.” (line 196-197)
10. Fig 2 is very helpful, but a threshold of >50% is provided when the text describes using a
51% threshold.
>>We have updated the Fig. 2 and replaced “>50%” with “≥51%”.
11. While minor, the software used to produce images was not stated.
>>We have added the following in methods: “Figures were created in MATLAB and Adobe
Illustrator (San Jose, CA, USA).” (line 220-241)
12. Line 162 - Explicitly cite why/where the 20 features considered comes from.
>>We have included the following to describe how we came to the 20 footwear features:
“. These 20 features were chosen from a list of 31 running shoe footwear features that were
identified based on an initial literature review, market analysis, and internal discussion. Two
influential studies during this process were reports from (6) and (13). The initial list of 31 was
reduced to 23 features by removing or joining related features that were reflected in other
features or similar in their function, respectively (e.g., remove midfoot midsole hardness and
only retain forefoot and rearfoot midsole hardness). Pilot testing with four footwear science
experts (not included in the main study) indicated that 23 features resulted in a questionnaire
that would require more than an hour to complete and could potentially lead to a high-drop out
rate. Therefore, we limited the number of footwear features to 20, by removing features for
which pilot participants indicated low relevance (e.g. upper overlays or varus alignment). In
return, the option was added for experts of the main study to suggest footwear features, that
should be added to the questionnaire.” (line 169-180)
13. The inclusion of 2 aims and 3 purposes is somewhat confusing. I recommend removing the
aims from your "Introduction" as they do not match the "Methods" and "Results" sections as
obviously.
>>We have removed the two aims from the introduction.
14. Please ensure that permissions for any adapted images (i.e. Figs 1 & 4) are provided as
necessary.
>>Figures 1 and 4 have been removed as we have replaced the Hoitz article (currently still in
review) with another recent running shoe construction review paper (Sun et al., 2020). “Sun, X,
Lam, WK, Zhang X., Wang J, & Fu W (2020). Systematic Review of the Role of Footwear
Constructions in Running Biomechanics: Implications for Running-Related Injury and
Performance. Journal of Sports Science and Medicine,19, 20-37”
15. A limitation that seems somewhat overlooked is that the definitions of runner levels changed
throughout iterations. As these definitions changed, so too may have respondents'
recommended properties. While the 3 repetitions and consensus measures may help to quell
these concerns, it seems important to consider the implications of these interconnected moving
targets.
>>We have added the following to the limitations as you suggested:
“The recommended footwear feature properties may have been influenced by a dynamic
definition of the runner levels, which changed slightly throughout the study. These changing
definitions, however, seemed to have little effect on expert opinions on the footwear feature
properties as the verifying consensus level was generally higher than the original consensus
level (Table 4, last vs. second-to-last column).” (line 403-408)
16. If possible, I would like to know more about your "Additional Delphi Questions" results in the
discussion. I read some of the statements in your raw data set and found the additional insights
very compelling. You do a good job of introducing some of the identified themes in your
"Discussion" but I feel that a bit more would elevate the current manuscript.
>>We have integrated some expert feedback in the running level definitions discussion
paragraph (line 434-451). Please see our response below #20 and #21.
17. Tables 5 and 6 both seem to provide complementary results. Is there a way to combine
them or make the more exclusive from one another?
>>We have eliminated Table 6 and added a column for “% Participant in agreement with
consensus” to Table 5.
18. Consider a CONSORT diagram so readers can better understand the development of the
expert panel round by round.
>>We have included a consort diagram, as suggested, to show the number of experts in each
round as the new Figure 1.
19. Please expand on how your panel may or may not influence your conclusions in the
"Discussion" (e.g. Where they all from the US? Do they disproportionately represent companies
with financial interests in designing complicated shoes? Etc.).
>>We have included the following to expand on our panel in the limitation:
“Furthermore, the final recommendation may have been biased as more experts that completed
the survey were male (e.g., 22/26 of the final participants). This expert panel was otherwise
diverse as nine countries were represented.” (line 401-403)
20. Please discuss how providing the expert panel with definitions in round 1 for running level as
opposed to forming definitions built by the panel may have influenced your conclusions.
21. Please expand on the results of your running level definitions in your "Discussion" section.
>>We have expanded our running level definitions discussion that includes discussion of #20:
“The footwear experts came to a consensus on the running level definitions through slight
adjustments to the initial definitions proposed and derived from literature. We opted to provide
initial running level definitions to our expert panel rather than letting the panel formulate
definitions independently. This latter approach would have required additional Delphi rounds
prior to the recommendation of footwear features and their properties. Panel formulated
definitions may have resulted in different running level definitions compared to the approach
presented here and different running level definitions could have led to altered footwear feature
recommendations. However, the experts’ consensus on the running level definitions were in
agreement with prior literature. This is exhibited by the novice runner level definition which is
similar to a definition created based on subjective running questionnaires (7). The experts did
recommend an increased workload for high caliber runners in comparison to literature (7) as
participant feedback resulted in the distance per week to be increased from >30 km/week to >50
km/week. These definitions may be viewed more as guidelines as one footwear expert
mentioned that “Even elite athletes perform training runs with different intensities, durations, on
different surfaces and so on. For each of these runs they might select a different type of
footwear.” This comment touches on the competing requirements for running shoes and there
may be multiple “correct” shoes for a given running level, especially in the high caliber
category.” (lines 362-377)
Reviewer #3: General
The paper is well written and the study uses appropriate methodology for reaching consensus
regarding standards for classifying runners as well as for recommendations for running
footwear.
>>Thank you for your compliments and suggestions.
One major concern that I have is that while the data was collected anonymously, the country
and region of the country is provide din the raw data. This information along with the
acknowledgment to specific participants, makes it quite easy to identify the responses of many
of the participants in the raw data. The country and region data collected in the survey needs to
be deleted to de-identify the data and preserve anonymity of the participants responses.
>>We have de-identified the raw data by removing the country and region for each participant.
Another concern I have is the use of a manuscript in review as a major reference for this study.
The Hoitz et al, manuscript that is listed as in review is not available to the reviewers of the
current manuscript. As such it is difficult to discern how the current manuscript contributes to the
literature. Moreover, depending on when or if the Hoitz, et al manuscript is accepted, it may not
be available to the readers of the current manuscript. It would be acceptable to reference a
manuscript that has been accepted and is in press.
>> We have removed the citation in question (as the mentioned manuscript is still in review) and
replaced it with the following: Sun, X, Lam, WK, Zhang X., Wang J, & Fu W (2020). Systematic
Review of the Role of Footwear Constructions in Running Biomechanics: Implications for
Running-Related Injury and Performance. Journal of Sports Science and Medicine,19, 20-37
Minor
Line 111: the phase “reached out to”, is awkward perhaps “contacted” or similar
>>We have updated the phrasing as recommended. (line 112)
Table 3 or discussion of runner classification. While consensus was reached on runner
classification, was consensus reached on how to classify runners who may meet standards
across categories (e.g. run at novice speed but with the habit or experience of recreational
runners). For example, for a runner to be in a category do they have to meet 4 of the 5
categories or … ?
>>While we did not specify how many criteria had to be fulfilled in order to decide the runner’s
category at the beginning of the survey, we acknowledge your points and added the following to
the limitation section:
“A limitation of the consensus process for the running level definitions was that we did not
specify to the experts how many of the of the categories a runner must match to be considered
a “novice”, “recreational”, or “high caliber” runner. As such, the definitions may lead to minor
variations when different footwear experts categorize runners.” (lines 408-411)
Table 6. I re-read the methods paragraph describing the manner of reaching consensus multiple
times, lines 181-194. I also read the results paragraph regarding shoe properties, lines 283 to
293, multiple times. However, it is not clear to be which specific variables qualified to be
presented in table 6.
>>We have eliminated Table 6 and added the “% Participants in agreement with consensus”
column from Table 6 to Table 5.
| Shoe feature recommendations for different running levels: A Delphi study. | 07-16-2020 | Honert, Eric C,Mohr, Maurice,Lam, Wing-Kai,Nigg, Sandro | eng |
PMC8609846 | Baygutalp et al.
BMC Sports Science, Medicine and Rehabilitation (2021) 13:145
https://doi.org/10.1186/s13102-021-00375-0
RESEARCH
Impacts of different intensities of exercise
on inflammation and hypoxia markers in low
altitude
Fatih Baygutalp1*, Yusuf Buzdağlı2, Murat Ozan3, Mitat Koz4, Nurcan Kılıç Baygutalp5 and Gökhan Atasever6
This study will be presented as an oral presentation in Eastern Black Sea Rheumatology Days on 18-19
December 2021 in Turkey.
Abstract
Background: This study aims to determine and compare the effects of exercise modalities with different intensities
on the secretion of key inflammation and hypoxia markers in amateur athletes.
Methods: Twenty-three athletes with a mean age of 20.1 years, living at low altitude (1850 m) participated in this
study. The participants’ maximal oxygen consumption values (VO2 max) were determined with an incremental cycle
exercise test as 54.15 ± 6.14 mL kg min−1. Athletes performed four protocols: at rest, 50% VO2 max, 75% VO2 max
and 100% VO2 max (until exhaustion) with one-week intervals. 50% VO2 max, 75% VO2 max sessions were performed
continuously for 30 min on a bicycle ergometer and 100% VO2 max session was performed by cycling until exhaus-
tion. Blood samples were obtained at rest and immediately after each exercise session. Serum tumor necrosis factor
alpha (TNF-α), C-reactive protein (CRP), interleukin-10 (IL-10), and hypoxia inducible factor-1 alpha (HIF-1α) levels were
measured.
Results: There were significant differences in serum TNF-α levels in 75% VO2 max and 100% VO2 max sessions
(489.03 ± 368.37 and 472.70 ± 365.21 ng/L, respectively) compared to rest conditions (331.65 ± 293.52 ng/L). Serum
CRP levels of 50% VO2 max and 75% VO2 max sessions (1.19 ± 0.50; 1.07 ± 0.52 mg/L) were significantly higher
than the rest condition (0.74 ± 0.35 mg/L). There were significant differences in serum IL-10 levels of rest condition
and 50% VO2 max; 50% VO2 max, and 100% VO2 max sessions (328.09 ± 128.87; 446.36 ± 142.84; 347.44 ± 135.69;
324.88 ± 168.06 pg/mL). Serum HIF-1α levels were significantly higher in 75% VO2 max session compared to rest
(1.26 ± 0.16; 1.08 ± 0.19 ng/mL) (P < 0.05 for all comparisons).
Conclusions: Both inflammatory and anti-inflammatory pathway is induced on different exercise intensities. Exer-
cise protocols performed until exhaustion may lead to activation of inflammatory pathways and hypoxia-induced
damage.
Keywords: Anti-inflammatory cytokine, Exercise, Health, Hypoxia, Pro-inflammatory cytokine
© The Author(s) 2021. Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which
permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the
original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or
other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line
to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory
regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this
licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/. The Creative Commons Public Domain Dedication waiver (http:// creat iveco
mmons. org/ publi cdoma in/ zero/1. 0/) applies to the data made available in this article, unless otherwise stated in a credit line to the data.
Background
The optimal exercise type and intensity should be care-
fully determined, particularly in acute exercise protocols.
While moderate exercise generally improves immune
function, excessive amounts of prolonged, high-intensity
Open Access
*Correspondence: [email protected]
1 Department of Physical Medicine and Rehabilitation, Ataturk University
Faculty of Medicine, Erzurum, Turkey
Full list of author information is available at the end of the article
Page 2 of 9
Baygutalp et al. BMC Sports Science, Medicine and Rehabilitation (2021) 13:145
exercise may lead to impairments in immune function
[1]. There are promising results from comparison stud-
ies demonstrating the equality or superiority of high-
intensity intermittent training (HIIT) programs over
low-intensity regular exercise programs based on car-
diorespiratory and metabolic parameters [2–4]. For this
reason, there is a growing interest in the studies with
high-intensity intermittent training protocols since this
training type is considered helpful for people who cannot
exercise regularly, and HIIT protocols are time efficient
[5, 6].
Exercise and intense training affect hormonal release,
creating adaptive responses that will facilitate the organ-
ism to cope with exercise stress [7]. Further, exercise
may be considered as a medicine against metabolic syn-
drome. The inflammatory response is the body’s reaction
and defence against homeostasis disorders, particularly
infection and injury [8]. It has long been recognized that
exercise is related to anti-inflammatory pathways [9, 10].
However, the pro-inflammatory pathway may be acti-
vated, in eccentric exercise protocols [11, 12]. Therefore,
optimal exercise protocol should be used for athletes and
sedentary people to improve outcomes and prevent mus-
culoskeletal damages, cardiovascular, neurological, and
endocrinological side effects [13, 14].
The release of pro-inflammatory cytokines and anti-
inflammatory cytokines into the circulation in response
to exercise varies according to exercise type, duration,
and intensity [1, 13].
Tumor necrosis factor alpha (TNF-α) is an essen-
tial mediator of the acute inflammatory response [15].
Interleukin-10 (IL-10) is one of the most important anti-
inflammatory cytokines [5]. Increased serum TNF- α
and C-reactive protein (CRP) levels and decreased IL-10
levels can be regarded as typical signs of a pro-inflamma-
tory state [9]. In addition, the IL-10/TNF-α ratio can be
used as an indicator of the beneficial effects of exercise
[16]. Many studies investigate the inflammatory response
in exercise [11, 12] or training sessions [9, 17], and the
results are inconsistent. These inconsistencies may arise
from various exercise or training protocols (type, dura-
tion and intensity), blood sampling timing, lack of con-
trol group, small sample sizes, ethnicity and biological
variations.
CRP is the main acute phase protein in tissue damage
and other inflammatory conditions and is a sensitive and
objective marker [18]. As a result of a systematic review
on CRP, it was found that in trained athletes, when a
single exercise protocol was applied, CRP temporarily
increased as the acute phase response after exercise. In
contrast, those who did higher levels of physical activity
in longitudinal studies had lower CRP levels. In this con-
text, although physical activity has been found to raise
the CRP level acutely, it has been found that chronically
physical activity reduces CRP levels [19].
Hypoxia is one of the stress factors that can promote
the inflammation process, including pro-inflammatory
and anti-inflammatory pathways. Also inflammatory tis-
sues usually become hypoxic [20]. Hypoxia or decreased
oxygen levels result in many changes at the cellular
level, including mitochondrial biogenesis and angio-
genesis in rest [21, 22] and exercise conditions [23].
Peroxisome
proliferator-activated
receptor-gamma
coactivator 1 alpha (PGC-1α), hypoxia-inducible fac-
tor-1 alpha (HIF-1α) and vascular endotheial growth
factor (VEGF) play key roles in these adaptation mecha-
nisms. HIF-1α is a hypoxia-induced transcription factor
that transcribes more than 100 enzymes and proteins
involved in cellular responses caused by hypoxia [24].
There is a relationship between the body’s response
to inflammation or exercise stress and its response to
hypoxia, and in both cases, the hypoxia-inducible factor-
1a (HIF-1a) signaling pathway can be induced [25]. It is
known that both exercise and hypoxia can alter mRNA
expression and protein release of pro-and anti-inflam-
matory cytokines, activate lymphocytes, alter chemokine
receptors, or induce other signaling pathways of the
hypoxic inflammatory response [26, 27].
This study was conducted in a low altitude (1850 m)
city [28], which is the highest city of Turkey. People living
in altitudes higher than sea level have reduced hypoxic
ventilatory response, decreased pulmonary hyperten-
sion under hypoxia, increased heart rate, and improved
peripheral oxygen saturation [29]. The fact that the
hypoxia and inflammatory responses were evaluated on
different exercise intensities and that the study was con-
ducted in a low altitude region is prominent in our study.
For these reasons, in this study, the acute effects of exer-
cise intensity on hypoxia and inflammatory responses in
an amateur athlete group living in a low altitude region
were investigated. The combined effects of hypoxia
marker and exercise on inflammatory pathways were
assessed.
Methods
Twenty-three amateur male athletes (soccer) living at
low altitude (1850 m), training 2 h/day, 5 days/week,
were included in this study. Inclusion criteria were; to
be an amateur athlete between the ages of 18–22, male
gender, living in this location for at least 5 years, vol-
unteering to participate in research being healthy and
not having a chronic or acute illness. Exclusion criteria
were; having any chronic disease, using any medication
or stimulants, smoking and alcohol use, having limitation
of movement due to injury for any reason. Twenty-three
Page 3 of 9
Baygutalp et al. BMC Sports Science, Medicine and Rehabilitation (2021) 13:145
people who met the inclusion criteria were included to
the study. Demographical characteristics of athletes are
given in Table 1. Athletes were evaluated at rest and at
three different exercise intensities: 30 min of exercise on
a bicycle ergometer at 50% of the predetermined maxi-
mal oxygen consumption capacity (VO2 max) values, 75%
of the predetermined VO2 max and exercise at 100% VO2
max-until the individual is exhausted. Venous blood sam-
ples were taken at rest state (1st session) and immediately
after each exercise sessions (2nd, 3rd and 4th sessions).
Ethical ıssues
The informed consent form was obtained from all partic-
ipants, and they were enlightened with all matters related
to the study. The study was approved by the Clinical
Research Ethics Committee of Ataturk University Faculty
of Medicine (27.05.2021).
Study design
The study design is summarized in Fig. 1. 2nd session can
be defined as mild intensity exercise (50% for 30 min),
3rd session as moderate-intensity exercise (75% for
30 min) and 4th session as high-intensity exercise (100%
to exhaustion).
Exercise sessions were carried out at one-week inter-
vals to prevent physiological adaptation.
Data collection
Demographical characteristics and height measurement
The athletes’ ages in the study were recorded based on
their identity information, and the sports ages based on
their declarations. The height of the athletes was meas-
ured with a mechanical measuring rod (Seca 216, Medis-
ave UK Co., UK).
Body composition measurement
Body composition parameters (weight and body fat per-
centage) were measured with the BOD POD body com-
position tracking system (Cosmed, USA). Body mass
index (BMI) was calculated with the following formula:
BMI = weight/height2 (kg/m2).
Maximal oxygen consumption capacity (VO2 max)
measurement (pre‑test − incremental cycle exercise test)
The participants were subjected to a maximal exercise
test with an exercise protocol with increasing intensity
following the test completion criteria used in the bicycle
Table 1 Demographical characteristics of athletes
SD: standart deviation, CI: confident interval, BMI: body mass index, BFP: body
fat percentage, VO2 max: maximal oxygen consumption
Male athletes (n = 23)
Mean ± SD
95% CI
Age (year)
20.12 ± 0.15
13.51–26.49
BMI (kg/m2)
23.44 ± 1.29
22.44–23.56
BFP
15.19 ± 7.19
11.89–18.11
Sports age (year)
10.21 ± 2.31
9.00–11.00
VO2 max (mL kg min−1)
54.15 ± 6.14
51.34–56.66
Fig. 1 Study design
Page 4 of 9
Baygutalp et al. BMC Sports Science, Medicine and Rehabilitation (2021) 13:145
ergometer protocols used for maximal oxygen consump-
tion capacity and power measurements.
Participants first cycled on the bicycle ergometer for
5 min (50–60 RPM) to warm up. Then, the warming was
completed by stretching for 2 min. As soon as the par-
ticipant is fully ready, the test is started with the start
command and the continuously increasing load test is
applied. The participant started the test at 60 revolutions
per minute and by pedaling at 150 watts. Then, 30 watts
were increased every 2 min, and the trial continued until
the test pedal speed fell below 50 revolutions per minute
or the subject could not continue anymore. The Rating
of Perceived Exertion (RPE) was defined with the Borg
scale to determine the VO2 max of the participants [30].
Participant-reported Borg scale scores were used at each
load-increasing phase of the exercise test in the first ses-
sion they visited the laboratory and immediately after
each session.
The observation of three of the criteria simultaneously
was accepted to indicate that the maximal oxygen use
capacity was reached, then the test was terminated. The
criteria were; reporting a Borg scale score of 20 by the
participant, the oxygen consumption does not increase
despite the increase in workload, the ratio of carbon diox-
ide production to oxygen consumption RER (respiratory
exchange ratio) reaches 1.15 and above, the heart rate is
85% and above the maximum number of heart rates, the
increase in the number of heart rates despite the increas-
ing workload [31]. In the gas analysis, minute ventila-
tion (VE), oxygen volume per minute (VO2), produced
carbon dioxide volume (VCO2) per minute were directly
measured and recorded. At the same time, the heart rate
(HRmax) at which the athletes reached the maximal oxy-
gen use and the perceived difficulty values at each step
of increase were also recorded. Before the measurement
sessions, the Cosmed K5 oxygen analyzer was calibrated
with high-grade calibration gases provided by the man-
ufacturers. Gas was pumped from the flow meters with
a 3-L calibration syringe following the manufacturer’s
recommendations and heated for a minimum of 15 min.
Mask size was determined individually before the first
test, and measurements were taken with the same size in
subsequent sessions.
Exercise protocols
All of the test protocols were carried out in our univer-
sity’s Sports performance laboratory and measurement
center. Athletes were not allowed to perform vigorous
exercise, using drugs, caffeine, alcohol and performance-
enhancing ergogenic supplements from 48 h before
exercise protocols. Before starting exercise protocols
(pre-test), the participants’ maximal oxygen consumption
values (VO2 max) were determined using the oxygen
analyzer K5 (Cosmed, USA) as a pre-test with the gradu-
ally increasing load exercise test on the bicycle ergom-
eter. Participants were required to cycle continuously
for 30 min on a bicycle ergometer at 50% and 75% of the
predetermined maximal oxygen consumption capacity
values.
Finally the participants were required to cycle at
100% VO2max until exhaustion.The term 100% VO2
max defines the situation when the participant reaches
exhaustion during exercise. [32]. The mean time to
exhaustion of athletes was 16.35 ± 3.38 min. Venous
blood samples were taken immediately after each session.
All measurements were taken at the same time of day
(morning time).
Biochemical analysis
5 mL of venous blood samples were taken from each
athlete. After the serum was obtained, the samples were
aliquoted and stored at -80 C° until analysis. Serum CRP,
TNF-α, interleukin-10 and HIF-1α levels in all samples
were analyzed by ELISA method with commercial kits
(Bioassay Technology Laboratory-BT Lab, China pro-
duced all kits). Samples were collected once and meas-
ured duplicated.
Sample size calculation
The minimum number of patients required for the study
was calculated in the G Power sample calculation pro-
gram (version 3.1.9.4) at the level of Type I error (α) 0.05)
and Type II error (1-β) 0.95, with an effect size (Cohen’s
f) of 0.4 (large) for a priori calculation of ANOVA test for
4 repeated groups. Accordingly, the minimum number of
samples was determined as 16. We included 23 partici-
pants to the study, in order to prevent a limitation caused
by small sample size.
Statistical analyses
Statistical analysis was performed in SPSS 23.0 pack-
age program. Kolmogorov–Smirnov test was used to
determine the normality of data. Descriptive statistical
analysis, repeated measures ANOVA test, and Pearson
correlation analysis were performed. Data were pre-
sented as mean ± SD (standard deviation). Kolmogo-
rov–Smirnov test revealed that data were distributed
normally, and repeated measures ANOVA test was used
to compare biochemical values of different sessions. Val-
ues of P < 0.05 at a 95% confidence interval were consid-
ered statistically significant. Eta-squared value (η2) was
used to determine effect sizes within the ANOVA calcu-
lation. η2 values of 0.01, 0.06, and 0.14 were interpreted
as “small”, “medium” and “large” effect sizes, respectively.
Page 5 of 9
Baygutalp et al. BMC Sports Science, Medicine and Rehabilitation (2021) 13:145
Results
Serum IL-10, TNF-α, CRP and HIF-1α values obtained
at rest conditions and different exercise sessions are
given in Table 2. Additionally, IL-10/TNF-α ratio was
used as a positive predictor of exercise and presented the
results in Table 2. Results show that IL-10/TNF-α ratio
was decreased in 100% VO2 max session compared to
both rest and 50% VO2 sessions (P = 0.008 and P = 0.041,
respectively).
The pairwise comparisons of biochemical values
between rest state and different sessions were per-
formed with the repeated measures ANOVA test, and
the results are summarized in Table 2. Results showed
significant differences in serum TNF-α levels between
rest condition and 75% VO2 max; rest and 100% VO2
max session. There were significant differences in
serum CRP levels between rest and 50% VO2 max; rest
and 75% VO2 max sessions. There were significant dif-
ferences in serum IL-10 levels between rest and 50%
VO2 max, 50% VO2 max, and 100% VO2 max sessions.
There were significant differences in serum HIF-1α lev-
els between rest and 75% VO2 max session (P < 0.05 for
all comparisons). All other comparisons were not sta-
tistically significant (P > 0.05 for all other pairs). The
alterations in pro-inflammatory and anti-inflammatory
pathways are shown in Fig. 2 with the results of IL-10
and TNF-α.
Pearson correlation analyses were performed to evalu-
ate the relationships between biochemical parameters in
rest conditions and each exercise session. Results showed
a high negative correlation between serum HIF-1α
and TNF-α levels on 50% VO2 max session (r: − 0.634,
P = 0.003). There was a moderate positive correlation
between serum HIF-1α and IL-10 levels at 75% VO2 max
session (r: 0.593, P = 0.006) (Fig. 3).
Correlation analysis showed that serum HIF-1α levels
were negatively related to serum TNF-α levels and posi-
tively related to serum IL-10 levels. Changes in HIF-1α
concentrations during exercise may have negatively
affected the pro-inflammatory pathway and positively
affected the anti-inflammatory pathway as a protection
mechanism.
Discussion
In this study, the acute effects of different exercise inten-
sities on serum IL-10, TNF-α, CRP and HIF-1α levels
were reported for the first time. Additionally, the study
was conducted in a low altitude (1850 m) city. Exercise
practice until exhaustion caused significant pro-inflam-
matory effects (demonstrated with TNF-α) and the
optimal IL-10 response on 50% VO2 max decreased to
nearly baseline level as the exercise intensity reached to
100% VO2 max. Thus, we can suggest that exercise inten-
sity should not reach to exhaustion due to there was no
improvement in the anti-inflammatory marker IL-10 and
there was an increment in the pro-inflammatory marker
TNF-α with the potential increase in inflammation.
There is a high altitude camping center for athletes in our
Table 2 Biochemical values of athletes
η2: Eta-squared value
a,b,c,d Show repeated measures ANOVA Bonferroni post-hoc test P values
a Significant difference (P < 0.05) between rest state and 50% VO2 max session
b Significant difference (P < 0.05) between rest state and 75% VO2 max session
c Significant difference (P < 0.05) between rest state and 100% VO2 max session
d Significant difference (P < 0.05) between 50% VO2 and 100% VO2 max sessions
Rest state
50% VO2 max
75% VO2 max
100% VO2 max
η2
IL-10 (pg/mL)
328.09 ± 128.87a
446.36 ± 142.84d
347.44 ± 135.69
324.88 ± 168.06
0.546
TNF-α (ng/L)
331.65 ± 293.52b,c
395.59 ± 319.82
472.70 ± 365.21
489.03 ± 368.37
0.309
IL-10/ TNF-α
1.63 ± 1.20 c
1.49 ± 0.93d
1.34 ± 0.97
0.99 ± 0.67
0.566
CRP (mg/L)
0.74 ± 0.35a,b
1.19 ± 0.50
1.07 ± 0.52
0.97 ± 0.55
0.773
HIF-1α (ng/mL)
1.08 ± 0.19b
1.12 ± 0.32
1.26 ± 0.16
1.18 ± 0.21
0.453
Fig. 2 Alterations of cytokine levels in rest state and different
seessions
Page 6 of 9
Baygutalp et al. BMC Sports Science, Medicine and Rehabilitation (2021) 13:145
city, and athletes from all over the country use this center.
For this reason, the study has regional and national added
value.
Several studies have investigated the responses of pro-
inflammatory cytokines, inflammatory cytokines and
inflammatory markers to different exercise intensities
and modalities, and these studies report distinct results
[11, 13, 16, 17, 33, 34]. The conflicting results from pre-
vious literature may arise from differences in exercise
intensities, exercise modalities, VO2 max capacities, the
timing of blood sampling and biological variations.
In a study conducted with 20 soccer players with a
mean age of 25.75 ± 3.99 years, participants were sub-
jected to a single bout high high-intensity interval train-
ing and plasma IL-6, IL-1 and TNF- α levels determined
before and immediately after training. There was a sig-
nificant increase in plasma interleukin-6 levels after exer-
cise; however, no significant increase in IL-1 and TNF- α
levels showing an anti-inflammatory condition might
occur through high-intensity interval training sessions
[34]. The training protocol of this study and the exer-
cise protocol of the current research is different, and it’s
known that metabolic changes may occur differently in
training and exercise. However, the type of sports and
age of the participants in the two studies are similar.
We observed increments in both anti-inflammatory and
pro-inflammatory cytokines; anti-inflammatory marker
(IL-10) was increased at 50% VO2 session, and pro-
inflammatory marker (TNF-α) was increased 75% VO2
and 100% VO2 sessions, and inflammatory (CRP) marker
was increased at 50% VO2 and 75% VO2 sessions. We
used high-intensity exercises, and researchers have used
high-intensity interval training (HIIT). We could not
show the anti-inflammatory effects of high intensity exer-
cise in our study, although other researchers have shown
the anti-inflammatory effects in their study using high-
intensity interval training (HIIT).
The TNF-α level is decreased by moderate exercise
(exercise intensity HRmax 60–70%) [33], and mRNA
expression of TNF-α is known to be slightly elevated in
skeletal muscle by endurance exercise [35]. In the previ-
ous study, a gradient increment was observed in serum
TNF-α levels as exercise intensity increases. The high-
est TNF-α response to exercise was found at 100% VO2
max session when the athlete presents his maximum
endurance.
A systematic review including 18 articles investigat-
ing the effects of moderate and intense exercise on
inflammatory response concluded that intense long
exercise protocols might activate pro-inflammatory
pathways. Instead of this, moderate or high-intensity
intermittent exercise protocols with suitable rest condi-
tions may be preferred [1]. We are in line with this con-
clusion since we observed an inflammatory profile by
determining high TNF- α and CRP levels in 75% VO2
max and 100% VO2 max sessions and optimal IL-10
concentration at 50% VO2 max session. Although there
is evidence of minimal pro-inflammatory cytokine
response and high anti-inflammatory cytokine release
from a study conducted on athletes competing in an
ironman triathlon race [36], it should be considered
that triathlon race is a type of ultra-endurance exer-
cise. We suggest IL-10 levels were not increased as
expected at 75% VO2 max and 100% VO2 max exercise
Fig. 3 Scatter-dot graphs of Pearson correlation analysis. A Significant negative correlation between HIF-1α and TNF-α at 50% VO2 max session; (r:
− 0.634, P = 0.003). B Significant positive correlation between HIF-1α and IL-10 at 75% VO2 max session (r: 0.593, P = 0.006).
Page 7 of 9
Baygutalp et al. BMC Sports Science, Medicine and Rehabilitation (2021) 13:145
intensities because of pro-inflammatory effects. CRP
level partially supported this suggestion, being signifi-
cantly higher in the 75% VO2 max session than the rest
state. Among studies investigating the impact of exer-
cise on CRP release, most of them reveal increased
CRP levels immediately after moderate [37]and intense
exercise [38]. Yet, a study reports no effect of exercise
modality on acute CRP response [39]. As determined
the highest CRP value in 75% VO2 max session and
elevated values in 75% VO2 max session compared to
rest state in the present study, we can conclude that
CRP partly acts together with the pro-inflammatory
pathways. However, we could not determine any signifi-
cant correlation between CRP and TNF- α. Although
we determined optimal IL-10 levels and relatively low
TNF- α levels (compared to 75% VO2 max and 100%
VO2) at 50% VO2 max session, we can not recommend
using this intensity to athletes since this intensity is
not related to training for fitness improvements/adap-
tations, and as well as for soccer as the participants in
the current investigation were indeed soccer players.
Further, 50% intensity will not be adequate to stress the
body to induce an adaptation. Acute exercise sessions
lead to a complex cascade of inflammatory and pro-
inflammatory pathways [40–42]. We are in line with
this conclusion with altered TNF-α, IL-10, and CRP
levels among sessions.
Considering the current study results and related stud-
ies, we can speculate that moderate intensity exercise
with durations longer than 30 min (providing higher
endurance than the present study) may be beneficial to
prevent/reduce pro-inflammatory response.
It is known that disease-induced hypoxia is closely
related to the activation of inflammatory pathways, but
less information is available about the effects of exercise-
induced hypoxia on inflammation. There is a relationship
between hypoxia and the release of pro-inflammatory
cytokines. Moreover, HIF-1α is important in control-
ling excessive inflammation [23]. Also, hypoxia, inflam-
mation, and exercise can induce the HIF-1α pathway. It
was shown that skeletal muscle HIF-1 protein content
increased by 120% with hypoxia, and HIF-1α released in
response to hypoxia was triggered by the effect of exer-
cise [43]. In the present study, serum HIF-1α levels were
significantly increased in 75% VO2 max session com-
pared to rest state in a high-lander athlete population liv-
ing in this location for at least 5 years.
At hypoxia conditions in the exercising person, the
inflammatory pathways are regulated differently. The
hypoxic and exercise stimuli are stronger in vivo than the
hypoxic or inflammatory stimuli isolated in vitro [24].
Of note, when considering HIF-1α results, it should
be kept in mind that high interindividual variability may
be seen in the expression of HIF and its target genes in
response to inflammatory or hypoxic stimuli, and sin-
gle nucleotide polymorphisms (SNPs) are thought to be
involved in these changes [25].
There is a relation between hypoxia and inflammation.
Hypoxia can induce inflammation, and inflamed tissues
may become hypoxic [20]. Limited studies investigat-
ing the effects of exercise on inflammatory pathways in
hypoxic conditions revealed no changes in pro-inflam-
matory cytokines, and increases in anti-inflammatory
cytokines, indicating the positive effects of exercising in
hypoxic conditions.
The triple relation of exercise, inflammatory pathways,
and oxygen consumption in a low altitude location were
investigated in a previous study. HIF-1 α response is
maximum on 75% VO2 max session and decreases from
this maximum value on exhaustion. This result follows
the finding that high-intensity exercise in hypoxia can
further induce HIF-1α expression [43]. It is well known
that high-lander athletes show better exercise perfor-
mance and greater VO2 max capacity than sea-landers
since athletes have adapted to hypoxia, and maybe some
have a genetic basis, thanks to the effect of altitude [44,
45].
Although there is no agreement to define the term
“high-intensity”, it widely refers to exercise intensity
higher than 75% VO2 max [23]. A speculative model
suggests that HIF-1α and PGC-1α act as mediators in
the adaptation of skeletal muscle. The mediators lead to
upregulation of mitochondrial biogenesis, angiogenesis
via activation of VEGF and a shift in the skeletal mus-
cle fibre type. Both high-intensity exercise/training and
hypoxia lead to this mechanism to upregulate skeletal
muscle adaptation [23].
We observed the optimum HIF-1 α response in a 75%
VO2 max session in the present study.
HIF-1 α response did not increase when the exercise
intensity was reached from 75% VO2 max to 100% VO2
max in the present study. We attribute this because the
athletes have developed a physiological adaptation to
hypoxia thanks to living at low altitudes. Correlation
analyses revealed a high negative correlation between
serum HIF-1α and TNF-α levels on 50% VO2 max ses-
sion (r: − 0.634, P = 0.003) and a moderate positive cor-
relation between serum HIF-1α and IL-10 levels at 75%
VO2 max session (r: 0.593, P = 0.006). Results suggest
that increased HIF-1α levels reflect the pro-inflammatory
condition in 50% VO2 max session and the anti-inflam-
matory condition in 75% VO2 max session. We deter-
mined maximum TNF-α response and similar IL-10
response compared to baseline in 100% VO2 max ses-
sion. We can conclude that the pro-inflammatory effects
of hypoxia and anti-inflammatory effects of the exercise
Page 8 of 9
Baygutalp et al. BMC Sports Science, Medicine and Rehabilitation (2021) 13:145
was probably due to activating the release of anti-inflam-
matory cytokines and downregulating toll-like receptor
(TLR) signalling [23].
Studies with different exercise protocols have shown
that high-intensity exercise (above 75% of the peak
power output) provides similar or even higher benefits
than a low-intensity continuous exercise in improving
heart health, respiratory health, and metabolic health.
Increases in peak power outputs during exercise result
in increased metabolic responses, compromising skel-
etal muscle integrity, which can cause early onset of
fatigue and exhaustion. Therefore, the selection of exer-
cise intensity should be made carefully to avoid undesir-
able consequences. Taken together, TNF-α, IL-10, CRP,
and HIF-1α results, we again suggest that exercise inten-
sity should not reach to exhaustion. Despite its original-
ity, the current study has a limitation. It would be better
ELISA results should be supported with western blotting
analysis and mRNA expression levels of proteins.
Conclusions
There is a tight connection between hypoxia and inflam-
mation, and studies investigating the effects of exercise
intensity in hypoxic and inflammatory pathways are
limited. There is no available study in any athletic popu-
lation reporting the acute changes on serum IL-10, TNF-
α, CRP and HIF-1α levels induced by different exercise
intensities. We noted that both inflammatory and anti-
inflammatory pathway is induced on different exercise
intensities. As the need for oxygen increases, the inflam-
matory pathway (by TNF- α and CRP) is induced, and
anti-inflammatory cytokine IL-10 reaches optimal value
on exercise intensity of 50% VO2 max. Exercise regimens
(not reached to exhaustion) are recommended to prevent
inflammation, hypoxia-induced damage, and existing
muscle damage progression if any. Further studies on dif-
ferent athlete groups should be conducted to determine
the optimum exercise intensity and maximum benefit.
Abbreviations
VO2 max: Maximal oxygen consumption values; TNF-α: Tumor necrosis factor
alpha; IL-10: Interleukin-10; CRP: C-reactive protein; HIF-1 α: Hypoxia inducible
factor-1 alpha.
Acknowledgements
The authors would like to thank the athletes that took part in this study. The
authors would like to thank to Prof. Mostafa Abdelaty HASSIBELNABY for his
scientific contribution to the study.
Authors’ contributions
FB: concept, design, inspection of all participants, writing article, critical review
of the article; YB: performing exercise protocol, writing article, critical review
of the article; MO: performing exercise protocol, writing article, critical review
of the article; MK: critical review of the article; NKB: biochemical analysis,
statistical analysis, critical review of the article; GA: determining exercise
protocol, critical review of the article. All authors read and approved the final
manuscript.
Funding
None.
Availability of data and materials
The datasets used and/or analyzed during the current study are available from
the corresponding author on reasonable request.
Declarations
Ethics approval and consent to participate
The informed consent form was obtained from all participants, and they were
enlightened with all matters related to the study. The study was approved
by the Clinical Research Ethics Committee of Ataturk University Faculty of
Medicine (27.05.2021).
Consent for publication
Not applicable.
Competing interests
Not applicable.
Author details
1 Department of Physical Medicine and Rehabilitation, Ataturk University
Faculty of Medicine, Erzurum, Turkey. 2 Department of Physical Education
and Sports, Erzurum Technical University Faculty of Sport Sciences, Erzurum,
Turkey. 3 Department of Physical Education and Sports, Ataturk University
Kazım Karabekir Education Faculty, Erzurum, Turkey. 4 Department of Sports
Health Sciences, Ankara University Faculty of Sport Sciences, Ankara, Turkey.
5 Department of Biochemistry, Ataturk University Faculty of Pharmacy, Erzu-
rum, Turkey. 6 Department of Recreation, Ataturk University Faculty of Sport
Sciences, Erzurum, Turkey.
Received: 25 June 2021 Accepted: 15 November 2021
References
1.
Cerqueira É, Marinho DA, Neiva HP, Lourenço O. Inflammatory effects of
high and moderate intensity exercise—a systematic review. Front Physiol.
2019;10:1550.
2.
Beauchamp MK, Nonoyama M, Goldstein RS, Hill K, Dolmage TE,
Mathur S, et al. Interval versus continuous training in individuals with
chronic obstructive pulmonary disease–a systematic review. Thorax.
2010;65(2):157–64.
3.
Rognmo Ø, Moholdt T, Bakken H, Hole T, Mølstad P, Myhr NE, et al.
Cardiovascular risk of high- versus moderate-intensity aerobic exercise in
coronary heart disease patients. Circulation. 2012;126(12):1436–40.
4.
Nicolò A, Bazzucchi I, Haxhi J, Felici F, Sacchetti M. Comparing continuous
and intermittent exercise: an “isoeffort” and “isotime” approach. PLoS ONE.
2014;9(4):e94990.
5.
Weston KS, Wisløff U, Coombes JS. High-intensity interval training in
patients with lifestyle-induced cardiometabolic disease: a systematic
review and meta-analysis. Br J Sports Med. 2014;48(16):1227–34.
6.
Thum JS, Parsons G, Whittle T, Astorino TA. High-intensity interval training
elicits higher enjoyment than moderate intensity continuous exercise.
PLoS ONE. 2017;12(1):e0166299.
7.
Gleeson M. Immune function in sport and exercise. J Appl Physiol (1985).
2007;103(2):693–9.
8.
Medzhitov R. Origin and physiological roles of inflammation. Nature.
2008;454(7203):428–35.
9.
Gleeson M, Bishop NC, Stensel DJ, Lindley MR, Mastana SS, Nimmo MA.
The anti-inflammatory effects of exercise: mechanisms and implica-
tions for the prevention and treatment of disease. Nat Rev Immunol.
2011;11(9):607–15.
10. Pedersen BK. Anti-inflammatory effects of exercise: role in diabetes and
cardiovascular disease. Eur J Clin Invest. 2017;47(8):600–11.
Page 9 of 9
Baygutalp et al. BMC Sports Science, Medicine and Rehabilitation (2021) 13:145
•
fast, convenient online submission
•
thorough peer review by experienced researchers in your field
•
rapid publication on acceptance
•
support for research data, including large and complex data types
•
gold Open Access which fosters wider collaboration and increased citations
maximum visibility for your research: over 100M website views per year
•
At BMC, research is always in progress.
Learn more biomedcentral.com/submissions
Ready to submit your research
Ready to submit your research ? Choose BMC and benefit from:
? Choose BMC and benefit from:
11. Fielding RA, Manfredi TJ, Ding W, Fiatarone MA, Evans WJ, Cannon JG.
Acute phase response in exercise. III. Neutrophil and IL-1 beta accumula-
tion in skeletal muscle. Am J Physiol. 1993;265(1 Pt 2):R166–72.
12. O’Reilly KP, Warhol MJ, Fielding RA, Frontera WR, Meredith CN, Evans WJ.
Eccentric exercise-induced muscle damage impairs muscle glycogen
repletion. J Appl Physiol (1985). 1987;63(1):252–6.
13. Nobari H, Cholewa JM, Pérez-Gómez J, Castillo-Rodríguez A. Effects of
14-weeks betaine supplementation on pro-inflammatory cytokines and
hematology status in professional youth soccer players during a competi-
tion season: a double blind, randomized, placebo-controlled trial. J Int
Soc Sports Nutr. 2021;18(1):42.
14. Nobari H, Vahabidelshad R, Pérez-Gómez J, Ardigò LP. Variations of train-
ing workload in micro- and meso-cycles based on position in elite young
soccer players: a competition season study. Front Physiol. 2021;12:668145.
15. Zelová H, Hošek J. TNF-α signalling and inflammation: interactions
between old acquaintances. Inflamm Res. 2013;62(7):641–51.
16. Batista ML Jr, Rosa JC, Lopes RD, Lira FS, Martins E Jr, Yamashita AS, et al.
Exercise training changes IL-10/TNF-alpha ratio in the skeletal muscle of
post-MI rats. Cytokine. 2010;49(1):102–8.
17. Ogawa K, Sanada K, Machida S, Okutsu M, Suzuki K. Resistance exercise
training-induced muscle hypertrophy was associated with reduc-
tion of inflammatory markers in elderly women. Mediat Inflamm.
2010;2010:171023.
18. Tang JH, Gao DP, Zou PF. Comparison of serum PCT and CRP levels in
patients infected by different pathogenic microorganisms: a systematic
review and meta-analysis. Braz J Med Biol Res. 2018;51(7):e6783.
19. Kasapis C, Thompson PD. The effects of physical activity on serum
C-reactive protein and inflammatory markers: a systematic review. J Am
Coll Cardiol. 2005;45(10):1563–9.
20. Eltzschig HK, Carmeliet P. Hypoxia and inflammation. N Engl J Med.
2011;364(7):656–65.
21. Carmeliet P, Dor Y, Herbert JM, Fukumura D, Brusselmans K, Dewerchin M,
et al. Role of HIF-1alpha in hypoxia-mediated apoptosis, cell proliferation
and tumour angiogenesis. Nature. 1998;394(6692):485–90.
22. Semenza GL. Regulation of mammalian O2 homeostasis by hypoxia-
inducible factor 1. Annu Rev Cell Dev Biol. 1999;15:551–78.
23. Li J, Li Y, Atakan MM, Kuang J, Hu Y, Bishop DJ, et al. The molecular adap-
tive responses of skeletal muscle to high-intensity exercise/training and
hypoxia. Antioxidants (Basel). 2020;9(8):656.
24. Kammerer T, Faihs V, Hulde N, Stangl M, Brettner F, Rehm M, et al.
Hypoxic-inflammatory responses under acute hypoxia: in vitro experi-
ments and prospective observational expedition trial. Int J Mol Sci.
2020;21(3):1034.
25. Lindholm ME, Rundqvist H. Skeletal muscle hypoxia-inducible factor-1
and exercise. Exp Physiol. 2016;101(1):28–32.
26. Choi S, Chung H, Hong H, Kim SY, Kim SE, Seoh JY, et al. Inflammatory
hypoxia induces syndecan-2 expression through IL-1β-mediated FOXO3a
activation in colonic epithelia. FASEB J. 2017;31(4):1516–30.
27. Dai T, Hu Y, Zheng H. Hypoxia increases expression of CXC chemokine
receptor 4 via activation of PI3K/Akt leading to enhanced migra-
tion of endothelial progenitor cells. Eur Rev Med Pharmacol Sci.
2017;21(8):1820–7.
28. Bärtsch P, Saltin B, Dvorak J. Consensus statement on playing football at
different altitude. Scand J Med Sci Sports. 2008;18(Suppl 1):96–9.
29. Nell HJ, Castelli LM, Bertani D, Jipson AA, Meagher SF, Melo LT, et al. The
effects of hypoxia on muscle deoxygenation and recruitment in the
flexor digitorum superficialis during submaximal intermittent handgrip
exercise. BMC Sports Sci Med Rehabil. 2020;12:16.
30. Borg G. Perceived exertion as an indicator of somatic stress. Scand J
Rehabil Med. 1970;2(2):92–8.
31. Edvardsen E, Hem E, Anderssen SA. End criteria for reaching maximal oxy-
gen uptake must be strict and adjusted to sex and age: a cross-sectional
study. PLoS ONE. 2014;9(1):e85276.
32. Buchfuhrer MJ, Hansen JE, Robinson TE, Sue DY, Wasserman K, Whipp BJ.
Optimizing the exercise protocol for cardiopulmonary assessment. J Appl
Physiol Respir Environ Exerc Physiol. 1983;55(5):1558–64.
33. Amin MN, El-Mowafy M, Mobark A, Abass N, Elgaml A. Exercise-induced
downregulation of serum interleukin-6 and tumor necrosis factor-alpha
in Egyptian handball players. Saudi J Biol Sci. 2021;28(1):724–30.
34. Ghorbani P, Kordi MR, Gaeeni AA, Arbab G, Nafar MH. Changes in inflam-
matory factors in elite soccer players folllowing single bout high intensity
interval training. Braz J Biomotricity. 2013;7(1):52–8.
35. Nieman DC, Davis JM, Henson DA, Walberg-Rankin J, Shute M, Dumke
CL, et al. Carbohydrate ingestion influences skeletal muscle cytokine
mRNA and plasma cytokine levels after a 3-h run. J Appl Physiol (1985).
2003;94(5):1917–25.
36. Suzuki K, Peake J, Nosaka K, Okutsu M, Abbiss CR, Surriano R, et al.
Changes in markers of muscle damage, inflammation and HSP70 after an
Ironman Triathlon race. Eur J Appl Physiol. 2006;98(6):525–34.
37. Draganidis D, Chatzinikolaou A, Jamurtas AZ, Carlos Barbero J, Tsoukas D,
Theodorou AS, et al. The time-frame of acute resistance exercise effects
on football skill performance: the impact of exercise intensity. J Sports Sci.
2013;31(7):714–22.
38. Fatouros IG, Destouni A, Margonis K, Jamurtas AZ, Vrettou C, Kouretas
D, et al. Cell-free plasma DNA as a novel marker of aseptic inflammation
severity related to exercise overtraining. Clin Chem. 2006;52(9):1820–4.
39. Mendham AE, Donges CE, Liberts EA, Duffield R. Effects of mode and
intensity on the acute exercise-induced IL-6 and CRP responses in a sed-
entary, overweight population. Eur J Appl Physiol. 2011;111(6):1035–45.
40. Allen J, Sun Y, Woods JA. Exercise and the regulation of inflammatory
responses. Prog Mol Biol Transl Sci. 2015;135:337–54.
41. Hennigar SR, McClung JP, Pasiakos SM. Nutritional interventions and the
IL-6 response to exercise. Faseb j. 2017;31(9):3719–28.
42. Peake JM, Neubauer O, Walsh NP, Simpson RJ. Recovery of the immune
system after exercise. J Appl Physiol (1985). 2017;122(5):1077–87.
43. Van Thienen R, Masschelein E, D’Hulst G, Thomis M, Hespel P. Twin resem-
blance in muscle HIF-1α responses to hypoxia and exercise. Front Physiol.
2016;7:676.
44. Marconi C, Marzorati M, Cerretelli P. Work capacity of permanent residents
of high altitude. High Alt Med Biol. 2006;7(2):105–15.
45. Linthorne NP. Improvement in 100-m sprint performance at an altitude of
2250 m. Sports (Basel). 2016;4(2):29.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in pub-
lished maps and institutional affiliations.
| Impacts of different intensities of exercise on inflammation and hypoxia markers in low altitude. | 11-22-2021 | Baygutalp, Fatih,Buzdağlı, Yusuf,Ozan, Murat,Koz, Mitat,Kılıç Baygutalp, Nurcan,Atasever, Gökhan | eng |
PMC10651037 | PONE-D-23-25168
Dose response of running on blood biomarkers of wellness in the generally healthy
Reviewer: Dr. Subir Gupta
Upon a meticulous review of the article in question, I wish to commend the authors for crafting
a piece that not only carries immense scientific weight but is also articulated with great clarity.
Such insightful work surely merits publication in your distinguished journal. It's admirable how
the authors have navigated through a myriad of physiological and biochemical variables (blood
biomarkers) across five distinct participant categories and presented their results with lucidity.
The experimental framework is robust, the statistical evaluations are apt, and the narrative
progresses seamlessly. The references provided are both relevant and adequate. Nevertheless,
I'd like to offer a few observations and suggestions:
Original Title: “Dose response of running on blood biomarkers of wellness in the generally
healthy.”
Proposed Title: “Dose-response relationship between running and blood biomarkers of
wellness in generally healthy individuals.”
Page 2, Line 8: The mention of “exposure to sunlight” seems somewhat out of context. Could
the authors clarify its relevance or indicate if it has been discussed elsewhere in the article?
Page 17, Lines 17-18: The text reads: "These observations suggest that elite endurance
runners………to their magnesium status."
Comments: It would be helpful to clarify whether the professional athletes (PRO) participating
in this study are specifically elite endurance runners. Kindly integrate this distinction into the
main text if accurate.
Page 19, Lines 1-2: The assertion: “Indeed whether exercise………..is inconclusive,” needs
to be substantiated with a relevant citation.
Table 1: Please include standard deviation (SD) values. I also recommend expressing exercise
duration in terms of "h/week" instead of "hr".
| Dose response of running on blood biomarkers of wellness in generally healthy individuals. | 11-15-2023 | Nogal, Bartek,Vinogradova, Svetlana,Jorge, Milena,Torkamani, Ali,Fabian, Paul,Blander, Gil | eng |
PMC7435036 | Physiological Reports. 2020;8:e14551.
| 1 of 13
https://doi.org/10.14814/phy2.14551
wileyonlinelibrary.com/journal/phy2
DOI: 10.14814/phy2.14551
O R I G I N A L R E S E A R C H
Indices of leg resistance artery function are independently related
to cycling V̇O2max
Jayson R. Gifford1,2
| Brady E. Hanson1 | Meagan Proffit1,2 | Taysom Wallace1 |
Jason Kofoed1 | Garrett Griffin1 | Melina Hanson1
This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided the original
work is properly cited.
© 2020 The Authors. Physiological Reports published by Wiley Periodicals LLC on behalf of The Physiological Society and the American Physiological Society
1Department of Exercise Sciences, Brigham
Young University, Provo, UT, USA
2Program of Gerontology, Brigham Young
University, Provo, UT, USA
Correspondence
Jayson R. Gifford, Department of Exercise
Sciences, Brigham Young University,
Provo, UT 84602, USA.
Email: [email protected]
Funding information
Bobbit Heart Disease Award, Grant/Award
Number: 1
Abstract
Purpose: While maximum blood flow influences one's maximum rate of oxygen
consumption (V̇O2max), with so many indices of vascular function, it is still unclear
if vascular function is related to V̇O2max in healthy, young adults. The purpose of
this study was to determine if several common vascular tests of conduit artery and
resistance artery function provide similar information about vascular function and
the relationship between vascular function and V̇O2max.
Methods: Twenty-two healthy adults completed multiple assessments of leg vas-
cular function, including flow-mediated dilation (FMD), reactive hyperemia (RH),
passive leg movement (PLM), and rapid onset vasodilation (ROV). V̇O2max was
assessed with a graded exercise test on a cycle ergometer.
Results: Indices associated with resistance artery function (e.g., peak flow during
RH, PLM, and ROV) were generally related to each other (r = 0.47–77, p < .05),
while indices derived from FMD were unrelated to other tests (p < .05). Absolute
V̇O2max (r = 0.57–0.73, p < .05) and mass-specific V̇O2max (r = 0.41–0.46, p < .05)
were related to indices of resistance artery function, even when controlling for fac-
tors like body mass and sex. FMD was only related to mass-specific V̇O2max after
statistically controlling for baseline artery diameter (r = 0.44, p < .05).
Conclusion: Indices of leg resistance artery function (e.g., peak flow during RH,
PLM, and ROV) relate well to each other and account for ~30% of the variance in
V̇O2max not accounted for by other factors, like body mass and sex. Vascular inter-
ventions should focus on improving indices of resistance artery function, not conduit
artery function, when seeking to improve exercise capacity.
K E Y W O R D S
flow-mediated dilation, passive leg movement, rapid onset vasodilation, vascular function,
V̇O2max
2 of 13 |
GIFFORD et al.
1 |
INTRODUCTION
One's maximum rate of oxygen consumption (V̇O2max)
strongly influences exercise performance and is also a
strong predictor of cardiovascular risk (Poole, Behnke, &
Musch, 2020). While many systems may limit V̇O2max
(Wagner, 2008), the cardiovascular system often serves as a
significant bottleneck, with untrained individuals often ex-
hibiting a much lower cardiac output and muscle blood flow
than endurance-trained individuals (Gifford et al., 2016;
Levine, 2008). Evidence indicates that cardiac output and
muscle blood flow during exercise are both strongly in-
fluenced by the ability of the peripheral vasculature to di-
late (Bada, Svendsen, Secher, Saltin, & Mortensen, 2012;
Hanson, Proffit, & Gifford, 2020; Joyner & Casey, 2015).
While large conduit arteries, like the brachial or femoral ar-
teries, may dilate during exercise (Tremblay & Pyke, 2018),
it is the dilation of the network of small resistance arteries,
whose total cross-sectional area far exceeds that of the large
conduit arteries (Wiedman, 1963), that primarily regulates
the increase in blood flow during exercise (Joyner & Casey,
2015; VanTeeffelen & Segal, 2006).
With fluctuations in the radius of the arterial circulation
having such a profound impact on blood flow and cardiac
output, several studies have sought to determine if the ability
of the vasculature to dilate, often termed vascular function
(Gifford & Richardson, 2017), is related to exercise capacity
(Montero, 2015).
In
such
studies
vascular
function
has
usually
(Montero, 2015) been quantified with a technique that
measures the vasodilator ability of a conduit artery (Flow-
Mediated Dilation, FMD) in a region not majorly involved
in most tests of V̇O2max, the arm. Despite vascular function
being measured in a conduit artery that does not perfuse the
main exercising muscles, most studies demonstrate a sig-
nificant, positive relationship between brachial FMD and
V̇O2max during running or cycling exercise (Montero, 2015).
However, as noted by Montero (Montero, 2015), the com-
parison of conduit artery function of an upper limb to the
V̇O2max elicited by a lower-body exercise (e.g., cycling or
running) is problematic since upper-limb vascular function
is not reflective of lower-limb vascular function (Thijssen,
Rowley, et al., 2011), and exercise training is known to elicit
local adaptations in conduit artery structure that may mask
any adaptation in local vascular function (Green, Spence,
Rowley, Thijssen, & Naylor, 2012). Moreover, given the neg-
ligible role of conduit arteries in regulating exercise blood
flow (Joyner & Casey, 2015), the relevance of conduit artery
function to exercise capacity is unclear.
Multiple noninvasive assessments intended to interrogate
the function of the resistance arteries (Limberg et al., 2020)
of the lower limbs have been developed in recent years.
Indeed, tests such as reactive hyperemia (RH) in response to
cuff occlusion and removal, the hyperemic response to pas-
sive leg movement (PLM), and the rapid onset of vasodila-
tion and hyperemia in response to a single muscle contraction
(ROV) have been shown to be NO dependent (Broxterman
et al., 2017; Casey, Walker, Ranadive, Taylor, & Joyner, 2013;
Gifford & Richardson, 2017; Limberg et al., 2020) and related
to peak blood flow during knee extension exercise (Hanson
et al., 2020). Nevertheless, it is not clear if these distinct indi-
ces of vascular function actually reflect the same underlying
physiology and are related to each other. It is also unclear how
relevant these indices of resistance artery function (Limberg
et al., 2020) are to exercise capacity. If these indices of resis-
tance artery vascular function truly are representative of the
function of the resistance arteries, which are largely respon-
sible for blood flow control (Joyner & Casey, 2015), one may
expect them to relate well to exercise capacity.
To date, the relationships between the various tests of vas-
cular function in the lower limb and their relevance to exer-
cise capacity have not been extensively explored. Therefore,
the purpose of this study was twofold. First, we sought to
determine how well the various indices of conduit and re-
sistance artery function relate to each other. Second, we
sought to determine if aerobic exercise capacity, assessed by
V̇O2max during cycling exercise, is related to various indices
of conduit artery and resistance artery function.
2 |
METHODS
2.1 | Subjects
Twenty-two young, healthy subjects (13 males, 9 females,
18–30 years old) completed the study. All subjects were
healthy, nonobese, nonsmokers, free from medications
that would affect their hemodynamic responses to exercise
(Gifford & Richardson, 2017). Data for females were col-
lected within the first 7 days of the menstrual cycle to mini-
mize variability attributable to hormonal fluctuations. Prior
to starting the study, the Institutional Review Board (IRB)
at Brigham Young University (BYU) found the study to be
safe, ethical, and in agreement with the main principles out-
lined in the Declaration of Helsinki. Prior to participation, all
subjects provided informed consent. While the study was per-
formed in accordance principles outlined in the Declaration
of Helsinki, it was not registered on clinicaltrials.gov before
data collection.
2.2 | Procedures
Subjects reported to the laboratory on three occasions hav-
ing fasted for 4 hr, rested from exercise and refrained from
alcohol or caffeine consumption for ~24 hr (Gifford &
|
3 of 13
GIFFORD et al.
Richardson, 2017). Each visit was separated by a minimum
of 24 hr. All data collection was completed on the subject's
right leg, regardless of leg dominance.
On the first visit, body measurements including height (cm),
body mass (kg), and body mass index (BMI, kg·m−2). After
resting supine for 20 min, vascular function was assessed by
the FMD and RH techniques on the superficial femoral artery
as described below. Subsequently, the maximum rate of oxygen
consumption during cycling (i.e., V̇O2max) was assessed with a
graded exercise test (25 watt increments per minute) on a cycle
ergometer (Excaliber Sport, Lode, Groningen, Netherlands)
with a Parvo metabolic cart (True One, Parvo-Medics Inc.,
Sandy, Utah, USA) (Gifford et al., 2016)). The greatest power
sustained for 1 min during the graded exercise test was identified
as Graded Exercise Test Max (GXTmax). Following 30 min of
rest from the initial graded exercise test, a constant-load test
(100% GXTmax) until exhaustion test was performed to verify
the initial V̇O2max results (Poole & Jones, 2017). The verifi-
cation V̇O2max for all subjects was within ±5% of the initial
V̇O2max, supporting the attainment of V̇O2max. The higher of
the two values was recorded as the final V̇O2max.
On the second visit, vascular function was assessed first
with the passive leg movement (PLM) technique in triplicate.
Following a ~ 5-min recovery period, vascular function was
then assessed by the rapid onset vasodilation (ROV) tech-
nique elicited by single kick knee extension exercise as de-
scribed below.
2.3 | Assessments of vascular function
2.3.1 | Flow-mediated dilation (FMD) and
reactive hyperemia (RH)
During the first visit subjects reported to the laboratory to have
vascular function assessed via FMD and RH on the superficial
femoral artery according to current recommendations (Harris,
Nishiyama, Wray, & Richardson, 2010) and as previously
described (Hanson et al., 2020). While lying in the supine po-
sition, a 9 cm blood pressure cuff (Hokanson Inc., Bellevue,
WA, USA) was placed on the thigh proximal to the knee-
cap. Following a 20-min acclimation/resting period, baseline
measurements (diameter and blood flow) were gathered for
60 s at the superficial femoral artery ~10 cm proximal to the
cuff with a GE Logiq E ultrasound (General Electric Medical
Systems, Milwaukee, WI, USA) operating with a B-mode
frequency of 9 MHz and a Doppler frequency of 5 MHz. The
cuff was then inflated for 5 min to 250 mmHg. Blood velocity
and diameter data were collected for 2 min immediately after
the release of the cuff pressure. Following the study, artery
diameter was analyzed frame-by-frame by automated edge
detection software (Quipu srl., Pisa, Italy) and averaged into
1-s bins corresponding to 1-s average velocities. A 3-s rolling
average was applied to smooth diameter and velocity data.
Blood flow (ml·min−1) was calculated using the equation:
blood flow=[
(mean blood velocity)×(휋 ×(vessel radius2) ) x 60]
,
where mean blood velocity is expressed in cm·s−1 and radius
is expressed in cm. FMD measurements were expressed as a
percent change in diameter and calculated with the equation:
Shear rate was calculated with the following equation
Shear rate= 8×mean blood velocity
Diameter
. Subsequently FMD was also
normalized for total shear area under the curve (i.e., FMD/
shear) as recommended and described by Harris et al (Harris
et al., 2010). Peak flow during RH following the release of
the cuff was identified as the greatest 1-s average of flow
achieved following cuff release (Harris et al., 2010).
2.3.2 | Passive leg movement (PLM)
The hyperemic response induced by PLM, which is NO-
dependent (Broxterman et al., 2017; Mortensen, Askew,
Walker, Nyberg, & Hellsten, 2012; Trinity et al., 2012)
and strongly related to acetylcholine-induced hyperemia
(Mortensen et al., 2012), was utilized to assess thigh vascular
function according to recently published guidelines (Gifford
& Richardson, 2017). Subjects were seated in an upright posi-
tion with knees fully extended (180°) for a 20-min acclimation
period before any data were collected. Subsequently, resting
blood flow was measured for 60 s at the common femoral artery
utilizing a GE Logiq E ultrasound (General Electric Medical
Systems, Milwaukee, WI, USA) operating with a B-mode
frequency of 9 MHz, a Doppler frequency of 5 MHz, and an
insonation angle of 60°. Subsequently, researchers manually
moved the subject's leg back and forth from the extended posi-
tion of the knee (180°) to the flexed position (90°), at a rate
of 60 knee extensions per min, while the subjects stayed re-
laxed with no voluntary muscle contraction, while blood flow
was measured at the common femoral artery throughout. This
procedure was completed three times with a ~15-min period of
rest between each trial. Blood flow data were analyzed second-
by-second and a 3-s rolling average was applied to smooth the
data. The peak blood flow and the area under the curve (PLM
Total Flow) were identified for each of the three trials and then
averaged together (Gifford & Richardson, 2017). The data pre-
sented in this manuscript are the average of the three trials.
2.3.3 | Rapid onset vasodilation (ROV)
The hyperemic response to a single muscle contraction (e.g.,
one leg extension) has also been shown to be NO dependent
FMD (%)= (Peak Diameter−Baseline Diameter)
Baseline Diameter
×100.
4 of 13 |
GIFFORD et al.
(Casey et al., 2013) and is indicative of the responsiveness of
the vasculature to an exercise stimulus (Credeur et al., 2015;
Hughes, Ueda, & Casey, 2016). For this study, the hyperemic
response to a single knee extension of 60 Nm of work was
used to quantify ROV. Subjects were seated in an upright
position with legs hanging over the end of a seat with knees
in a flexed position (knee at 90° flexion at rest). The right
ankle was then connected to the cable of a knee extension
machine (a basic pulley system that vertically displaces a se-
lected amount of weight – N.K. Products, Lake Elsinore, CA,
USA). Subjects then fully extended their leg so that the verti-
cal displacement distance associated with a fully extended
kick could be measured using a standard tape measure. This
displacement distance was subsequently used to calculate
the total work performed during the different kicks. Subjects
were then familiarized with the kicking motion at various dif-
ferent weights.
Following 20-min recovery, ROV was assessed in du-
plicate in response to a full knee extension totaling 60 Nm
of work. Repeated trials were separated by at least 2 min of
recovery. As subjects of different leg lengths displaced the
weights to different distances, the mass each subject lifted
was adjusted for the absolute work kick so that the total
work (i.e., Total Work = mass × gravity × displacement
distance) was 60 Nm when extending the leg through a full
90° range of motion. For each kick, 1 min of baseline data
was collected while the leg was rested in a flexed position.
Subsequently, subjects extended the knee to ~180° and then
passively allowed the weight to flex the knee back to 90° with
no engagement of knee flexor or extensor muscles during the
knee flexion phase (e.g., active contraction during knee ex-
tension and no contraction during flexion). Femoral blood
flow was assessed, as described for the PLM technique, for
1 min of baseline prior to contraction, during the kick and for
1 min following the kick. Data were subsequently analyzed
second-by-second and a 3-s rolling average was applied to
smooth the data. The peak blood flow (ROV Peak Flow) was
subsequently identified as the greatest 1-s average of blood
flow, while the total flow response (ROV Total Flow) was
identified as the area under the curve for 60 s. As each ex-
ercise was performed in duplicate, the data reported in this
manuscript are the average of both trials.
2.4 | Statistical analysis
Test–retest reliability of the variables that were performed
in repeated measures (PLM-based indices in triplicate and
ROV-based indices in duplicate) was assessed with intra-
class correlation (ICC) using a two-way mixed model based
on absolute agreement. As the average of the multiple meas-
urements was used for the analysis in this study, the ICC
for the average of the repeated measures, not the ICC for an
individual measure, is reported. Criteria for classifying the
level of reliability of measurements were based up those set
forth by Koo & Li (2016), in which an ICC between 0.50
and 0.75 is evidence of “moderate reliability”, an ICC be-
tween 0.75 and 0.90 is evidence of “good reliability”, and an
ICC > 0.90 is evidence of “excellent reliability”.
Pearson correlation and a linear regression were utilized
to determine the relationship between the various assess-
ments of vascular function and other variables. Categorical
data, like sex, were dummy coded into correlations. Part/
partial correlation was utilized to determine the amount of
unique variance shared by two variables when removing that
related to a third variable. Principal components analysis
was utilized to combine the large amount of information pro-
vided by the multiple indices of vascular function into fewer,
discrete variables based on the shared variance among the
different indices of vascular function. Specifically, major
variables derived from the tests of vascular function (FMD
% dilation, FMD/shear, RH Peak Flow, RH Total Flow, PLM
Peak Flow, PLM Total Flow, ROV Peak Flow, and ROV Total
Flow) were entered into a principal components analysis with
orthogonal rotation (varimax). Factors with an eigenvalues
greater than 0.7 were accepted and only variables with load-
ings greater than 0.7 were included in a factor (Field, 2009).
An independent sample t-test was conducted to identify sex
differences among the indices of vascular function. Alpha
was set at p ≤ .05 a priori.
All statistical analyses were completed using SPSS ver-
sion 26 (SPSS Inc.). Data are expressed as the mean ± SE
unless otherwise stated.
3 |
RESULTS
3.1 | Test–retest reliability of PLM and
ROV measurements
The repeated measurements of PLM Peak Flow and PLM
Total Flow both exhibited “excellent reliability” with ICC
equal to 0.91. The repeated measurements of ROV Peak
Flow exhibited “excellent reliability” with ICC equal to
0.96. The repeated measurements of ROV Total Flow exhib-
ited “moderate reliability” with ICC equal to 0.72. As men-
tioned in the methods section, the average of the repeated
measurements was utilized for all subsequent analyses in
this study.
3.2 | Relationship between the various
assessments of vascular function
As illustrated in Figure 1 and further described in Table 1,
the relationships between the multiple indices of vascular
|
5 of 13
GIFFORD et al.
function were examined with Pearson correlation. In general,
indices derived from resistance artery function tests (i.e., RH,
PLM, and ROV) were related to each other (p < .05), but
not to indices derived from conduit artery function tests (e.g.,
FMD, p > .05).
Principal components analysis of the variables listed in
Table 1 was utilized to group indices that share substantial
variance to condense the multiple indices of vascular func-
tion to fewer factors. In essence, this analysis determines
the extent to which the various assessments of vascular
function represent similar or distinct factors. The Kaiser-
Meyer-Olkin measure (KMO = 0.54) supported the sam-
pling adequacy for the factor analysis. Visual analysis of
a scree plot indicated a breakpoint at two factors, support-
ing the inclusion of two different factors with eigenvalues
greater than 0.7. Factor #1 was exclusively comprised of
factors related to FMD (FMD % dilation and FMD/shear)
with loading factors of 0.72 and 0.89, respectively. Factor
#2 was comprised of the following variables with the load-
ing factors indicated in parentheses: RH Peak Flow (0.84),
RH Total Flow (0.75), PLM Peak Flow (0.83), PLM Total
Flow (0.74), and ROV Peak Flow (0.84). ROV total flow
was not included in either factor.
3.3 | Relationship between indices of
vascular function and V̇O2max
As illustrated in Figure 2 and further described in Table 2,
variables associated with FMD were unrelated to mass-
specific and absolute V̇O2max (p = .12–0.40). Meanwhile,
variables associated with the second factor revealed in fac-
tor analysis (e.g., RH Peak Flow, PLM Peak Flow, and
ROV Peak Flow) exhibited moderate-to-strong correla-
tions with absolute and mass-specific V̇O2max (r = 0.56–
73, p < .05).
FIGURE 1
Relationship between Different Indices of Vascular Function. (a) Relationship between the peak flow achieved during passive
leg movement (PLM) and the peak flow achieved during the rapid onset vasodilation (ROV) test. (b) Relationship between PLM peak flow
and the peak flow observed during a reactive hyperemia (RH) test. (c) Relationship between the peak flow achieved during the ROV and RH
tests. (d) Relationship between flow-mediated dilation (FMD) of the superficial femoral artery and the peak flow achieved during an ROV test.
(e) Relationship between FMD of the superficial femoral artery and the peak flow achieved during RH. (f) Relationship between FMD of the
superficial femoral artery and the peak flow achieved during PLM. A solid trendline indicates a significant relationship between the two variables
(p ≤ .05) while a dotted trendline indicates a nonsignificant relationship between the two variables (p > .05). Light gray circles represent data for
females and dark gray circles represent data for males
6 of 13 |
GIFFORD et al.
TABLE 1
Relationship between various indices of vascular function
FMD (%
Dilation)
FMD (%/
Shear)
RH peak flow
(ml/min)
RH total flow
(ml)
PLM peak flow
(ml/min)
PLM total
flow (ml)
ROV peak flow
(ml/min)
ROV total flow
(ml)
Factor 1
FMD (%
Dilation)
-
r = 0.47
p = .04
r = −0.14
p = .55
r = −0.01
p = .99
r = −0.11
p = .64
r = 0.01
p = .96
r = −0.33
p = .15
r = −0.04
p = .85
FMD (%/ Shear)
r = 0.47
p = .04
-
r = −0.21
p = .37
r = −0.54
p = .01
r = −0.31
p = .18
r = −0.07
p = .78
r = −0.32
p = .17
r = 0.18
p = .45
Factor 2
RH peak flow
(ml/min)
r = −0.14
p = .55
r = −0.21
p = .37
-
r = 0.82
p < .01
r = 0.47
p = .03
r = 0.31
p = .17
r = 0.77
p < .01
r = 0.36
p = .11
RH total flow
(ml)
r = −0.01
p = .99
r = −0.54
p = .01
r = 0.82
p < .01
-
r = 0.40
p = .08
r = 0.23
p = .32
r = 0.57
p = .01
r = 0.09
p = .71
PLM peak flow
(ml/min)
r = −0.11
p = .64
r = −0.31
p = .18
r = 0.47
p = .03
r = 0.40
p = .08
-
r = 0.89
p < .01
r = 0.64
p < .01
r = 0.24
p = .28
PLM total flow
(ml)
r = 0.01
p = .96
r = −0.07
p = .78
r = 0.31
p = .17
r = 0.23
p = .32
r = 0.89
p = .01
-
r = 0.45
p = .04
r = 0.18
p = .41
ROV peak flow
(ml/min)
r = −0.33
p = .15
r = −0.32
p = .17
r = 0.77
p < .01
r = 0.57
p = .01
r = 0.64
p < .01
r = 0.45
p = .04
-
r = 0.63
p < .01
ROV total flow
(ml)
r = −0.04
p = .85
r = 0.18
p = .45
r = 0.36
p = .11
r = 0.09
p = .71 .01
r = 0.24
p = .28
r = 0.18
p = .41
r = 0.63
p < .01
Note: The terms “Factor 1” and “Factor 2” at the left of the table refer to the variables that were grouped together via principal components analysis. Significant relationships are in bold font
Abbreviations: FMD, flow-mediated dilation; PLM: passive leg movement; RH: Reactive Hyperemia; ROV: Rapid onset vasodilation.
|
7 of 13
GIFFORD et al.
3.4 | Other factors that relate to the
indices of vascular function
As described in Table 3, factors related to a subject's anat-
omy, sex, and body mass were related to the outcomes of the
vascular function tests. Notably, FMD exhibited a negative
correlation with the artery diameter at baseline (r = −0.64,
p = .002), such that individuals with larger arteries tended
to exhibit lower FMD (Table 3). Meanwhile, body mass was
positively related with RH Peak Flow (r = 0.47, p = .01),
PLM Peak Flow (r = 0.62, p < .01), and ROV Peak Flow
(r = 0.53, p = .01).
3.4.1 | Sex differences in indices of
vascular function
PLM Peak Flow (Female: 1,140 ± 98 ml min−1, Male:
1626 ± 114 ml min−1; p = .006) and PLM Total Flow
(Female: 333 ± 57 ml, Male: 566 ± 56 ml; p = .01) were
both significantly greater in males than females. ROV peak
flow also tended to be greater in males than females (Female:
1762 ± 149 ml min−1, Male: 2,196 ± 186 ml min−1; p = .10),
while FMD (% Dilation) tended to be lower in males than fe-
males (Female: 6.98 ± 0.78%, Male: 4.73 ± 0.85%; p = .09).
The sex difference in PLM Peak Flow and Total Flow disap-
peared when controlling for body mass (p = .98), which was
significantly different between the females and males in the
study (57.00 ± 1.69 kg vs. 82.35 ± 1.69 kg, respectively,
p < .01).
3.5 | Relationship between vascular
function and V̇O2max when controlling for
other variables
Recognizing that several other factors may potentially influ-
ence the responses observed in the different vascular func-
tion tests (see Table 3), the relationship between V̇O2max
and the various indices of vascular function was examined
when controlling for potentially confounding variables.
When controlling for the variation in FMD accounted for by
baseline diameter, FMD was found to be significantly related
to the mass-specific V̇O2max (r = 0.44, p = .04; Table 4).
Moreover, when simultaneously accounting for the variance
related to body mass, sex, and BMI with partial correlation,
RH Peak Flow, PLM Peak, and ROV Peak Flow were still
significantly related to absolute V̇O2max (r = 0.49–0.59,
p < .05) and mass-specific V̇O2max (r = 0.46–0.55, p ≤ .05).
Finally, stepwise linear regression was performed to
explore the possibility of predicting V̇O2max with vascu-
lar function data and other subject characteristics. Of the
five variables entered into the regression (body mass, sex,
height, BMI, and PLM Peak Flow), only body mass, PLM
FIGURE 2
The Relationship between
Vascular Function and The Maximum Rate
of Oxygen Consumption (V̇O2max) during
Cycling. The relationship between absolute
V̇O2max and (a) flow-mediated dilation
(FMD) of the superficial femoral artery (b)
peak flow during Reactive hyperemia (RH),
(c) peak flow during passive leg movement
(PLM) and (d) peak flow during a rapid
onset vasodilation (ROV) test. V̇O2max.
A solid trendline indicates a significant
relationship between the two variables
(p ≤ .05), while a dotted trendline indicates
a nonsignificant relationship between the
two variables (p > .05). Light gray circles
represent data for females and dark gray
circles represent data for males
8 of 13 |
GIFFORD et al.
Peak Flow, and BMI were retained by the stepwise regres-
sion, yielding the following equation (R2 = 0.83, p < .01):
Absolute V̇O2max = 970.82 + 55.83 (Body Mass) + 0.68
(PLM Peak Flow) – 121.75 (BMI).
4 |
DISCUSSION
The purpose of this study was to determine how well the vari-
ous indices of vascular function relate to each other and if
aerobic capacity, assessed by V̇O2max, is related to these in-
dices of vascular function. The results of this inquiry yielded
two major findings. First, in agreement with current thought
(Limberg et al., 2020; Thijssen, Black, et al., 2011), the as-
sessments of conduit artery function (FMD and its deriva-
tives) and resistance artery function (derivatives of RH, PLM,
and ROV) appear to reflect two different aspects of vascular
function, with the indices derived from the RH, PLM, and
ROV being strongly correlated with each other, but not with
FMD and its derivatives. The second major finding of this
study is that leg vascular function, especially resistance ar-
tery function, is strongly related to V̇O2max, accounting for
approximately 30% of the variance in V̇O2max not accounted
for by known influencers, like body mass, sex, and BMI.
4.1 | Are the various indices of vascular
function interchangable with one another?
Multiple methods exist for quantifying a person's vascular
function, yet it is unclear if these various methods are related
to each other. Therefore, in the current study vascular func-
tion was measured in multiple ways (FMD, RH, PLM, and
ROV) on a group of young, healthy adults. As illustrated in
Figure 1 and further described in Table 1, lower limb vas-
cular function assessed by the resistance artery tests RH,
PLM, and ROV exhibits strong relationships with each other
(r = 0.54–0.83, p < .05), supporting the notion that they
reflect some of the same physiological processes (Limberg
et al., 2020). This comes in agreement with data from
Rossman, Groot, Garten, Witman, & Richardson (2016) and
Walker et al. (2016) who observed significant correlations
between PLM-induced hyperemia and RH in various popu-
lations. However, as was the case for Rossman et al. (2016),
vascular function assessed by FMD of the superficial femo-
ral artery was not related to the other measurements of vas-
cular function (e.g., PLM-induced hyperemia) examined in
the current study (Figure 1, Table 1).
The lack of relationship between FMD and the other vari-
ables should not be interpreted as evidence of superiority
or inferiority of one test over another, but as an indication
that these validated tests of vascular function capture differ-
ent aspects of cardiovascular physiology. Indeed, principal
components analysis, which consolidated the various indices
of vascular function into two different factors, supports the
idea that the results of the various tests capture two general
aspects of vascular physiology. As illustrated in Table 1,
Factor 1 is comprised exclusively of FMD and factors de-
rived from the FMD test, which have been suggested to rep-
resent conduit artery function (Thijssen, Black, et al., 2011).
Meanwhile, Factor 2 was comprised of the main indices de-
rived from RH, PLM, and ROV, all of which have recently
been referred to as tests of resistance artery or resistance
vessel function (Limberg et al., 2020). Thus, the current data
indicate that the tests of resistance artery function used in
the current study are relatively interchangeable, but that tests
reflecting conduit artery function should not be considered
as surrogates for tests of resistance artery function, or vice
versa.
TABLE 2
Relationship between different indices of vascular
function and maximum rate of oxygen consumption (V̇O2max)
achieved during cycling
Mass-Specific V̇O2max
(ml/kg/min)
Absolute V̇O2max
(ml/min)
FMD (%
Dilation)
r = 0.20
p = .40
r = −0.24
p = .31
FMD (%/shear)
r = −0.27
p = .24
r = −0.35
p = .12
RH Peak Flow
(ml/min)
r = 0.26
p = .25
r = 0.49
p = .02
RH Total Flow
(ml)
r = 0.41
p = .06
r = 0.44
p = .04
PLM Peak Flow
(ml/min)
r = 0.48
p = .03
r = 0.72
p < .01
PLM Total
Flow (ml)
r = 0.42
p = .05
r = 0.65
p < .01
ROV Peak Flow
(ml/min)
r = 0.36
p = .11
r = 0.58
p < .01
ROV Total
Flow (ml)
r = 0.10
p = .65
r = 0.03
p = .91
Body Mass (kg)
r = 0.16
p = .49
r = 0.84
p < .01
Body Mass
Index (kg/m2)
r = −0.16
p = .49
r = 0.64
p < .01
Sex
(Female = −1,
Male = +1)
r = 0.01
p = .98
r = 0.79
p < .01
Note: Note that sex has been dummy coded with females being coded as −1
and males being entered as + 1. In this dummy coding scenario, a negative
correlation indicates greater values are associated with the female sex, while a
positive correlation indicates greater values are associated with the male sex.
Significant relationships are in bold font
Abbreviations: FMD, flow-mediated dilation; PLM, passive leg movement; RH,
reactive hyperemia; ROV, rapid onset vasodilation.
|
9 of 13
GIFFORD et al.
4.2 | Are conduit and/or resistance artery
function related to V̇O2max?
The overarching aim of this study was to answer the ques-
tion, “Is vascular function related to V̇O2max?” However,
the data in Table 1 make it clear that one must clarify
which aspect of vascular function is of interest when
answering this question, since indices of conduit artery
function and resistance artery function are not well cor-
related. As illustrated in Figure 2, resistance artery, but
not conduit artery, function was strongly related to ab-
solute V̇O2max, meaning that an individual with a large
hyperemic response to the vascular tests would be likely
to achieve a greater maximal rate of oxygen consumption
and power output (e.g., GXTmax) during a graded exer-
cise test. Meanwhile, mass-specific V̇O2max was only re-
lated to resistance artery function assessed by PLM Peak
Flow (r = 0.46, p = .03) and RH Total Flow (r = 0.41,
p = .05), but not conduit artery function assessed by
FMD (r = 0.20, p = .40). The strong relationship between
Baseline Artery
Diameter (mm)
Body Mass
(kg)
BMI
(kg/m2)
Sex
(Female = −1,
Male = +1)
FMD (%
Dilation)
r = −0.64
p < .01
r = −0.47
p = .03
r = −0.48
p = .03
r = −0.38
p = .09
FMD (%/shear)
r = −0.49
p = .03
r = −0.27
p = .26
r = −0.33
p = .17
r = −0.15
p = .54
RH Peak Flow
(ml/min)
r = 0.65
p < .01
r = −0.47
p = .03
r = 0.48
p = .03
r = 0.30
p = .19
RH Total Flow
(ml)
r = 0.48
p = .03
r = 0.28
p = .22
r = 0.31
p = .17
r = 0.17
p = .46
PLM Peak
Flow (ml/min)
r = 0.85
p < .01
r = 0.62
p < .01
r = 0.59
p < .01
r = 0.56
p < .01
PLM Total
Flow (ml)
r = 0.74
p < .01
r = 0.58
p < .01
r = 0.54
p < .01
r = 0.54
p = .01
ROV Peak
Flow (ml/min)
r = 0.75
p < .01
r = 0.53
p = .01
r = 0.61
p < .01
r = 0.35
p = .11
ROV Total
Flow (ml)
r = 0.18
p = .43
r = 0.01
p = .99
r = 0.15
p = .50
r = 0.03
p = .98
Note: Note that sex has been dummy coded with females being coded as −1 and males being entered as +1.
Significant relationships are in bold font.
Abbreviations: BMI, body mass index; FMD, flow-mediated dilation; PLM, passive leg movement; ROV,
rapid onset vasodilation.
TABLE 3
Relationship between
indices of vascular function and other
subject characteristics
Mass-Specific V̇O2max
(ml/kg/min)
Absolute V̇O2max
(ml/min)
FMD (% Dilation)
Controlling for baseline diameter
r = 0.45
p = .04
r = 0.35
p = .14
Peak Reactive Hyperemia (ml/
min)
Controlling for body mass, sex,
and BMI
r = 0.46
p = .05
r = 0.49
p = .04
PLM Peak Flow (ml/min)
Controlling for body mass, sex,
and BMI
r = 0.53
p = .02
r = 0.58
p = .01
ROV Peak Flow (ml/min)
Controlling for body mass, sex,
and BMI
r = 0.55
p = .01
r = 0.59
p < .01
Note: Note that sex has been dummy coded with females being coded as −1 and males being entered as +1.
Significant relationships are in bold font.
Abbreviations: BMI, body mass index; FMD, flow-mediated dilation; PLM, passive leg movement; ROV,
rapid onset vasodilation.
TABLE 4
Partial correlations
between indices of vascular function and
the maximum rate of oxygen consumption
(V̇O2max) during cycling exercise when
controlling for potentially confounding
variables
10 of 13 |
GIFFORD et al.
resistance artery function and V̇O2max in these healthy
young adults is consistent with previous studies that have
reported relationships between V̇O2max and the hypere-
mic responses to RH (Robbins et al., 2011) and ROV (19)
in various populations.
It makes sense that V̇O2max would be more related
to resistance artery function than conduit artery func-
tion since V̇O2max is strongly influenced by maximum
blood flow (Gifford et al., 2016; Levine, 2008) which is
primarily controlled by the dilation and constriction of
the myriad of resistance arteries (Dodd & Johnson, 1991;
Joyner & Case y, 2015). Along these lines, our group re-
cently reported that factors associated with resistance ar-
tery function (e.g., PLM Peak Flow and ROV Peak Flow)
were very predictive of peak blood flow achieved during
knee extension exercise, while FMD was not (Hanson
et al., 2020). A large PLM, RH, or ROV response seems
to be indicative of a limb with a network of resistance
arteries that can accommodate high rates of blood flow,
thereby facilitating a greater V̇O2max. Thus, interven-
tions targeting resistance artery function may potentially
have more impact on exercise tolerance in healthy adults
than interventions seeking to improve conduit artery
function. Future studies could potentially further exam-
ine the relationship between conduit artery function and
V̇O2max by measuring conduit artery diameter during
a V̇O2max test. Unfortunately, such precise diameter
measurements are not currently possible during cycling
exercise.
Contrary
to
our
findings,
previous
research
(Montero, 2015) has indicated that FMD is typically re-
lated to V̇O2max, most commonly the mass-specific
V̇O2max. The reason for the disagreement between find-
ings may be due to measurement location. In contrast to
most previous studies, which measured FMD in the arm,
the current study compared vascular function, including
FMD, assessed in the lower limb to cardiorespiratory fit-
ness assessed during a predominantly lower-limb exercise
like cycling or running. Indeed, the aforementioned me-
ta-analysis (Montero, 2015) concluded “further studies
are needed to elucidate the association of cardiorespi-
ratory fitness with lower limb endothelial function.” As
mentioned earlier, exercise-induced adaptations to arterial
structure and diameter appear to be of a greater magnitude
in exercise-trained muscles than in nontrained muscles
(Rowley et al., 2012). It is possible that exercise-induced
adaptations in the diameter of the superficial femoral ar-
tery masked the relationship between FMD in the lower
limb and V̇O2max. Thus, further investigation into the
relationship between vascular function and V̇O2max,
when controlling for potentially confounding variables, is
warranted.
4.3 | What other factors influence the
indices of vascular function?
It is important to recognize that although these indices of vas-
cular function are related to NO bioavailability and endothe-
lial function (Casey & Joyner, 2011; Green, 2005; Mortensen
et al., 2012), multiple other factors, besides endothelial func-
tion, can influence the results of these vascular function
tests. As listed in Table 3, the measures of vascular func-
tion utilized in the current study are sensitive to several fac-
tors that should be considered when interpreting the results
of a test. For example, in agreement with previous research
(Anderson et al., 1995; Celermajer et al., 1992), FMD was
negatively related to baseline artery diameter, such that indi-
viduals with a large diameter artery at baseline tend to exhibit
a lower FMD. In the initial paper to link brachial artery FMD
to coronary endothelial dysfunction (Anderson et al., 1995),
the authors indicated that baseline brachial artery diameter
was the strongest predictor of a decreased FMD, not the coro-
nary endothelial dysfunction for which the paper is famous.
With ~41% of the variation in FMD in the current sample
being related to baseline diameter (i.e., R2 = 0.41, p < .01),
it is possible that the arterial enlargement associated with ha-
bitual exercise (Green et al., 2012) may have masked any
potential relationship between FMD and V̇O2max in the cur-
rent study.
As depicted in Table 3 tests of resistance artery func-
tion are strongly related with body mass and BMI, such
that larger individuals with larger thighs tend to exhibit a
greater RH Peak Flow, PLM Peak Flow, and ROV Peak
Flow. It is not possible to conclude why this relationship
exists from the current data, but it seems likely that larger
limbs have a larger vascular network, which can accommo-
date greater flows. Whatever the mechanism, the influence
of body mass on the measures of resistance artery function
is not trivial and should be considered when interpreting
these tests, especially when relating vascular function to
V̇O2max, which is also strongly influenced by body mass
(Proctor & Joyner, 1997).
Sex is also related to resistance artery function (Table 3),
with males exhibiting a greater peak flow response to PLM.
A similar tendency was also observed with ROV Peak Flow
(p = .10). However, this sex difference in resistance artery
function appears to be driven by differences in body mass
between females and males (males were 25.35 ± 2.73 kg
heavier than the females in this study, p < .01), since the sex
differences in PLM Peak Flow disappeared when statistically
removing variance in PLM Peak Flow accounted for by body
mass (p = .98).
In addition to the factors mentioned above, previous re-
search has revealed other factors that must be considered when
performing and interpreting tests of vascular function. For
|
11 of 13
GIFFORD et al.
example, the placement of the cuff proximal or distal to the site
of measurement may impact the results of an FMD test (Doshi
et al., 2001), the frequency of movement and the range of mo-
tion of PLM (Gifford et al., 2019), and the amount of work
performed during ROV (Tschakovsky et al., 2004) have been
shown to strongly impact the results. Therefore, these factors
should be considered when exploring the relationship between
vascular function and other variables, like V̇O2max.
4.4 | Is vascular function related to V̇O2max
when controlling for potentially confounding
variables?
As described above, several factors, independent of the
health of the vascular system, may impact the results of a
vascular function test. Thus, it is possible that the underlying
influences of variables, like artery diameter and body size,
either mask potential relationships between vascular func-
tion and V̇O2max or potentially account for them. Partial
correlations between the indices of vascular function and
V̇O2max were performed to statistically remove variance ac-
counted for potentially confounding variables. As described
in Table 4, when statistically controlling for the variance in
FMD related to baseline artery diameter, superficial femoral
artery FMD does exhibit the weak relationship with mass-
specific V̇O2max (r = 0.45, p = .05) that has been indicated
by studies measuring FMD in the arm (Montero, 2015). No
such relationships were observed with absolute V̇O2max
(p > .05). Thus, conduit artery function does appear to be
weakly related to mass-specific V̇O2max, but the relationship
is obscured by variation in artery diameter.
While indices of resistance artery function are related
to V̇O2max (Table 2), this relationship could potentially be
completely dependent upon body mass, sex, and BMI, which
are also strongly related to vascular function (Table 3) and
V̇O2max (Table 2). Thus, the partial correlation between the
indices of resistance artery function and V̇O2max was ex-
plored when simultaneously controlling for body mass, sex,
and BMI. As described in Table 4, the relationship between
resistance artery function and absolute V̇O2max persists,
while the relationship between resistance artery function
and mass-specific V̇O2max is apparently strengthened when
removing any variance in vascular function and V̇O2max
related to body mass, sex, and BMI. Similarly, previous re-
search indicated that PLM Peak Flow was related to peak
exercise blood flow in a mass-independent manner (Hanson
et al., 2020). Thus, the relationship between resistance artery
function and V̇O2max occurs independently and is not merely
a product of sex, mass, or BMI.
As described by Wagner (Wagner, 2008), V̇O2max can be
simultaneously influenced by the function of many systems,
including the lungs, heart, arteries, skeletal muscle mass, and
mitochondria. With so many factors influencing V̇O2max in
healthy young adults that resistance artery function accounts
for ~30% of the variance in V̇O2max not accounted for by
body mass, sex, and BMI is quite notable. Factors that were
not measured in the current study, like maximal cardiac out-
put, mitochondrial density, and muscle oxygen diffusion are
likely to account for some of the remaining variance (Gifford
et al., 2016; Wagner, 2008). Since V̇O2max is limited by dif-
ferent factors in different populations (Gifford et al., 2016;
Wagner, 2008), the amount of variance in V̇O2max ac-
counted for by resistance artery function likely differ in other
populations.
4.5 | Clinical relevance
Since V̇O2max strongly influences exercise performance and is
also a strong predictor of cardiovascular risk (Poole et al., 2020),
there is great interest in identifying what limits or reduces an
individual's V̇O2max (Wagner, 2008) so that appropriate steps
may be taken to improve it. With resistance artery function
being related to both maximum exercise blood flow (Hanson
et al., 2020) and V̇O2max (Table 2), noninvasive assessments,
like passive-leg movement (PLM)-induced hyperemia, may
conceivably be used to easily determine the likelihood that im-
pairments in muscle resistance artery function and leg blood
flow impair a person's V̇O2max. Since the PLM technique oc-
curs while the subject is in a completely rested state, this could
be particularly useful in scenarios in which direct assessment of
exercise blood flow may not be possible or practical.
Given the strong relationship between resistance artery
function and V̇O2max, vascular function data collected at rest
could potentially be used to predict V̇O2max. For example,
stepwise linear regression revealed that absolute V̇O2max
(expressed in ml·min−1) could be predicted (R2 = 0.83,
p < .01, n = 22) when considering the peak flow response to
PLM (expressed in ml·min−1), body mass (expressed in kg),
and BMI (expressed in kg·m−2):
Clearly, these data are very preliminary, and a much
larger, more heterogeneous sample is needed before a pre-
diction equation may be validated and standardized, but the
prospect of accurately predicting V̇O2max without breaking
a sweat is enticing.
4.6 | Conclusions
This study supports the notion that noninvasive indices of
vascular function generally reflect two different aspects of
Absolute ̇VO2max =970.82+55.83 (Body Mass)
+0.68 (PLM Peak Flow)−121.75 (BMI) .
12 of 13 |
GIFFORD et al.
vascular function: conduit artery function (e.g., FMD) and
resistance artery function (e.g., RH Peak Flow, PLM Peak
Flow, and ROV Peak Flow). Importantly, the results of the
tests within each aspect of vascular function (i.e., conduit or
resistance artery function) relate well to one another, such
that inferences about one test may be made based on the
results of another. While only a weak relationship between
conduit artery function (e.g. FMD) and V̇O2max is observed
when accounting for baseline artery diameter, resistance
artery function, assessed by multiple different tests, is con-
sistently and independently related to V̇O2max. While FMD
has been related to various aspects of cardiovascular health
(Broxterman et al., 2019), it is the function of the resist-
ance arteries, not the conduit arteries, that is tightly related
to exercise capacity and physical function. Thus, vascu-
lar interventions, like exercise training (Montero, Walther,
Diaz-Cañestro, Pyke, & Padilla, 2015), seeking to improve
exercise capacity should target resistance artery function, as
represented by factors like peak flow during PLM, RH, or
ROV.
ACKNOWLEDGMENTS
The authors acknowledge and thank the participants for their
gracious participation, and the peer reviewers for their ef-
forts in refining this manuscript. This study was funded by
the Bobbitt Heart Disease Research Award and the BYU
Graduate Student Mentorship Award. The authors have no
conflicts of interest to report.
AUTHOR CONTRIBUTIONS
JG: Designed and performed the study, analyzed the data,
and wrote the manuscript. BH: Designed and performed the
study and wrote the manuscript. MP: Performed the study,
analyzed the data, and approved the final manuscript. TW:
Performed the study, analyzed the data, and approved the
final manuscript. GG: Performed the study, analyzed the
data, and approved the final manuscript. JK: Performed the
study, analyzed the data, and approved the final manuscript.
MH: Performed the study, analyzed the data, and approved
the final manuscript.
ORCID
Jayson R. Gifford
https://orcid.org/0000-0002-6034-306X
REFERENCES
Anderson, T. J., Uehata, A., Gerhard, M. D., Meredith, I. T., Knab, S.,
Delagrange, D., … Selwyn, A. P. (1995). Close relation of endo-
thelial function in the human coronary and peripheral circulations.
Journal of the American College of Cardiology, 26, 1235–1241.
https://doi.org/10.1016/0735-1097(95)00327 -4
Bada, A. A., Svendsen, J. H., Secher, N. H., Saltin, B., & Mortensen, S.
P. (2012). Peripheral vasodilatation determines cardiac output in ex-
ercising humans: Insight from atrial pacing. Journal of Physiology,
590, 2051–2060.
Broxterman, R. M., Trinity, J. D., Gifford, J. R., Kwon, O. S., Kithas,
A. C., Hydren, J. R., … Richardson, R. S. (2017). Single passive
leg movement assessment of vascular function: The contribution of
nitric oxide. Journal of Applied Physiology, 123, 1468–1476.
Broxterman, R. M., Witman, M. A., Trinity, J. D., Groot, H. J., Rossman,
M. J., Park, S.- H.- S.-Y., … Richardson, R. S. (2019). Strong rela-
tionship between vascular function in the coronary and brachial ar-
teries. Hypertension, 74, 208–215. https://doi.org/10.1161/HYPER
TENSI ONAHA.119.12881
Casey, D. P., & Joyner, M. J. (2011). Local control of skeletal muscle
blood flow during exercise: Influence of available oxygen. Journal
of Applied Physiology, 111, 1527–1538.
Casey, D. P., Walker, B. G., Ranadive, S. M., Taylor, J. L., & Joyner, M.
J. (2013). Contribution of nitric oxide in the contraction-induced rapid
vasodilation in young and older adults. Journal of Applied Physiology,
115, 446–455. https://doi.org/10.1152/jappl physi ol.00446.2013
Celermajer, D. S., Sorensen, K. E., Gooch, V. M., Spiegelhalter, D.
J., Miller, O. I., Sullivan, I. D., … Deanfield, J. E. (1992). Non-
invasive detection of endothelial dysfunction in children and adults
at risk of atherosclerosis. Lancet, 340, 1111–1115. https://doi.
org/10.1016/0140-6736(92)93147 -F
Credeur, D. P., Holwerda, S. W., Restaino, R. M., King, P. M., Crutcher,
K. L., Laughlin, M. H., … Fadel, P. J. (2015). Characterizing rap-
id-onset vasodilation to single muscle contractions in the human
leg. Journal of Applied Physiology, 118(4), 455–464. https://doi.
org/10.1152/jappl physi ol.00785.2014
Dodd, L. R., & Johnson, P. C. (1991). Diameter changes in arterio-
lar networks of contracting skeletal muscle. American Journal of
Physiology-Heart and Circulatory Physiology, 260(3), H662–H670.
https://doi.org/10.1152/ajphe art.1991.260.3.H662
Doshi, S. N., Naka, K. K., Payne, N., Jones, C. J., Ashton, M., Lewis,
M. J., & Goodfellow, J. (2001). Flow-mediated dilatation following
wrist and upper arm occlusion in humans: The contribution of nitric
oxide. Clinical Science (Lond), 101, 629–635.
Field, A. (2009). Discovering statistics using SPSS, 3rd ed. Los Angeles:
SAGE Pulbications LTD.
Gifford, J. R., Bloomfield, T., Davis, T., Addington, A., McMullin,
E., Wallace, T., … Hanson, B. (2019). The effect of the speed and
range of motion of movement on the hyperemic response to pas-
sive leg movement. Physiological Reports, 7, e14064. https://doi.
org/10.14814/ phy2.14064
Gifford, J. R., Garten, R. S., Nelson, A. D., Trinity, J. D., Layec, G.,
Witman, M. A. H. H., … Richardson, R. S. (2016). Symmorphosis
and skeletal muscle V̇O2 max: In vivo and in vitro measures reveal
differing constraints in the exercise-trained and untrained human.
Journal of Physiology, 594, 1741–1751.
Gifford, J. R., & Richardson, R. S. (2017). CORP: Ultrasound assess-
ment of vascular function with the passive leg movement technique.
Journal of Applied Physiology, 123, 1708–1720.
Green, D. (2005). Point: Counterpoint Point : Flow-mediated dilation
does reflect nitric oxide-mediated endothelial function. Journal of
Applied Physiology, 99, 1233–1238.
Green, D. J., Spence, A., Rowley, N., Thijssen, D. H. J., & Naylor, L.
H. (2012). Vascular adaptation in athletes: Is there an “athlete’s ar-
tery”? Experimental Physiology, 295–304. https://doi.org/10.1113/
expph ysiol.2011.058826
Hanson, B. E., Proffit, M., & Gifford, J. R. (2020). Vascular function
is related to blood flow during high-intensity, but not low-intensity,
knee extension exercise. Journal of Applied Physiology, 128(3),
698–708. https://doi.org/10.1152/jappl physi ol.00671.2019
|
13 of 13
GIFFORD et al.
Harris, R. A., Nishiyama, S. K., Wray, D. W., & Richardson, R.
S. (2010). Ultrasound assessment of flow-mediated dilation.
Hypertension, 55, 1075–1085. https://doi.org/10.1161/HYPER
TENSI ONAHA.110.150821
Hughes, W. E., Ueda, K., & Casey, D. P. (2016). Chronic endurance ex-
ercise training offsets the age-related attenuation in contraction-in-
duced rapid vasodilation. Journal of Applied Physiology, 120,
1335–1342. https://doi.org/10.1152/jappl physi ol.00057.2016
Joyner, M. J., & Casey, D. P. (2015). Regulation of increased blood flow
(hyperemia) to muscles during exercise: A hierarchy of competing
physiological needs. Physiological Reviews, 95, 549–601.
Koo, T. K., & Li, M. Y. (2016). A guideline of selecting and reporting
intraclass correlation coefficients for reliability research. Journal
of Chiropractic Medicine, 15, 155–163. https://doi.org/10.1016/j.
jcm.2016.02.012
Levine, B. D. (2008). VO2max: What do we know, and what do we
still need to know? Journal of Physiology, 586, 25–34. https://doi.
org/10.1113/jphys iol.2007.147629
Limberg, J. K., Casey, D. P., Trinity, J. D., Nicholson, W. T., Wray, D. W.,
Tschakovsky, M. E., … Padilla, J. (2020). Assessment of resistance
vessel function in human skeletal muscle: Guidelines for experimen-
tal design, Doppler ultrasound, and pharmacology. American Journal
of Physiology. Heart and Circulatory Physiology, 318, H301–H325.
Montero, D. (2015). The association of cardiorespiratory fitness with
endothelial or smooth muscle vasodilator function. European
Journal of Preventive Cardiology, 22, 1200–1211. https://doi.
org/10.1177/20474 87314 553780
Montero, D., Walther, G., Diaz-Cañestro, C., Pyke, K. E., & Padilla,
J. (2015). Microvascular dilator function in athletes: A systematic
review and meta-analysis. Medicine & Science in Sports & Exercise,
47, 1485–1494.
Mortensen, S. P., Askew, C. D., Walker, M., Nyberg, M., & Hellsten, Y.
(2012). The hyperaemic response to passive leg movement is depen-
dent on nitric oxide: A new tool to evaluate endothelial nitric oxide
function. Journal of Physiology, 590, 4391–4400.
Poole, D. C., Behnke, B. J., & Musch, T. I. (2020). The role of vascu-
lar function on exercise capacity in health and disease. Journal of
Physiology, https://doi.org/10.1113/JP278931.
Poole, D. C., & Jones, A. M. (2017). Measurement of the maximum
oxygen uptake V̇O2max : V̇O2peak is no longer acceptable. Journal of
Applied Physiology, 122, 997–1002.
Proctor, D. N., & Joyner, M. J. (1997). Skeletal muscle mass and the
reduction of V̇O2(max) in trained older subjects. Journal of Applied
Physiology, 82, 1411–1415.
Robbins, J. L., Jones, W. S., Duscha, B. D., Allen, J. D., Kraus, W.
E., Regensteiner, J. G., … Annex, B. H. (2011). Relationship be-
tween leg muscle capillary density and peak hyperemic blood flow
with endurance capacity in peripheral artery disease. Journal of
Applied Physiology, 111, 81–86. https://doi.org/10.1152/jappl physi
ol.00141.2011
Rossman, M. J., Groot, H. J., Garten, R. S., Witman, M. A. H., &
Richardson, R. S. (2016). Vascular function assessed by passive leg
movement and flow-mediated dilation: Initial evidence of construct
validity. American Journal of Physiology-Heart and Circulatory
Physiology, 311, ajpheart.00421.2016.
Rowley, N. J., Dawson, E. A., Hopman, M. T. E., George, K. P., Whyte,
G. P., Thijssen, D. H. J., & Green, D. J. (2012). Conduit diame-
ter and wall remodeling in elite athletes and spinal cord injury.
Medicine and Science in Sports and Exercise, 44, 844–849. https://
doi.org/10.1249/MSS.0b013 e3182 3f6887
Thijssen, D. H. J., Black, M. A., Pyke, K. E., Padilla, J., Atkinson, G.,
Harris, R. A., … Green, D. J. (2011). Assessment of flow-mediated
dilation in humans: A methodological and physiological guideline.
American Journal of Physiology-Heart and Circulatory Physiology,
300, H2–12.
Thijssen, D. H. J., Rowley, N., Padilla, J., Simmons, G. H., Laughlin,
M. H., Whyte, G., … Green, D. J. (2011). Relationship between
upper and lower limb conduit artery vasodilator function in hu-
mans. Journal of Applied Physiology, 111, 244–250. https://doi.
org/10.1152/jappl physi ol.00290.2011
Tremblay, J. C., & Pyke, K. E. (2018). Flow-mediated dilation stim-
ulated by sustained increases in shear stress: A useful tool for as-
sessing endothelial function in humans? American Journal of
Physiology-Heart and Circulatory Physiology, 314, H508–H520.
Trinity, J. D., Groot, H. J., Layec, G., Rossman, M. J., Ives, S. J., Runnels, S.,
… Richardson, R. S. (2012). Nitric oxide and passive limb movement:
A new approach to assess vascular function. Journal of Physiology, ,
590(6), 1413–1425. https://doi.org/10.1113/jphys iol.2011.224741
Tschakovsky, M. E., Rogers, A. M., Pyke, K. E., Saunders, N. R., Glenn,
N., Lee, S. J., … Dwyer, E. M. (2004). Immediate exercise hyper-
emia in humans is contraction intensity dependent: Evidence for
rapid vasodilation. Journal of Applied Physiology, 96, 639–644.
VanTeeffelen, J. W. G. E., & Segal, S. S. (2006). Rapid dilation of arte-
rioles with single contraction of hamster skeletal muscle. American
Journal of Physiology-Heart and Circulatory Physiology, 290, 119–
127. https://doi.org/10.1152/ajphe art.00197.2005
Wagner, P. D. (2008). Systemic oxygen transport and utili-
zation. Journal of Breath Research, 2, 1–12. https://doi.
org/10.1088/1752-7155/2/2/024001
Walker, M. A., Hoier, B., Walker, P. J., Schulze, K., Bangsbo, J.,
Hellsten, Y., & Askew, C. D. (2016). Vasoactive enzymes and blood
flow responses to passive and active exercise in peripheral arterial
disease. Atherosclerosis, 246, 98–105. https://doi.org/10.1016/j.
ather oscle rosis.2015.12.029
Wiedman, M. P. (1963). Dimensions of blood vessels from distributing
artery to collecting vein. Circulation Research, 12, 375–378. https://
doi.org/10.1161/01.RES.12.4.375
How to cite this article: Gifford JR, Hanson BE,
Proffit M, et al. Indices of leg resistance artery
function are independently related to cycling V̇O2max.
Physiol Rep. 2020;8:e14551. https://doi.org/10.14814/
phy2.14551
| Indices of leg resistance artery function are independently related to cycling V̇O<sub>2</sub> max. | [] | Gifford, Jayson R,Hanson, Brady E,Proffit, Meagan,Wallace, Taysom,Kofoed, Jason,Griffin, Garrett,Hanson, Melina | eng |
PMC3409846 | EDITORIAL
Open Access
Born to run. Studying the limits of human
performance
Andrew Murray1* and Ricardo JS Costa2
Abstract
It is recognised that regular physical activity and a high level of fitness are powerful predictors of positive health
outcomes. There is a long and rich history of significant feats of human endurance with some, for example, the
death of the first marathon runner, Pheidippides, associated with negative health outcomes.
Early studies on endurance running used X-ray and interview techniques to evaluate competitors and comment on
performance. Since then, comparatively few studies have looked at runners competing in distances longer than a
marathon. Those that have, tend to show significant musculoskeletal injuries and a remarkable level of adaptation
to this endurance load.
The TransEurope Footrace Project followed ultra-endurance runners aiming to complete 4,500 Km of running in 64
days across Europe. This pioneering study will assess the impact of extreme endurance on human physiology;
analysing musculoskeletal and other tissue/organ injuries, and the body’s potential ability to adapt to extreme
physiological stress. The results will be of interest not only to endurance runners, but to anyone interested in the
limits of human performance.
Please see related article: http://www.biomedcentral.com/1741-7015/10/78
Keywords: Physical inactivity, ultra-marathon, endurance, runners, musculoskeletal, nutrition, hydration, race, Trans-
Continental
Background
Professor Steven Blair describes physical inactivity as “one
of the most important public health challenges of the 21st
Century” [1]. It is recognized that regular physical activity
and a high level of fitness are powerful predictors of posi-
tive health outcomes, with Professor Karim Khan, who is a
prominent sports and exercise medicine researcher, fram-
ing Blair’s data, to show that low fitness may be responsi-
ble for a larger attributable fraction of mortality than
“Smokadiabesity"- that is smoking, diabetes, and obesity
combined [2].
Can there ever be too much of a good thing? Can we
ever do too much physical activity? History suggests the
human body is perfectly adapted to run long distances.
Humans have an unmatched ability in the animal king-
dom to run these distances, capabilities that probably
emerged around 2 million years ago to assist with
persistence hunting - a tactic still used by the San Bush-
men of the Kalahari [3].
History celebrates the run in 490 BC from Marathon to
Athens by Pheidippides as the inspiration for the modern
marathon, whilst remembering that this hero of ancient
Greece died following his exertions. The traditional story
tells that Pheidippides had, in fact, run from Athens to
Sparta, a distance of 240 km in less than 48 hours shortly
before. This would be defined as an ultra-marathon,
which is considered any distance in excess of the stan-
dard marathon distance of 42.195 km. Of interest are the
physiological changes that accompany such extreme chal-
lenges. In a study published in BMC Medicine, Schutz
et al. [4] followed 44 ultra-marathon runners in the
TransEurope Footrace 2009, which is a distance of over
4,487 km from South Italy to North Cape. Here, they
recorded daily sets of data from magnetic resonance ima-
ging, psychometric, body composition and biological
measurements with the aim of uncovering new knowl-
edge on the physiological and pathological changes that
* Correspondence: [email protected]
1SportScotland Institute of Sport, Aithrey Road, Stirling, FK9 5PH, UK
Full list of author information is available at the end of the article
Murray and Costa BMC Medicine 2012, 10:76
http://www.biomedcentral.com/1741-7015/10/76
© 2012 Murray and Costa; licensee BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative
Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and
reproduction in any medium, provided the original work is properly cited.
accompany cellular and organ systems under extreme
strain.
Previous studies: What is known about ultra-endurance
running?
Humans have been racing across continents since the
1928 and 1929 U.S. Trans-Continental Races. In his
seminal work “Lore of Running” Tim Noakes notes that
2 separate medical reports were compiled from the
4,960 km, 84 stage 1928 race. J.T. Farrell et al. con-
cludes “the immediate effects of long-distance running
are inconsequential” from his team’s X-ray studies of
the heart, bones and joints; while Gordon and Baker
concluded only 40 of the 199 competitors were capable
of sustaining this physical workload [5]. Musculoskeletal
injuries and financial difficulties were cited as principal
reasons that only 55 of the 199 competitors finished [5].
Since these early studies, little research has been con-
ducted on extreme endurance runners. Numerous studies
have looked at musculoskeletal injuries in marathon run-
ners, but few exist of athletes running further than this.
Fallon studied musculoskeletal injuries in athletes run-
ning 1,005 km from Sydney to Melbourne finding injuries
to the knee, and ankle to be most prevalent [6]; while
Scheer and Murray amongst others also found lower
limb musculoskeletal injuries to be common [7]. Fallon
also described an injury fairly specific to ultra-endurance
runners, tendinopathy of the ankle dorsiflexors, a condi-
tion subsequently called “Ultra-marathoner’s ankle” [6].
Other studies have looked at immune status, nutrition,
and hydration in ultra-endurance runners. Perturbed
immune function is a common feature after endurance
type exercise, with clinical significance associated with
increased risk of illness, infection, suppressed tissue
repair and wound healing abilities, and exertional heat ill-
nesses [8]. Consuming adequate nutrition to meet nutri-
ent demands and maintaining euhydration has shown to
attenuate some of the immune perturbing effects of
extreme endurance running [9,10]; this not only has
immune and health consequences, but will affect running
performance on consecutive days of competition [11].
A perceived common outcome of running exercise in
the heat is dehydration, and thus much focus has pre-
viously been on promoting hyperhydration strategies dur-
ing ultra-endurance events. Evidence from previous
reviews and preliminary data from Coventry University
actually suggests that fluid-overconsumption behaviours
are a common feature of ultra-endurance running, with
large ingestions of plain water and insufficient sodium
replacement frequently observed [12]. This type of drink-
ing behaviour is associated with the manifestation of hypo-
natraemia, which recently has had much interest, with
cases of asymptomatic and symptomatic hyponatraemia
becoming better recognised during ultra-endurance events
[13].
O’Keefe et al. recently found prolonged endurance
exercise may cause pathologic remodelling of the heart
and be pro-arrythmogenic with atrial fibrillation as much
as five times more prevalent in the population studied
[14]. Moreover, interesting case studies exist. In his book
“Survival of the Fittest” Dr Mike Stroud described the
destruction of the human body when studying the effects
of a brutal 95 day Antarctic crossing, at times burning
>10,000 kcal·day-1 [15]. Unpublished blood values from
Andrew Murray’s ultra-endurance run from Scotland to
the Sahara desert showed a drop in haemoglobin to
10.8 g·dl-1, and a serum ferritin of 2 ng·ml-1, from pre-
viously normal values, despite a dietary iron intake 450%
of recommended nutritional intake (RNI), and body
weight being maintained.
What this study adds: assessing the impact of extreme
endurance on human physiology
The TransEurope Footrace Project followed 44 ultra-
endurance runners aiming to complete 4,500 km of run-
ning in 64 days across Europe. Comfortably the largest
and most comprehensive of its kind to date, it aimed at
demonstrating the feasibility of conducting a longitudinal
study over this period collecting a large and wide ranging
amount of data which included: 741 magnetic resonance
imaging (MRI) examinations, 5,720 urine samples, 244
blood samples, 205 electrocardiogram examinations,
1,018 bioelectrical impedance analysis measurements,
539 anthropological measurements, and 150 psychologi-
cal questionnaires. This pioneering study will assess the
impact of extreme endurance on human physiology; ana-
lysing musculoskeletal and other tissue/organ injuries,
and the body’s potential ability to adapt to extreme phy-
siological stress.
Although running is one of the most popular forms of
recreation worldwide, not many will wish to gallop across
continents. But ultra-endurance running is increasing
dramatically in popularity and this study will be of inter-
est to anyone with an interest in the limits of human per-
formance, and the ability of man to adapt to seemingly
impossible challenges. Comparisons with athletes under-
taking feats of endurance including cycling’s Tour de
France will be interesting.
Conclusions
There is a long and rich history of significant feats of
human endurance. Remarkably, studies have been con-
ducted on trans-continental races since 1928. Similarities
between these early studies and the TransEurope Foo-
trace Project include the distance covered, and the use of
imaging. However, advances in technology has meant
Murray and Costa BMC Medicine 2012, 10:76
http://www.biomedcentral.com/1741-7015/10/76
Page 2 of 3
that the TransEurope Footrace Project has been able to
acquire longitudinal data from a relatively large volunteer
cohort of ultra-marathon runners including data on mus-
culoskeletal, cardiac, and brain MRI, along with a raft of
other data on immune function, hydration and nutrition.
Data is likely to show that competing in such an event
can lead to significant musculoskeletal and other inju-
ries, but also that the human body is capable of adapting
to incredible endurance loads, and can run well in
excess of a marathon per day despite seemingly signifi-
cant medical issues. Like the runners, research in this
field will continue to move forward.
Author details
1SportScotland Institute of Sport, Aithrey Road, Stirling, FK9 5PH, UK.
2Department of Health Professions, Coventry University, Priory Road,
Coventry, CV1 5FB, UK.
Authors’ information
AM and RC have completed over 100 ultra-marathons between them. AM
completed a run across Europe in 2011 and is a Sports and Exercise
Medicine doctor. He has worked at numerous ultra-marathon events with
Marathon Medical Services. RC is a former professional triathlete and is
currently a Senior Lecturer and Researcher Fellow in Dietetics and Human
Nutrition at Coventry University. AM and RC have both produced original
research from ultra-marathon competition.
Received: 13 July 2012 Accepted: 19 July 2012 Published: 19 July 2012
References
1.
Blair SN: Physical inactivity: The biggest public health problem of the
21st Century. Br J Sports Med 2009, 43:1-2.
2.
Khan KM, Tunaiji HA: As different as Venus from Mars: time to distinguish
efficacy (can it work?) from effectiveness (does it work?). Br J Sports Med
2011, 45:759-760.
3.
Lieberman DE, Bramble DM: The evolution of marathon running:
capabilities in humans. Sports Med 2007, 37:288-90.
4.
Schulz , et al:, (to be added when published).
5.
Noakes T: From Learning from the experts in Lore of Running.Edited by:
Noakes T. Oxford: Oxford University Press; 2001:361-483.
6.
Fallon KE: Musculoskeletal injuries in the ultra-marathon: the 1990
Westfield Sydney to Melbourne run. Br J Sports Med 1996, 30:319-323.
7.
Scheer BV, Murray AD: Al Andalus Ultra Trail: An Observation of Medical
Interventions During a 219-km, 5-Day Ultramarathon stage race. Clin J
Sports Med 2011, 21:444-446.
8.
Walsh NP, Gleeson M, Shephard RJ, Gleeson M, Woods JA, Bishop NC,
Fleshner M, Green C, Pedersen BK, Hoffman-Goetz L, Rogers CJ, Northoff H,
Abbasi A, Simon P: Position statement. Part one: Immune function and
exercise. Exerc Immunol Rev 2011, 17:6-63.
9.
Walsh NP, Gleeson M, Pyne DB, Nieman DC, Dhabhar FS, Shephard RJ,
Oliver SJ, Bermon S, Kajeniene A: Position statement. Part two:
Maintaining immune health. Exerc Immunol Rev 2011, 17:64-103.
10.
Costa RJS, Walters R, Bilzon JLJ, Walsh NP: Effects of immediate
postexercise carbohydrate ingestion with and without protein on
neutrophil degranulation. Int J Sport Nutr Exerc Metab 2011, 21(3):205-213.
11.
American College of Sports Medicine, American Dietetic Association,
Dietitians of Canada: American College of Sports Medicine position stand.
Nutrition and athletic performance. Med Sci Sports Exerc 2009,
41(3):709-731.
12.
Hew-Butler T, Ayus JC, Kipps C, Maughan RJ, Mettler S, Meeuwisse WH,
Page AJ, Reid SA, Rehrer NJ, Roberts WO, Rogers IR, Rosner MH, Siegel AJ,
Speedy DB, Stuempfle KJ, Verbalis JG, Weschler LB, Wharam P: Statement
of the Second International Exercise-Associated Hyponatremia
Consensus Development Conference. Clin J Sport Med 2007, 18(2):111-121.
13.
Noakes TD, Sharwood K, Speedy D, Hew T, Reid S, Dugas J, Almond C,
Wharam P, Weschler L: Three independent biological mechanisms cause
exercise-associated hyponatremia: evidence from 2,135 weighed
competitive athletic performances. Proc Natl Acad Sci USA 2005,
102(51):18550-18555.
14.
O’Keefe JH, Patil HR, Lavie CJ: Potential Adverse Cardiovascular Effects
From Excessive Endurance Exercise. Mayo Clinic Proceedings 2012,
87(6):587-595.
15.
Stroud M: Survival of the Fittest.Edited by: Stroud M. London: Yellow
Jersey Press; 2004:.
Pre-publication history
The pre-publication history for this paper can be accessed here:
http://www.biomedcentral.com/1741-7015/10/76/prepub
doi:10.1186/1741-7015-10-76
Cite this article as: Murray and Costa: Born to run. Studying the limits of
human performance. BMC Medicine 2012 10:76.
Submit your next manuscript to BioMed Central
and take full advantage of:
• Convenient online submission
• Thorough peer review
• No space constraints or color figure charges
• Immediate publication on acceptance
• Inclusion in PubMed, CAS, Scopus and Google Scholar
• Research which is freely available for redistribution
Submit your manuscript at
www.biomedcentral.com/submit
Murray and Costa BMC Medicine 2012, 10:76
http://www.biomedcentral.com/1741-7015/10/76
Page 3 of 3
| Born to run. Studying the limits of human performance. | 07-19-2012 | Murray, Andrew,Costa, Ricardo J S | eng |
PMC6843975 | International Journal of
Environmental Research
and Public Health
Article
Analysis of the Association between Running
Performance and Game Performance Indicators in
Professional Soccer Players
Toni Modric 1,2, Sime Versic 1,2, Damir Sekulic 1,*
and Silvester Liposek 3,*
1
Faculty of Kinesiology, University of Split, 21000 Split, Croatia; [email protected] (T.M.);
[email protected] (S.V.)
2
HNK Hajduk Split, 21000 Split, Croatia
3
University of Maribor, 2000 Maribor, Slovenia
*
Correspondence: [email protected] (D.S.); [email protected] (S.L.); Tel.: +385-21-302-440 (D.S.)
Received: 2 October 2019; Accepted: 18 October 2019; Published: 21 October 2019
Abstract:
Running performance (RP) and game performance indicators (GPI) are important
determinants of success in soccer (football), but there is an evident lack of knowledge about
the possible associations between RP and GPI. This study aimed to identify associations between RP
and GPI in professional soccer players and to compare RP and GPI among soccer playing positions.
One hundred one match performances were observed over the course of half of a season at the
highest level of national competition in Croatia. Players (mean ± SD, age: 23.85 ± 2.88 years; body
height: 183.05 ± 8.88 cm; body mass: 78.69 ± 7.17 kg) were classified into five playing positions
(central defenders (n = 26), full-backs (n = 24), central midfielders (n = 33), wide midfielders (n = 10),
and forwards (n = 8). RP, as measured by global positioning system, included the total distance
covered, distance covered in five speed categories (walking, jogging, running, high-speed running,
and maximal sprinting), total number of accelerations, number of high-intensity accelerations, total
number of decelerations, and number of high-intensity decelerations. The GPI were collected by the
position-specific performance statistics index (InStat index). The average total distance was 10,298.4
± 928.7 m, with central defenders having the shortest and central midfielders having the greatest
covered distances. The running (r = 0.419, p = 0.03) and high-intensity accelerations (r = 0.493,
p = 0.01) were correlated with the InStat index for central defenders. The number of decelerations
of full-backs (r = −0.43, p = 0.04) and the distance covered during sprinting of forwards (r = 0.80,
p = 0.02) were associated with their GPI obtained by InStat index. The specific correlations between
RP and GPI should be considered during the conditioning process in soccer. The soccer training
should follow the specific requirements of the playing positions established herein, which will allow
players to meet the game demands and to perform successfully.
Keywords: GPS; football; accelerations; decelerations; efficacy
1. Introduction
Soccer is a highly complex team sport with changing dynamics and multistructural movements
played by two teams. Each team consists of 10 outfield players and a goalkeeper and the final game
achievement depends directly on the performance of all 11 players [1,2]. Therefore, performance
analysis is crucial in the evaluation of players’ achievement [3]. The global popularity of soccer has led
to the implementation of scientific and technological knowledge in everyday use, and this is particularly
evident within the field of performance analysis. One of the important aspects of performance analysis
is termed “running performance”, which is nowadays mostly evidenced by global positioning software
systems (GPS) [4].
Int. J. Environ. Res. Public Health 2019, 16, 4032; doi:10.3390/ijerph16204032
www.mdpi.com/journal/ijerph
Int. J. Environ. Res. Public Health 2019, 16, 4032
2 of 13
GPS technology is known to be highly applicable in evaluation of mobility and physical activity
patterns within the field of public health [5–7]. With the improvement of their accuracy/precision,
design, usability and safeness (Figure 1), the GPS-based devices are becoming prevalent even in
competitive sports, including soccer [8,9].
Figure 1. Global positioning device (GPS) used for the measurement of running performances in soccer.
Specifically, GPS allows collecting data about players’ running performance, such as the total
distance covered, the distance covered at different intensities (i.e., speeds), and the number of
accelerations and decelerations. Studies conducted so far have provided playing-position-specific
evidence with regard to running at different intensities, with midfielders covering the largest total
distance and wingers performing the most high-intensive sprints [10]. Furthermore, match running
performance in Brazilian professional soccer players indicated that winning teams, home playing
teams and teams that play against “weaker” opponents had the greatest total distance covered [11].
A study performed with <21- and <18-year old soccer players found that a 3–5–2 formation elicited
the highest total distance, with a 4–2–3–1 formation eliciting the highest number of accelerations
and decelerations [12]. The results of the previously cited studies that used GPS technology as
a measurement tool were generally consistent with those from investigations where authors used
different video-based computerized match analysis systems in the evaluation of players’ running
performances [13–17].
Game performance indicators are another set of variables that are used in performance analysis
in soccer. Basically, game performance indicators are defined as a “selection and combination of
variables that define some aspect of performance and that help achieve athletic success” [18]. The most
frequently used game performance indicators are passes, shots, crosses, dribbles, challenges etc. [19].
Currently, numerous video-based platforms that track performance indicators of soccer players are
available (InStat, Optasport, Wyscout). Such platforms quickly and accurately provide a large range of
data about game performance indicators, allowing the simultaneous analysis of the physical efforts,
movement patterns, and technical actions of players, both with and without the ball [20–22].
Previous studies conducted in the field of performance analysis in soccer found that both the
physical (i.e., total distance covered, high-intensity running, accelerations and decelerations) and
technical–tactical performances (i.e., shots, crosses, challenges, and dribbles) of players were correlated
with specific conditions such as match outcome (win/draw/loss), match location (home/away), type of
match (league/cup/friendly), and strength of the opponent team [19,23–28]. Situational variables, such as
ball possession, total shots, shots on target, crosses, dribbles, clearances, challenges, and interceptions,
and their influence on technical–tactical parameters were mostly evaluated by the variation of counts
of technical match actions, which include shots, passing, tackles, aerial duels, and dribbles [18,27,29,30].
Briefly, situational variables that discriminate among winning, drawing and losing were mostly those
related to ball possession and offensive actions (e.g., total shots, shots on goal, and crosses) [29,30],
Int. J. Environ. Res. Public Health 2019, 16, 4032
3 of 13
while some studies found that indicators of defensive efficacy (e.g., interceptions, clearance, and aerial
challenges) were the variables most related to the match outcome [27].
Although game performance indicators and running performance are often investigated separately,
to the best of our knowledge, there is no study that simultaneously observed both groups of performance
variables during official soccer matches. Additionally, there is no information about the relationship that
may exist between these two groups of variables. Therefore, the aim of this study was to identify possible
associations that may exist between running performance and game performance in professional
soccer players. Additionally, running performance and standard soccer performance variables were
compared among playing positions. Authors were of the opinion that a study of this type would
allow a better understanding of the relationships that exist between running performance and game
performance indicators and that such understanding would therefore improve the applicability of both
sets of variables in soccer training and competition.
2. Materials and Methods
2.1. Participants and Design
The participants in this study were professional soccer players from Croatia (mean ± SD, age:
23.85 ± 2.88 years; body height: 183.05 ± 8.88 cm; body mass: 78.69 ± 7.17 kg), and all were members
of one team competing at the highest national. Players were observed over one competitive half
season, resulting in 101 match performances which were used as cases for this study. All data were
collected during 14 matches of the Croatian Soccer League 2018/2019 season, and for the purpose
of this study only the results of those players who participated in the whole game were analyzed.
Players were classified in five groups based on playing positions: central defenders (CD; n = 26),
full-backs (FB; n = 24), central midfielders (CM; n = 33), wide midfielders (WM; n = 10), and forwards
(FW; n = 8), as suggested previously [10]. Sociodemographic and anthropometric data of observed
players are presented in Table 1. In the observed half-season, the team played seven home and seven
guest matches, with three wins, eight draws and three losses. At the end of the observed half-season,
the team ranked 6th of 10 teams which competed in Croatian Soccer League. The investigation was
approved by Ethical Board of the University of Split, Faculty of Kinesiology, Split, Croatia (approval
number: 2181-205-02-05-19-0020).
Table 1. Sociodemographic and anthropometric characteristics of the studied players with differences
among playing positions (F-test).
Age (years)
Body height (cm)
Body mass (kg)
Mean ± SD
Mean ± SD
Mean ± SD
Total sample (n = 101)
23.85 ± 2.88
183.05 ± 6.88
78.69 ± 7.17
Central Defenders (n = 26)
23.25 ± 2.21
192.25 ± 5.61
87.27 ± 7.38
Full-Backs (n = 24)
23.2 ± 3.56
176.6 ± 3.36
73.4 ± 4.34
Central Midfielders (n = 33)
22.66 ± 2.73
175 ± 6.08
76.51 ± 5.02
Wide Midfielders (n = 10)
26.0 ± 1.0
183 ± 3.46
76.2 ± 4.17
Forwards (n = 8)
27.0 ± 2.82
181.5 ± 0.7
85 ± 4.1
F-test (p)
1.51 (0.24)
5.92 (0.01)
5.04 (0.01)
2.2. Procedures
The variables in this study were two sets of soccer performance variables (running performance
and game performance indicators) and the final game outcome (observed as loss, draw, win).
Data on the running performance of the players were collected by GPS technology (Catapult S5
and X4 devices, Melbourne, Australia) with a sampling frequency of 10 Hz. Such device was already
investigated for metrics, and was found to be appropriately reliable and valid in sport settings (i.e., less
than 1% measurement error, and 80% of common variance with running speed measured by timing
gates) [31,32].
Int. J. Environ. Res. Public Health 2019, 16, 4032
4 of 13
The variables included the following:
total distance covered (m); distance in five speed
categories (walking (<7.1 km/h), jogging (7.2–14.3 km/h), running (14.4–19.7 km/h), high-speed running
(19.8–25.1 km/h), and maximal sprinting (>25.2 km/h)); number (frequency) of total accelerations
(>0.5 m/s2); number of high-intensity accelerations (>3 m/s2); number of total decelerations (less than
–0.5 m/s2) and number of high-intensity decelerations (less than –3 m/s2).
The game performance indicators for each player were determined by the position-specific InStat
index (InStat, Moscow, Russia). The InStat index is calculated on the basis of a unique set of key
parameters for each playing position (12–14 performance parameters, depending on the position
during the game), with a higher numerical value indicating better performance. The exact calculations
are trademarked and known only to the manufacturer of the platform. In most general terms, an
automatic algorithm considers the player’s contribution to the team’s success, the significance of their
actions, opponent’s level and the level of the competition they play in (i.e., the same performance done
in European Champions League and some national-level first division will not be rated with same
values). The rating is created automatically, and each parameter has a factor which changes depending
on the number of actions and events in the match. The weight of the action factors differs depending
on the player’s position. For example, grave mistakes done by CD and their frequency affect InStat
index to a greater extent than those done by FWD. The key factors included in the calculation of the
InStat index are position specific and include tackling, aerial duels, set pieces in defense, interceptions
(for CD); number of crosses, number of passes to the penalty area, pressing (for FB); playmaking,
number of key passes, finishing (for CM); pressing, dribbling, finishing, counterattacking (for WM);
shooting, finishing, pressing, dribbling (for FWD). In order to calculate the InStat Index, the player
has to spend a certain amount of time on the field and perform a minimum number of actions, but in
this study this issue was solved simply by including only those players who played the whole game
(as explained in Section 2.1).
2.3. Statistics
The normality of the distributions was checked by the Kolmogorov–Smirnov test, and the data are
presented as the means ± standard deviations. The homoscedasticity of all variables was confirmed by
Levene’s test. The statistical analyses were performed throughout several phases.
In the first phase the data obtained by InStat index were associated with final game outcome
by one-way analysis of variance (ANOVA). For this procedure the game outcomes (loss, draw, win)
were considered as the grouping (independent) variable, and differences were established for total
sample of players, and separately for each playing position. This allowed identification of the validity
of the InStat index as an indicator of the final game achievement for the total sample, and for the five
observed playing positions.
The second phase of data analysis comprised calculation of differences among playing positions
in running performance and InStat index. This was done by ANOVA with a consecutive Scheffe
post hoc test. Throughout these analyses the information of running performance specifics for each
playing position were obtained. Also, the analysis of differences in InStat allowed identification of
the applicability of the InStat index for the analysis of game achievement for each playing position.
To evaluate the effect sizes (ES), partial eta-squared values (η2) were presented (small ES: >0.02; medium
ES: >0.13; large ES: >0.26) [33].
In the third phase, the associations between running performance (obtained by GPS) and game
performance indicators (evaluated by InStat) were identified by calculating Pearson’s product moment
correlation coefficients.
For all analyses, Statistica 13.0 (TIBCO Software Inc., Greenwood Village, CO, USA) was used,
and a p < 0.05 was applied.
Int. J. Environ. Res. Public Health 2019, 16, 4032
5 of 13
3. Results
The ANOVA indicated significant (p < 0.05) association between the InStat index and match
outcome for the total sample (n = 101, F-test: 23.69, η2 = 0.30 (large E)), CD (n = 26, F-test: 3.89,
η2 = 0.24 (medium ES)), FB (n = 24, F-test: 4.98, η2 = 0.31 (large ES)) and CM (n = 33, F-test: 15.71,
η2 = 0.50 (large ES)). The InStat index was not significantly associated with game outcomes for WM
(n = 10, F-test: 0.98, η2 = 0.21 (medium ES)), and FW (n = 8, F-test: 2.61, η2 = 0.52 (large ES)) (Figure 2).
Figure 2. InStat index in relation to the outcome of the match for total sample (Total) and different
playing positions (CD, central defenders; FB, full-backs; CM, central midfielders; WM, wide midfielders;
FW, forwards); * indicates statistically significant differences at p < 0.05 derived by analysis of variance
The descriptive parameters for running performances and InStat index in total sample, and for
each playing positions are presented in Table 2. Significant ANOVA differences were found among
playing positions (p < 0.05) in all running performances, with large ES for differences in: (i) total
distance covered (η2 = 0.59); (ii) distance covered while jogging (η2 = 0.41); (iii) running (η2 = 0.62);
(iv) high-speed running (η2 = 0.53); (v) sprinting (η2 = 0.39); (vi) number of performed accelerations
(η2 = 0.27); (vii) number of decelerations (η2 = 0.45); (viii) number of high-intensity accelerations
(η2 > 0.30); and (ix) number of high-intensity decelerations (η2 = 0.41). Small ES was found for
differences in distance covered while walking (η2 = 0.11) (Table 3).
Specifically, CM covered the longest total distance (significant post-hoc differences when compared
to all other playing positions), the longest distance in jogging (significant post-hoc differences when
compared to all other playing positions), and the longest distance while running (significantly different
from CD and FB). WM covered the longest distance in high-speed running, and in sprinting (significant
post-hoc differences to CD, CM, and FW). CD carried out the highest number of accelerations and
highest number of decelerations (significantly different from FW). Finally, FW carried out the highest
number of high-intensity accelerations (significant post-hoc differences when compared to CD, FB, and
WM) and high-intensity decelerations (significantly different to WM) (Table 3).
The total running distance and high-intensity accelerations were correlated with the InStat index
for CD (r = 0.42 and r = 0.49, respectively). Furthermore, the number of decelerations was significantly
correlated with the InStat in FB (r = –0.43), while distance covered during sprinting was correlated
with InStat index in FW (r = 0.80). In general, the running performances of players in central and wide
midfield positions were not significantly associated with the InStat index (Table 4).
Int. J. Environ. Res. Public Health 2019, 16, 4032
6 of 13
Table 2. Descriptive statistics for running performances and game performance indicator (InStat).
Variables
Total
Central Defenders
Full-Backs
Central Midfielders
Wide Midfielders
Forwards
Mean ± SD
Mean ± SD
Mean ± SD
Mean ± SD
Mean ± SD
Mean ± SD
Total distance (m)
10,298.4 ± 928.68
9313.5 ± 599.4
10,368 ± 612
11,155.1 ± 635.3
10,264.8 ± 275.2
9796.7 ± 703.7
Walking (m)
4220.57 ± 362.33
4076.6 ± 378.3
4297.9 ± 338.5
4258.5 ± 340.7
4074.8 ± 194.3
4482.1 ± 442.2
Jogging (m)
4092.94 ± 569.73
3859 ± 380.2
3975.4 ± 372.8
4599.7 ± 471.4
3761.2 ± 324.1
3530 ± 729.9
Running (m)
1363.27 ± 339.68
999.2 ± 197.7
1320.7 ± 236.1
1674.9 ± 226.1
1526.5 ± 117.4
1184.4 ± 207.9
High-speed running (m)
461.83 ± 160.15
288.2 ± 63.8
533.9 ± 134.1
492.7 ± 139.9
640.7 ± 105.4
458.7 ± 94.7
Sprinting (m)
155.89 ± 97.13
87.7 ± 59.9
236.6 ± 97.2
123.7 ± 69.5
260.6 ± 68.8
137.1 ± 46.9
Accelerations (count)
716.19 ± 73.15
743.5 ± 56.2
710 ± 66.2
733.4 ± 72.4
688 ± 34.2
610.1 ± 83.7
Decelerations (count)
674.44 ± 69.29
714.1 ± 51.5
672.4 ± 56
681.9 ± 55.8
661.8 ± 36.7
536.6 ± 69
High-intensity accelerations (count)
3.16 ± 2.67
2.5 ± 1.8
3.1 ± 1.7
1.9 ± 2.2
7 ± 2.6
6 ± 2.9
High-intensity decelerations (count)
11.39 ± 6.27
6.1 ± 2.8
13.1 ± 4.9
11.5 ± 5.9
20.8 ± 5.5
11 ± 3.1
InStat (index)
284.5 ± 31.04
247.4 ± 29.2
243 ± 28.7
254.1 ± 29.3
251.1 ± 32.1
242 ± 49.5
Table 3. Differences among playing positions for running performances and game performance indicator (InStat) determined by analysis of variance (ANOVA), with
Scheffe post-hoc test differences.
Variables
ANOVA
Effect Size
Post hoc
F (p)
η2
Central Defenders
Full-Backs
Central Midfielders
Wide Midfielders
Forwards
Total distance (m)
35.02 (0.01)
0.59
FB, CM, WM
CD, CM
CD, FB, WM, FW
CD, CM
CM
Walking (m)
3.18 (0.02)
0.11
-
-
-
-
-
Jogging (m)
16.71 (0.01)
0.41
CM
CM
CD, FB, WM, FW
CM
CM
Running (m)
39.30 (0.01)
0.62
FB, CM, WM
CD, CM
CD, FB
CD, FW
CM
High-speed running (m)
29.30 (0.01)
0.53
FB, CM, WM, FW
CD
CD, WM
CD, CM, FW
CD, WM
Sprinting (m)
15.72 (0.01)
0.39
FB, WM
CD, CM, FW
FB, WM
CD, CM, FW
FB, WM
Accelerations (count)
9.06 (0.01)
0.27
FW
FW
FW
CD, CM, FW
Decelerations (count)
20.11 (0.01)
0.45
FW
FW
FW
FW
CD, FB, CM, WM
High-intensity accelerations (count)
8.53 (0.01)
0.30
WM, FW
WM, FW
WM, FW
CD, FB, CM
CD, FB, WM
High-intensity decelerations (count)
16.70 (0.01)
0.41
FB, CM, WM
CD, WM
CD, WM
CD, FB, CM, FW
WM
InStat (index)
0.64 (0.62)
0.03
-
-
-
-
-
Superscripted letters indicate significant post-hoc differences when compared to specific playing position (CD, central defenders; FB, full-backs; CM, central midfielders; WM – wide
midfielders; FW, forwards).
Int. J. Environ. Res. Public Health 2019, 16, 4032
7 of 13
Table 4. Pearson’s product moment correlations between running performances and game performance
indicator (InStat) for different playing positions.
Variables
Total
(n = 101)
Central
Defenders
(n = 26)
Full-Backs
(n = 24)
Central
Midfielders
(n = 33)
Wide
Midfielders
(n = 10)
Forwards
(n = 8)
Total distance
0.08
0.18
–0.04
–0.02
–0.17
0.01
Walking
−0.02
0.09
0.01
–0.12
0.07
0.04
Jogging
0.05
0.02
–0.10
0.05
–0.41
–0.05
Running
0.16
0.42 *
0.02
0.12
0.01
–0.13
High-speed running
0.02
–0.04
–0.06
–0.10
0.54
0.17
Sprinting
0.01
–0.24
0.17
–0.04
0.22
0.80 *
Accelerations
–0.01
0.12
–0.39
0.07
–0.24
–0.02
Decelerations
–0.09
0.07
–0.43 *
–0.05
–0.26
–0.33
High-intensity
accelerations
0.18
0.49 *
0.20
0.29
–0.08
0.26
High-intensity
decelerations
0.05
0.29
–0.04
–0.01
0.44
–0.18
* denotes statistical significance of p < 0.05.
4. Discussion
With regard to study aims there are two most important findings. First, the total distance covered
and the intensity of running varied according to the different playing positions. Second, running
performance parameters (e.g., the number of accelerations or decelerations and the distance covered
in different speed zones) affect successful performance in soccer for some playing positions. Prior to
discussion of these findings, an overview of the analyses done in order to evaluate the applicability
and validity of InStat index as a measure of final match outcome will be provided.
Studies have already investigated the association between different variables explaining situational
efficacy (i.e., game performance indicators) and match outcomes. For example, when losing the game,
teams had more ball possession [30,34,35] and performed more crosses and dribbles [27]. Additionally,
when winning, the teams performed more interceptions, clearances and aerial challenges, fewer
passes and dribbles [27], and less high-intensity activities [18,34]. However, previous studies regularly
investigated the performance indicators of the whole team, while there has been limited research
investigating the position-specific performances in relation to game outcome, even though technical
indicators have been considered good predictors of soccer match success [36]. Also, the quality of
technical skills in real-game performance, which is actually obtained throughout the InStat index and
other similar platforms, has been included as a main component in soccer talent identification and
development systems [37,38].
InStat index in soccer is based on wide range of team- and individual-statistics, which are linked
to the supporting video episodes. At the final stage, the calculated index should be related to final
game outcome, and consequently should be a valid measure of final team achievement (i.e., game
outcome). Results of this study indicated significant differences among game outcomes (loss, draw,
or win) in InStat index for the total sample and specifically for CD, FB and CM. Although the statistical
significance of the F-test did not reach statistical significance for WM and FW, this may be attributed to
small number of players in these groups (WM: 10 players, FW: 8 players) and consequent small number
of degrees of freedom [39]. Therefore, it might be said that the results presented here confirmed the
validity of InStat in evaluation of final game achievement in Croatian professional soccer. It is also
important to note that InStat index is specifically calculated for different positions on the basis of
position-specific parameters (please see Section 2 for more details). Therefore, the lack of differences
among playing positions in InStat (please see Tables 2 and 3 for more details) indicates that this index
might be observed as an applicable measure of position-specific game performance in soccer.
Int. J. Environ. Res. Public Health 2019, 16, 4032
8 of 13
4.1. Running Performances and Differences Among Playing Positions
Considering the different tactical roles of different playing positions in soccer games, recent studies
confirm that the distance covered during the match appears to be related to playing position [11,14,16,20].
Results of this study evidenced significant differences in running performance among playing positions,
and such results are generally in agreement with previous studies that investigated these issues in the
English Premier League, the Spanish first division, the Italian Serie A, the French League 1, and the
Brazilian first division [13,14,16,20,40]. Specifically, analysis of the Brazilian first division evidenced
that the total distance covered by FB, CM and WM was greater than that covered by CD and FW [13].
Supporting this, the lowest total distance was found for CD (9313 m on average). At the same time,
CM covered significantly more distance than players in all other positions (11,155 m, on average),
which is known to be related to specific playing duties (i.e., CM are responsible for the connection
between defense and attack, and such tactical roles require them to achieve greater distances) [14,16].
Previous studies performed indicated 10.7 km as the average total distance covered in Spanish
and English top divisions [10,40,41]. Meanwhile players observed herein covered total distance of
10.3 km in average. Therefore, it seems that the total distance covered is not the factor that distinguishes
Croatian players from those playing in elite European divisions. On the other hand, there is an evident
difference in the intensity of running. More precisely, top-level European soccer players cover 10% of
the total distance at a high intensity, which includes high-speed running and maximal sprinting [17,42].
Meanwhile, here presented results indicated that Croatian players perform 6.4% of the total distance
covered at a high-intensity running pace.
It is generally accepted that low-intensity activities, such as walking and jogging, are not crucial
in elite soccer performance [43]. However, knowledge of these indicators is important to properly
understand the position-specific demands. Thus, considering the percentage of the total distance,
the most time spent walking and jogging is observed in CD (an average of 85.2% (7935 m) of their total
distance covered (9313 m)). On the other hand, the least time spent walking and jogging is observed in
WF (76.3%), followed by CM (79.4%), FB (79.8%) and FW (81.8%). Collectively, these findings support
previous considerations that Croatian first division players generally play at a lower game pace when
compared to elite European national division players, who spend a much lower percentage of time in
low-intensity activities (from 74.9% to 79.6% of total distance) [10].
The distance covered while jogging among CM is significantly higher than for any other
position. As mentioned before, CM had the greatest total distance, which is directly influenced
by the distance covered while jogging. Furthermore, CM have the greatest distances in the “running
zones” (e.g., 14.4–19.7 km/h). Therefore, results support the findings from previous studies in which
authors reported similar figures and concluded that the physical performance of CM is characterized by
covering a high overall distance, especially at moderate to high speeds such as jogging and running [10].
High-intensity activities are usually defined as all activities with running speeds of 19.8 km/h
and above, and the distance covered at high intensities has been traditionally identified as a key
performance indicator of physical match performance [44] and one of the crucial elements of success
in soccer [43]. The results showed that the greatest amount of high-intensity running (high-speed
running + sprint running) is covered by WM, while the CD have the lowest values for these indicators.
This is consistent with previous investigations in which authors reported similar results for the English
Premier League and the Spanish first division [10,14,20,40].
It is known that outside players (e.g., WM and FB) perform significantly more sprints than
players in central playing positions [14]. Supporting this, our results showed that the greatest sprint
distance was covered by WM and FB. However, despite similar differences among playing positions
between our study and previous studies, values of high-intensity running in Croatian players were
evidently lower than those from the best European national competitions [10,40]. More specifically,
the mean high-intensity distance covered among all playing positions in the English Premier League
was 936 m, in the Spanish first division an average of 821 m was reported, while the average value for
high-intensity running in Croatian players was 652 m.
Int. J. Environ. Res. Public Health 2019, 16, 4032
9 of 13
The highest number of accelerations and decelerations was found for CD and the lowest for FW,
which is consistent with some similar studies on friendly matches in the Spanish first division [10].
Specifically, one of the most important tactical roles of FW is to keep the ball in possession in the
central position, so it is expected that FWs do not cover a large distance. On the other hand, CD
must be constantly prepared for defensive reactions. While trying to find appropriate positioning,
they frequently change running directions, but also the type of running (i.e., frontal running to make
a defensive line to catch opposing players in offsides and lateral shuffles to obtain better positions
versus FW). This certainly results in a high number of accelerations and decelerations for CD. However,
the kind of accelerometer unit and the way that the data are mathematically treated could have
a significant effect on the calculation of accelerations and decelerations, which actually limits the
comparability between different studies [10]. Specifically, while the capacity to accelerate and decelerate
plays a critical role in elite soccer, as it represents high energy demanding activities, the determination
of accelerations might still have unresolved methodological issues [10].
4.2. Associations between Game Performance Indicators and Running Performances— Playing Position
Approach
The results suggest that CD covered the shortest distance while running out of all playing
positions, and this is in agreement with previous reports where authors found that CD exert the fewest
high-intensity efforts compared to all other playing positions [10,14,20,40]. This is understandable
knowing that their technical roles (i.e., aerial duels, tackles, positioning, and interceptions of the balls
passed to the attackers) are generally more focused on the reactions or accelerations and then on
high-speed running. As a result, most of their high-intensity efforts are performed in the zone of
running (14.4–19.7 km/h) simply because they do not have many opportunities to develop running
speeds above the high-intensity zone threshold (>19.8 km/h). However, because of the positive
correlation between the InStat index and the distance covered while running (14.4–19.7 km/h) for
CD, running should be considered an important determinant that affects success for this position.
Furthermore, a positive correlation between the numbers of high-intensity accelerations and the InStat
index among CD shows that a greater number of high-intensity accelerations directly affects real game
performance for this playing position. Specifically, stepping out to the duels and putting pressure
on opponent players are two of the most important tactical roles of CD. If performed rapidly and
aggressively (in other words, with a high acceleration), the chances of winning a duel increase, which
consequently has positive repercussions on final match achievement as well.
The total number of decelerations was inversely associated with the InStat index among FB,
meaning that a higher number of decelerations negatively affected real game efficacy for that playing
position. Although FB are basically defensive players and their starting tactical line-up is in the first
third of the pitch, the main technical requirements for FB are the number of entries to the third part of
the pitch (i.e., pressing) and the number of crosses [38,45]. These duties are actually performed on the
opponent’s half. Therefore, some of the most important tactical roles of the FB are actually in attacking.
To create more of these activities, FB frequently have to move away from the starting tactical line-up,
which actually enables them to make crosses and press. Consequently, if FB have a higher number
of stoppings (i.e., decelerations), it probably negatively affects their ability to participate in attacking
actions and to perform crosses and entries to the third part of the pitch. Collectively, it seems that
soccer success of FB is more affected by their attacking activities, regardless of the fact that they are
defensive players.
Previously, it was highlighted that FW had evidently shorter sprinting distances than players in
the same position during games from other European competitions (please see previous discussion
for details). However, results indicated a strong correlation between the InStat index and the sprint
distance for this playing position, which led to conclusion that the sprint distance covered during
the game was a highly important determinant of overall game performance for FW. Indeed, FW are
positioned close to the opponent’s goal, and almost every sprint presents the opportunity to perform
Int. J. Environ. Res. Public Health 2019, 16, 4032
10 of 13
attacking actions. In addition, FW have the lowest number of tackles, interceptions and clearances
compared with other playing positions [38], which suggests that most of their activities are focused on
attacking. With the higher number of attacking actions, there is a growing chance to enter the penalty
area, shoot, and score. As a result, the number of attacking situations increases the likelihood for
positive game outcomes [28–30].
The main role of CM is to organize the offense by proper ball control and passes, rather than
by invasion into the opponent’s area [38]. Considering the lack of significant association between
running parameter and the InStat index for this playing position, it seems that CM soccer success is
more influenced by some variables, other than those obtained by GPS, such as ball possession, number
of key passes, dribbles, and shots. Also, the running indicators obtained by GPS measurements were
not correlated to InStat variables in WM, which may be observed as surprising since WM experience
the greatest physical requirements during the game, both in terms of total distance covered and
high-intensity running [10]. The possible explanation may be the previously discussed finding of the
small amount of overall distance covered in the studied Croatian players, which actually resulted in
truncated variance and consequently statistically/mathematically decreased the possibility of achieving
significant correlations.
4.3. Limitations and Strengths
The main limitation comes from the fact that this study observed only one team which was
observed during one half season. Therefore, some specific covariates (limited number of observed
players, strength of the opponent, specific tactical requirements) may influence reported results. Next,
in this study no data were collected about psycho-physiological responses of the players (e.g., heart rate
and RPE), which are known to be important determinants of overall performance. Further, this study
actually studied relatively simple “game-related outcomes” (i.e., running performances obtained by
GPS and game performance indicators obtained by InStat index), while sport-performances, especially
those in team sports, are far more complex (i.e., include interaction, cooperation, and opposition) [46].
Also, in this study relatively simple methodology was applied, while complex systems like sport games
may ask for more detailed experimental approaches and the use of mixed methods as an observational
methodology [47].
On the other hand, this study has several strengths. First, this is one of the first studies which
simultaneously evaluated two sets of performance variables (i.e., running performances and game
performance indicators) and probably the first one where associations between these two groups of
performances were analyzed. Also, the data were collected during official games, among professional
players, and at the highest national competitive level. Therefore, results are generalizable to similar
samples of participants and levels of competition. Furthermore, the position-specific approach in
identification of the relationships between running performances and game performance indicators is
important strength of the investigation. Therefore, despite the evident limitations, the authors believe
that this study may contribute to the knowledge on this field and initiate further research.
5. Conclusions
The total distance covered during the match did not distinguish Croatian first division players
from players who compete in elite European divisions (i.e., Spain and Germany). However, the players
studied here achieved total distances at lower running speeds than their peers involved in top-level
European competitions, which clearly indicates the lower game pace in Croatian soccer competition.
These findings can be useful for determining the physical requirements and profiles of the players
in the Croatian first division, especially with regard to international competitions (i.e., the European
League, Champions League).
This study confirmed the association between the running performance of players involved in
certain playing positions and overall game performance. Specifically, it seems that CD distance in
Int. J. Environ. Res. Public Health 2019, 16, 4032
11 of 13
the running zone and number of high-intensity accelerations, FB number of decelerations, and FW
sprinting distance are crucial physical requirements of team success.
Training prescriptions in soccer should be based on established requirements specific to the
playing positions, thereby ensuring that players are more able to fulfill their game duties and tactical
responsibilities over the soccer match. In further studies it would be important to identify possible
associations that might exist between different parameters of players’ conditioning status and indicators
of real game performance.
Author Contributions: Conceptualization: T.M.; Data curation: T.M. and S.V.; Formal analysis: T.M.; Funding
acquisition: T.M.; Investigation: T.M., S.V. and S.L.; Methodology: D.S.; Project administration: D.S.; Supervision:
D.S.; Validation: S.V.
Funding: This research received no external funding.
Acknowledgments: Authors are particularly grateful to all players who volunteered to participate in the research.
Conflicts of Interest: The authors declare no conflict of interest.
References
1.
Leontijevi´c, B.; Jankovi´c, A.; Tomi´c, L. Attacking performance profile of football teams in different national
leagues according to uefa rankings for club competitions. Facta Univ. Ser. Phys. Educ. Sport 2019, 697–708.
[CrossRef]
2.
Kubayi, A.; Toriola, A. Physical demands analysis of soccer players during the extra-time periods of the
UEFA Euro 2016. South Afr. J. Sports Med. 2018, 30, 1–3. [CrossRef]
3.
Carling, C.; Williams, A.M.; Reilly, T. Handbook of Soccer Match Analysis: A Systematic Approach to Improving
Performance; Routledge: London, UK, 2007.
4.
Ehrmann, F.E.; Duncan, C.S.; Sindhusake, D.; Franzsen, W.N.; Greene, D.A. GPS and injury prevention in
professional soccer. J. Strength Cond. Res. 2016, 30, 360–367. [CrossRef] [PubMed]
5.
Sanchez, M.; Ambros, A.; Salmon, M.; Bhogadi, S.; Wilson, R.T.; Kinra, S.; Marshall, J.D.; Tonne, C. Predictors
of Daily Mobility of Adults in Peri-Urban South India. Int. J. Environ. Res. Public Health 2017, 14. [CrossRef]
[PubMed]
6.
Tandon, P.S.; Saelens, B.E.; Zhou, C.; Christakis, D.A. A Comparison of Preschoolers' Physical Activity
Indoors versus Outdoors at Child Care. Int. J. Environ. Res. Public Health 2018, 15. [CrossRef] [PubMed]
7.
Mennis, J.; Mason, M.; Coffman, D.L.; Henry, K. Geographic Imputation of Missing Activity Space Data
from Ecological Momentary Assessment (EMA) GPS Positions. Int. J. Environ. Res. Public Health 2018, 15.
[CrossRef]
8.
Canton, A.; Torrents, C.; Ric, A.; Goncalves, B.; Sampaio, J.; Hristovski, R. Effects of Temporary Numerical
Imbalances on Collective Exploratory Behavior of Young and Professional Football Players. Front. Psychol.
2019, 10, 1968. [CrossRef]
9.
Sampaio, J.; Macas, V. Measuring tactical behaviour in football. Int. J. Sports Med. 2012, 33, 395–401.
[CrossRef]
10.
Mallo, J.; Mena, E.; Nevado, F.; Paredes, V. Physical demands of top-class soccer friendly matches in relation
to a playing position using global positioning system technology. J. Hum. Kinet. 2015, 47, 179–188. [CrossRef]
11.
Aquino, R.; Carling, C.; Palucci Vieira, L.; Martins, G.; Jabor, G.; Machado, J.; Puggina, E. Influence of
situational variables, team formation, and playing position on match running performance and social
network analysis in brazilian professional soccer players. J. Strength Cond. Res. 2018. [Epub ahead of print].
[CrossRef]
12.
Tierney, P.J.; Young, A.; Clarke, N.D.; Duncan, M.J. Match play demands of 11 versus 11 professional football
using Global Positioning System tracking: Variations across common playing formations. Hum. Mov. Sci.
2016, 49, 1–8. [CrossRef] [PubMed]
13.
Barros, R.M.; Misuta, M.S.; Menezes, R.P.; Figueroa, P.J.; Moura, F.A.; Cunha, S.A.; Anido, R.; Leite, N.J.
Analysis of the distances covered by first division Brazilian soccer players obtained with an automatic
tracking method. J. Sports Sci. Med. 2007, 6, 233–242. [PubMed]
14.
Di Salvo, V.; Baron, R.; Tschan, H.; Montero, F.C.; Bachl, N.; Pigozzi, F. Performance characteristics according
to playing position in elite soccer. Int. J. Sports Med. 2007, 28, 222–227. [CrossRef] [PubMed]
Int. J. Environ. Res. Public Health 2019, 16, 4032
12 of 13
15.
Gregson, W.; Drust, B.; Atkinson, G.; Salvo, V. Match-to-match variability of high-speed activities in premier
league soccer. Int. J. Sports Med. 2010, 31, 237–242. [CrossRef] [PubMed]
16.
Vigne, G.; Gaudino, C.; Rogowski, I.; Alloatti, G.; Hautier, C. Activity profile in elite Italian soccer team. Int. J.
Sports Med. 2010, 31, 304–310. [CrossRef] [PubMed]
17.
Carling, C. Influence of opposition team formation on physical and skill-related performance in a professional
soccer team. Eur. J. Sport Sci. 2011, 11, 155–164. [CrossRef]
18.
Lago-Peñas, C.; Lago-Ballesteros, J. Game location and team quality effects on performance profiles in
professional soccer. J. Sports Sci. Med. 2011, 10, 465–471.
19.
Sarmento, H.; Marcelino, R.; Anguera, M.T.; CampaniÇo, J.; Matos, N.; LeitÃo, J.C. Match analysis in football:
A systematic review. J. Sports Sci. 2014, 32, 1831–1843. [CrossRef]
20.
Dellal, A.; Chamari, K.; Wong, d.P.; Ahmaidi, S.; Keller, D.; Barros, R.; Bisciotti, G.N.; Carling, C. Comparison
of physical and technical performance in European soccer match-play: FA Premier League and La Liga.
Eur. J. Sport Sci. 2011, 11, 51–59. [CrossRef]
21.
Carling, C.; Bloomfield, J.; NELSON, L.; Reilly, T. The role of motion analysis in elite soccer: Contemporary
performance measurement techniques and work rate data. Sports Med. 2012, 38, 389. [CrossRef]
22.
Drust, B.; Atkinson, G.; Reilly, T. Future perspectives in the evaluation of the physiological demands of
soccer. Sports Med. 2007, 37, 783–805. [CrossRef] [PubMed]
23.
Lago-Peñas, C. The role of situational variables in analysing physical performance in soccer. J. Hum. Kinet.
2012, 35, 89–95. [CrossRef] [PubMed]
24.
Moreno, E.; Gómez, M.A.; Lago, C.; Sampaio, J. Effects of starting quarter score, game location, and quality
of opposition in quarter score in elite women’s basketball. Kinesiology 2013, 45, 48–54.
25.
Mackenzie, R.; Cushion, C. Performance analysis in football: A critical review and implications for future
research. J. Sports Sci. 2013, 31, 639–676. [CrossRef] [PubMed]
26.
Taylor, B.J.; Mellalieu, D.S.; James, N.; Barter, P. Situation variable effects and tactical performance in
professional association football. Int. J. Perform. Anal. Sport 2010, 10, 255–269. [CrossRef]
27.
Taylor, J.B.; Mellalieu, S.D.; James, N.; Shearer, D.A. The influence of match location, quality of opposition,
and match status on technical performance in professional association football. J. Sports Sci. 2008, 26, 885–895.
[CrossRef] [PubMed]
28.
Liu, H.; Yi, Q.; Giménez, J.-V.; Gómez, M.-A.; Lago-Peñas, C. Performance profiles of football teams in the
UEFA Champions League considering situational efficiency. Int. J. Perform. Anal. Sport 2015, 15, 371–390.
[CrossRef]
29.
Castellano, J.; Casamichana, D.; Lago, C. The use of match statistics that discriminate between successful
and unsuccessful soccer teams. J. Hum. Kinet. 2012, 31, 137–147. [CrossRef]
30.
Lago-Peñas, C.; Lago-Ballesteros, J.; Dellal, A.; Gómez, M. Game-related statistics that discriminated winning,
drawing and losing teams from the Spanish soccer league. J. Sports Sci. Med. 2010, 9, 288–293.
31.
Johnston, R.J.; Watsford, M.L.; Kelly, S.J.; Pine, M.J.; Spurrs, R.W. Validity and interunit reliability of 10 Hz
and 15 Hz GPS units for assessing athlete movement demands. J. Strength Cond. Res. 2014, 28, 1649–1655.
[CrossRef]
32.
Castellano, J.; Casamichana, D.; Calleja-Gonzalez, J.; Roman, J.S.; Ostojic, S.M. Reliability and Accuracy of
10 Hz GPS Devices for Short-Distance Exercise. J. Sports Sci. Med. 2011, 10, 233–234. [PubMed]
33.
Cohen, J. Statistical Power Analysis for the Behavioral Sciences; Lawrence Erlbaum Assosiates: New York, NY,
USA, 1988.
34.
Lago, C. The influence of match location, quality of opposition, and match status on possession strategies in
professional association football. J. Sports Sci. 2009, 27, 1463–1469. [CrossRef]
35.
Lago, C.; Martín, R. Determinants of possession of the ball in soccer. J. Sports Sci. 2007, 25, 969–974. [CrossRef]
[PubMed]
36.
Lago-Peñas, C.; Lago-Ballesteros, J.; Rey, E. Differences in performance indicators between winning and
losing teams in the UEFA Champions League. J. Hum. Kinet. 2011, 27, 135–146. [CrossRef]
37.
Vaeyens, R.; Lenoir, M.; Williams, A.M.; Philippaerts, R.M. Talent identification and development programmes
in sport. Sports Med. 2008, 38, 703–714. [CrossRef] [PubMed]
38.
Yi, Q.; Jia, H.; Liu, H.; Gómez, M.Á. Technical demands of different playing positions in the UEFA Champions
League. Int. J. Perform. Anal. Sport 2018, 18, 926–937. [CrossRef]
Int. J. Environ. Res. Public Health 2019, 16, 4032
13 of 13
39.
Gaddis, M.L. Statistical methodology: IV. Analysis of variance, analysis of co variance, and multivariate
analysis of variance. Acad. Emerg. Med. 1998, 5, 258–265. [CrossRef]
40.
Bradley, P.S.; Sheldon, W.; Wooster, B.; Olsen, P.; Boanas, P.; Krustrup, P. High-intensity running in English
FA Premier League soccer matches. J. Sports Sci. 2009, 27, 159–168. [CrossRef]
41.
Rampinini, E.; Sassi, A.; Azzalin, A.; Castagna, C.; Menaspa, P.; Carlomagno, D.; Impellizzeri, F.M.
Physiological determinants of Yo-Yo intermittent recovery tests in male soccer players. Eur. J. Appl. Physiol.
2010, 108, 401. [CrossRef]
42.
Andrzejewski, M.; Chmura, J.; Pluta, B.; Konarski, J.M. Sprinting activities and distance covered by top level
Europa league soccer players. Int. J. Sports Sci. Coach. 2015, 10, 39–50. [CrossRef]
43.
Di Salvo, V.; Gregson, W.; Atkinson, G.; Tordoff, P.; Drust, B. Analysis of high intensity activity in Premier
League soccer. Int. J. Sports Med. 2009, 30, 205–212. [CrossRef] [PubMed]
44.
Mohr, M.; Krustrup, P.; Bangsbo, J. Match performance of high-standard soccer players with special reference
to development of fatigue. J. Sports Sci. 2003, 21, 519–528. [CrossRef] [PubMed]
45.
Van Lingen, B. Coaching Soccer: The Official Coaching Book of the Dutch Soccer Association; Reedswain: Spring
City, PA, USA, 1998.
46.
Pic, M.; Lavega-Burgués, P.; March-Llanes, J. Motor behaviour through traditional games. Educ. Stud. 2018,
45, 1–14. [CrossRef]
47.
Pic, M. Performance and Home Advantage in Handball. J. Hum. Kinet. 2018, 63, 61–71. [CrossRef]
© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access
article distributed under the terms and conditions of the Creative Commons Attribution
(CC BY) license (http://creativecommons.org/licenses/by/4.0/).
| Analysis of the Association between Running Performance and Game Performance Indicators in Professional Soccer Players. | 10-21-2019 | Modric, Toni,Versic, Sime,Sekulic, Damir,Liposek, Silvester | eng |
PMC5803184 | ORIGINAL RESEARCH
Effect of speed endurance training and reduced training
volume on running economy and single muscle fiber
adaptations in trained runners
Casper Skovgaard1,2, Danny Christiansen3, Peter M. Christensen1,2, Nicki W. Almquist1,
Martin Thomassen1 & Jens Bangsbo1
1 Department of Nutrition, Exercise and Sports, Section of Integrative Physiology, University of Copenhagen, Copenhagen, Denmark
2 Team Danmark (Danish Elite Sports Organization), Copenhagen, Denmark
3 Institute of Sport, Exercise and Active Living (ISEAL), Victoria University, Melbourne, Australia
Keywords
Intense training, muscle fiber type-specific
adaptations, muscular adaptations, sprint
interval training.
Correspondence
Jens Bangsbo, University of Copenhagen,
Department of Nutrition, Exercise and Sports,
Section of Integrative Physiology, August
Krogh Building, Universitetsparken 13, 2100
Copenhagen O, Denmark.
Tel: +45 35 32 16 23
Fax: +45 35 32 16 00
E-mail: [email protected]
Funding Information
The study was supported by a grant from
Team Danmark (Danish elite sports
organization), Copenhagen, Denmark.
Received: 26 October 2017; Revised: 7
January 2018; Accepted: 9 January 2018
doi: 10.14814/phy2.13601
Physiol Rep, 6 (3), 2018, e13601,
https://doi.org/10.14814/phy2.13601
Abstract
The aim of the present study was to examine whether improved running
economy with a period of speed endurance training and reduced training vol-
ume could be related to adaptations in specific muscle fibers. Twenty trained
male (n = 14) and female (n = 6) runners (maximum oxygen consumption
(VO2-max): 56.4 4.6 mL/min/kg) completed a 40-day intervention with 10
sessions of speed endurance training (5–10 9 30-sec maximal running) and a
reduced (36%) volume of training. Before and after the intervention, a muscle
biopsy was obtained at rest, and an incremental running test to exhaustion
was performed. In addition, running at 60% vVO2-max, and a 10-km run was
performed in a normal and a muscle slow twitch (ST) glycogen-depleted con-
dition. After compared to before the intervention, expression of mitochondrial
uncoupling protein 3 (UCP3) was lower (P < 0.05) and dystrophin was higher
(P < 0.05) in ST muscle fibers, and sarcoplasmic reticulum calcium ATPase 1
(SERCA1) was lower (P < 0.05) in fast twitch muscle fibers. Running econ-
omy at 60% vVO2-max (11.6 0.2 km/h) and at v10-km (13.7 0.3 km/h)
was ~2% better (P < 0.05) after the intervention in the normal condition, but
unchanged in the ST glycogen-depleted condition. Ten kilometer performance
was improved (P < 0.01) by 3.2% (43.7 1.0 vs. 45.2 1.2 min) and 3.9%
(45.8 1.2 vs. 47.7 1.3 min) in the normal and the ST glycogen-depleted
condition, respectively. VO2-max was the same, but vVO2-max was 2.0%
higher (P < 0.05; 19.3 0.3 vs. 18.9 0.3 km/h) after than before the inter-
vention. Thus, improved running economy with intense training may be
related to changes in expression of proteins linked to energy consuming pro-
cesses in primarily ST muscle fibers.
Introduction
Speed endurance training (SET; 10–40 sec repeated “all-
out” efforts with rest periods lasting >5 times the exercise
bouts) with a concomitant reduced training volume has
been found to improve endurance performance in associ-
ation with better running economy at submaximal speeds
in trained runners (Bangsbo et al. 2009; Bangsbo 2015).
However, the mechanisms causing the improved running
economy are not clearly identified, but may be related to
metabolic changes in the trained muscles (Saunders et al.
2004).
Training-induced improvement in running economy
may be due to higher mitochondrial efficiency, that is,
higher ATP/O2, which could be due to reduced uncou-
pled respiration. The mitochondrial uncoupling protein 3
ª 2018 The Authors. Physiological Reports published by Wiley Periodicals, Inc. on behalf of
The Physiological Society and the American Physiological Society
This is an open access article under the terms of the Creative Commons Attribution License,
which permits use, distribution and reproduction in any medium, provided the original work is properly cited.
2018 | Vol. 6 | Iss. 3 | e13601
Page 1
Physiological Reports ISSN 2051-817X
(UCP3) is suggested to be involved in thermogenesis by
dispersing energy as heat instead of converting it to ATP
(Gong et al. 1997; Boss et al. 2000) and improved run-
ning economy may therefore be related to reduced levels
of muscle UCP3. In agreement, cross-sectional studies
have shown that endurance-trained subjects have lower
muscle UCP3 expression and better running economy
than untrained subjects (Russell et al. 2003a,b; Mogensen
et al. 2006). However, Iaia et al. (2009) found no change
in whole muscle UCP3 level although running economy
improved after 4 weeks of SET and a 65% reduced train-
ing volume. Thus, studies should investigate whether
changes in the single muscle fiber expression of UCP3
could be related to changes in running economy.
The transfer of muscle force produced by the acto-
myosins involves a secondary matrix of proteins that trans-
mit the muscle force along and between muscle fibers and
out to the extracellular matrix. Cytoskeleton proteins, such
as dystrophin, have been identified as playing a role in this
muscle force transmission (Rybakova et al. 2000; Prins
et al. 2009) and changes in the expression of these proteins
could influence the integrity and the strength of the muscle
(Hughes et al. 2015). Hence, increased expression of mus-
cle dystrophin may result in increased rate of force devel-
opment, increased muscular power output and greater
storage and return of elastic energy thereby lowering the
cost of running (i.e., improve running economy).
Another potential cause of training-induced improve-
ments in running economy is lowered muscle expression
of the sarcoplasmic reticulum (SR) Ca2+-ATPase (SERCA)
pumps, as they are suggested to be responsible for up to
50% of the ATP used during muscle activity (Clausen
et al. 1991; Walsh et al. 2006; Smith et al. 2013). Studies
have shown that speed endurance training modulates
skeletal muscle fiber type distribution in soccer players
(Gunnarsson et al. 2012) and runners (Skovgaard et al.
2014), which has been found together with lowered
SERCA1 expression (Skovgaard et al. 2014) and improved
running economy. Muscle fibers with high SERCA1
expression have a faster release and uptake of Ca2+ (Del-
bono and Meissner 1996; Froemming et al. 2000) and
lowered expression of SERCA1 may therefore reduce the
energy turnover during exercise.
An increase in the respiratory capacity of skeletal mus-
cle permits the use of less oxygen per mitochondrial res-
piratory chain for a given submaximal running speed
(Saunders et al. 2004). Slow twitch (ST) muscle fibers
have higher mitochondrial content and are more depen-
dent on oxidative metabolism than fast twitch (FT) mus-
cle fibers (Berchtold et al. 2000; Schiaffino and Reggiani
2011). However, Jansson and Kaijser (1977) reported that,
unlike a control group of varying physical fitness, there
was no difference in succinate dehydrogenase muscle
activity between ST and FT fibers in gastrocnemius mus-
cle of elite orienteers, suggesting that FT fibers have the
ability to metabolically adapt to high oxidative demands
(Jansson and Kaijser 1977). Metabolic adaptations in FT
fibers may therefore contribute to improving running
economy after intense training, such as SET, targeting
both ST and FT fibers (Egan and Zierath 2013). In sup-
port, augmented mRNA response related to mitochon-
drial
biogenesis
(peroxisome
proliferator-activated
receptor-c coactivator-1, PGC-1a) and metabolism (hex-
okinase II and pyruvate dehydrogenase kinase-4, PDK4)
in trained subjects was observed following a SET session
(Skovgaard et al. 2016). Furthermore, PGC-1a mRNA has
been shown to increase in an exercise intensity-dependent
manner (Egan et al. 2010; Nordsborg et al. 2010). Regular
intense training may therefore lead to higher oxidative
capacity, possibly due to oxidative adaptations in FT
fibers, which in turn could contribute to the improved
running economy as a result of the intense training (Iaia
et al. 2008; Bangsbo et al. 2009; Iaia and Bangsbo 2010;
Skovgaard et al. 2014).
In vitro studies have shown that the energy cost of con-
traction is higher in FT than ST fibers (Crow and Kushmer-
ick 1982; Barclay et al. 1993; He et al. 2000). This was
confirmed in vivo by Krustrup et al. (2008) who observed
that the oxygen uptake for at given exercise intensity was
higher when ST fibers were blocked by a neuromuscular
blocking agent. And reports by Krustrup et al. (2004), who
depleted the ST fibers the day before submaximal exercise,
that the glycogen depletion of ST fibers enhanced the
recruitment of FT fibers and elevated the energy require-
ment by 7% (Krustrup et al. 2004). By using the approach,
of depleting ST fibers the day before exercise (Krustrup
et al. 2004), before and after a SET period, it may be possi-
ble to study whether a change in running economy is
caused by specific adaptations in FT fibers.
Thus, the aims of the present study were in trained run-
ners to investigate the effect of intensified training, in the
form of speed endurance training and a reduced volume of
aerobic training, on running economy and adaptation of
single muscle fibers. We hypothesized that FT muscle fibers
would adapt to the training by lowered expression of
UCP3 and SERCA1, and increased expression of dys-
trophin and CS, which would be associated with improved
running economy and 10-km running performance.
Methods
Subjects
Twenty-six trained runners commenced the study. Six sub-
jects did not complete the intervention period due to per-
sonal circumstances (n = 4) or low adherence to the
2018 | Vol. 6 | Iss. 3 | e13601
Page 2
ª 2018 The Authors. Physiological Reports published by Wiley Periodicals, Inc. on behalf of
The Physiological Society and the American Physiological Society
Muscle Fiber Type Adaptations to Speed Endurance Training in Runners
C. Skovgaard et al.
training program (n = 2). Thus, a total of twenty trained
male (n = 14) and female (n = 6) runners with an average
age, height, body mass, and maximum oxygen consump-
tion
(VO2-max)
of
28.1 4.5 years,
177.5 9.9 cm,
72.5 10.6 kg, and 56.4 4.6 mL/min/kg, respectively,
(males: 28.8 4.8 years, 181.8 7.9 cm, 77.8 6.6 kg,
58.1 3.4 mL/min/kg;
females:
27.4 3.7 years,
169.0 5.6 cm,
59.9 6.9 kg,
52.5 4.9 mL/min/kg;
means SD), completed the study. After receiving written
and oral information about the study and the possible risks
and discomforts associated with the experimental proce-
dures, all subjects gave their written informed consent to
participate. The study conformed to the Code of Ethics of
the World Medical Association (Declaration of Helsinki)
and was approved by the Ethics Committee of the capital
region of Copenhagen (Region Hovedstaden).
Design
The study lasted 40 days and consisted of 10 sessions of
supervised speed endurance training (SET) and 10 ses-
sions
of
aerobic
moderate-intensity
(AM)
training
(Fig. 1). Total running distance during the intervention
period was reduced (P < 0.05) by 36% compared to
before
the
intervention
(mean SE,
16 1
vs.
25 2 km/week).
Screening and familiarization
Before being included in the study, subjects performed a
10-km running test and an incremental treadmill test to
exhaustion with pulmonary VO2 measured by a breath-
by-breath gas analyzing system (Oxycon Pro; Viasys
Healthcare, Hoechberg, Germany), and heart rate (Polar
Team2 transmitter; Polar Electro Oy, Kempele, Finland)
collected throughout the test.
Training
SET was performed on day two and six of an 8-day cycle
at Østerbro Stadium, Copenhagen, on an outdoor 400-m
running track. In first and final SET session, subjects
completed six bouts of 30-sec running. The first bout was
performed
with
near-maximal
intensity,
whereas
the
remaining five bouts were performed with maximal inten-
sity and distance covered was measured. For the remain-
ing eight SET sessions, subjects completed ten bouts of
30-sec “all-out” running. In all sessions, running bouts
were separated by 3.5 min of recovery (walking ~200 m
to the start-line). SET sessions were supervised, but the
subjects performed the SET sessions on their own, if they
were unable to participate in the supervised training
(85 4% adherence to the supervised SET).
AM training was performed on the first and fifth day
during the 8-day cycle. These sessions were not super-
vised, but subjects kept a training log to record exercise
distance, time and intensity. A Polar FT7 (Polar Electro
Oy, Kempele, Finland) or personal watch with HR moni-
tor was used to record exercise intensity and training logs
was continuously analyzed. The adherence to the AM
training sessions was 93 3% with a weekly duration of
68 5 min and with an average heart rate of 83 1%
of HRmax.
Testing
Tests were performed on separate days interspersed by at
least 48 hours, on the same treadmill in the Exercise
Physiology laboratory at August Krogh Institute, Depart-
ment of Nutrition, Exercise and Sports, University of
Copenhagen, before and after the intervention. Tests
included: (1) an incremental running test to exhaustion
(INC); (2) repeated bouts of 6-min submaximal running
followed by a 10-km running test on a running track in a
normal condition; (3) repeated bouts of 6-min submaxi-
mal running followed by a 10-km running test on a run-
ning track in a ST glycogen-depleted condition; (4) a
muscle biopsy and a blood sample collected at rest after
an overnight fast (Fig. 1).
All tests were carried out at the same time of day. Sub-
jects refrained from strenuous physical activity, alcohol
and caffeine 24 h before testing. Subjects were instructed
to keep a diary journal 2 days before and during the first
series of tests, and to replicate this diet when tested again.
0
8
16
24
32
40
days
Pre testing:
•
INC
•
10-km normal
•
•
10-km depleted
muscle and blood
sampling
•
muscle and blood
sampling
Post testing:
•
INC
•
10-km normal
•
10-km depleted
Figure 1. Testing before (Pre) and after (Post) 5 blocks/40 days of speed endurance training and reduced training volume in trained runners.
Small grey, black and white boxes on the timeline are days with aerobic moderate-intensity training, speed endurance training and rest days,
respectively. INC: incremental test to exhaustion.
ª 2018 The Authors. Physiological Reports published by Wiley Periodicals, Inc. on behalf of
The Physiological Society and the American Physiological Society
2018 | Vol. 6 | Iss. 3 | e13601
Page 3
C. Skovgaard et al.
Muscle Fiber Type Adaptations to Speed Endurance Training in Runners
The incremental running test to exhaustion
INC consisted of 2 min of walking at 5 km/h, 6 min at
the subject’s individual average 10-km running pace
determined at the 10-km screening test before the inter-
vention (v10 km; 13.7 0.3 km/h), and 2 min at 14 or
15 km/h (dependent on v10 km), after which the speed
increased by 1 km/h every minute until exhaustion. Dur-
ing INC, VO2-max, defined as the highest average value
achieved over a 30-sec period (Howley et al. 1995), and
maximal
incremental
speed
(vVO2-max)
{[vVO2-
max = Vf + (Ti/60)],
where
Vf
is
the
final
velocity
obtained and Ti is the time spent at the final speed level}
were determined. Attaining of maximal heart rate (HR)
(judged against the screening test) and an RER value of
>1.15 were used as criterions. During the last part of the
test, the subjects were verbally encouraged to continue
their effort until voluntary termination of the test. Before
the test, body mass was measured and subjects wore a
Polar Team2 HR monitor around their chest for continu-
ous HR recordings. Pulmonary VO2 was measured by use
of Oxycon Pro, which was calibrated prior to each test.
Muscle and blood sampling
Sampling of muscle and blood was performed between 7
and 11 AM after an overnight fast. Using the Bergstr€om
procedure (Bergstrom 1962), a muscle biopsy was col-
lected with a 5-mm needle from a standardized depth of
5 cm in the middle of m. vastus lateralis of the right leg
at rest using local anesthesia (1 mL; 20 mg/L lidocaine
without adrenaline). The muscle sample (~100 mg wet
weight) was immediately frozen in liquid N2 and stored
at 80°C until further analysis. Next, a catheter was
inserted in the antecubital vein, and a ~7-mL blood sam-
ple was collected and stored on ice until being analyzed.
10-km running tests
Both before and after the intervention, two 10-km run-
ning tests were performed on a 400-m outdoor running
track (Østerbro Stadium, Copenhagen) under similar
weather conditions (~20°C, partly cloudy, light winds)
between the beginning of July and end of August. The
10-km tests were conducted in a randomized order either
without (normal) or after a muscle ST glycogen depletion
protocol that was performed the day before the test (see
later). Each 10-km running test was preceded by two
bouts of 6 min of running, separated by 20 min of rest,
on a treadmill at the subject’s individual 60% vVO2-max
(11.6 0.2 km/h) with respiratory and HR measure-
ments. After these bouts, subjects biked to Østerbro Sta-
dium (1-km, slow pace) for the 10-km test.
Muscle slow-twitch glycogen depletion protocol
The protocol was based on the findings from the study
by Krustrup et al. (2004) who used a 3-h cycling protocol
(~50% VO2-max) to deplete ST fibers the day before 20-
min of submaximal cycling. The authors reported that the
glycogen depletion of ST fibers (51 and 44% of the ST
fibers were empty and almost empty of glycogen, respec-
tively, and less than 2% of the FT fibers were empty of
glycogen) enhanced the recruitment of FT fibers (Krus-
trup et al. 2004). The protocol is verified by previous
findings that ST fibers are exclusively active at 50% VO2-
max when subjects have normal muscle glycogen levels
(Gollnick et al. 1974; Vøllestad and Blom 1985).
The subjects completed a 3-h exercise protocol consist-
ing of 60 min of cross-training, 30 min of cycling,
30 min of running, and 60 min of striding at a low speed
to deplete glycogen in ST muscle fibers of the calves and
thigh muscles. The protocol was chosen to minimize
muscle soreness from eccentric contractions while mim-
icking the movement pattern of running. During the pro-
tocol,
subjects’
HR
was
monitored
to
ensure
they
exercised at 60–65% of HRmax (~50% VO2-max). Average
HR during the 3-h depletion protocol was the same
before
and
after
the
intervention
(120 1
vs.
120 1 bpm; 63 0 vs. 63 0% HRmax). The protocol
started at 6:30 PM and finished around 10:00 PM and sub-
jects were allowed water ad libitum. After termination of
the protocol, subjects were given a diet consisting of 5E%
carbohydrate, 35E% protein, and 60E% fat with a total
energy intake of 30 kJ/kg body mass at dinner and 10 kJ/
kg at breakfast. Breakfast was consumed 2 h before the
10-km running test, which started at 8:00 AM.
Whole muscle protein expression
Western blotting was performed to determine protein
expression
as
described
previously
(Skovgaard
et al.
2014). In short, ~2.5 mg dry weight (dw; freeze-dried for
a minimum of 24 h) of each muscle sample was dissected
free from blood, fat, and connective tissue. Samples were
homogenized for 1 min at 28.5 Hz (Qiagen Tissuelyser II;
Retsch) in a fresh batch of ice-cold buffer containing (in
mM) 10% glycerol, 20 Na-pyrophosphate, 150 NaCl, 50
HEPES (pH 7.5), 1% NP-40, 20 b-glycerophosphate, 2
Na3VO4, 10 NaF, 2 PMSF, 1 EDTA (pH 8), 1 EGTA (pH
8), 10 lg/mL aprotinin, 10 lg/mL leupeptin, and 3 ben-
zamidine, after which they rotated for 1 h at 4°C, and
centrifuged at 18,320g for 20 min at 4°C to exclude
nondissolved structures. The supernatant (lysate) was col-
lected and used for further analysis. Total protein concen-
tration in each sample was determined by a BSA standard
kit (Thermo Scientific), and samples were mixed with 69
2018 | Vol. 6 | Iss. 3 | e13601
Page 4
ª 2018 The Authors. Physiological Reports published by Wiley Periodicals, Inc. on behalf of
The Physiological Society and the American Physiological Society
Muscle Fiber Type Adaptations to Speed Endurance Training in Runners
C. Skovgaard et al.
Laemmli buffer (7 mL 0.5 mol/L Tris-base, 3 mL glycerol,
0.93 g DTT, 1 g SDS, and 1.2 mg bromophenol blue)
and ddH2O to reach equal protein concentration before
protein expression was determined by western blotting.
Equal amounts of total protein (6–12 lg depending on
the protein of interest) were loaded in each well of precast
gels (Millipore). All samples from each subject were loaded
on the same gel. Proteins were separated according to their
molecular weight by SDS-PAGE and semi-dry transferred
to a 0.45 lm PVDF membrane (Bio-Rad). The membranes
were blocked in either 2% skimmed milk or 3% BSA in
TBST, including 0.1% Tween-20 before an overnight incu-
bation with rocking in primary antibody at 4°C. The pri-
mary antibodies used were: (ab. cat number and company,
respectively):
sarcoplasmic
reticulum
Ca2+-ATPase
1
(SERCA1; MA3-912; Thermo Scientific), sarcoplasmic
reticulum Ca2+-ATPase 2 (SERCA2; N-19 Sc-8095; Santa
Cruz Technology), actin (A2066; Sigma Aldrich), mito-
chondrial uncoupling protein 3 (UCP3; AB3046; Milli-
pore). The membranes were then incubated for 1 h at
room temperature in horseradish peroxidase conjugated
secondary antibody (rabbit anti-sheep (P-0163, DAKO),
rabbit anti-goat (P-0449, DAKO), goat anti-mouse (P-
0447, DAKO) or goat anti-rabbit IgM/IgG (4010-05; South-
ern Biotech), depending on the primary antibody source).
The protein bands were visualized with ECL (Millipore)
and recorded with a digital camera (ChemiDoc MP Imag-
ing System, Bio-Rad Laboratories). For each muscle sam-
ple, protein expression was determined in duplicate on
individual gels. Quantification of the band intensity was
performed using Image Lab version 4.0 (Bio-Rad Labora-
tories). Each band was normalized to two control samples
of human, whole-muscle homogenate that were loaded
onto every gel.
Single muscle fiber protein expression
To determine the protein expression of citrate synthase
(CS), UCP3 as well as SERCA- and myosin heavy chain
(MHC) isoforms in different muscle fiber types, 88 5
single-fiber segments were collected from each freeze-dried
muscle biopsy. Individual segments were isolated under a
microscope at room temperature using fine jeweler’s for-
ceps, and were individually incubated for 1 h at room tem-
perature
in
microfuge
tubes
containing
10 lL
of
denaturing buffer (0.125 mol/L Tris-HCl, 10% glycerol,
4% SDS, 4 mol/L urea, 10% mercaptoethanol, and 0.001%
bromophenol blue, pH 6.8) (Murphy, 2011). The dena-
tured segments were stored at 80°C until being analyzed
for fiber type and grouped accordingly as described below.
The fiber type of fiber segments was determined using
dot blotting. 1.5 lL of each denatured sample was spotted
onto two PVDF membranes, which were pre-activated in
95%
ethanol
and
pre-equilibrated
in
transfer
buffer
(25 mmol/L Tris, 192 mmol/L glycine, pH 8.3, 20%
methanol). After drying completely at room temperature,
the membranes containing samples were reactivated in
ethanol and re-equilibrated in transfer buffer, before
being blocked in 5% skim milk in TBST for 5–30 min.
One membrane was then incubated by gentle rocking
with MHCI antibody (1:200 in 1% BSA with PBST;
mouse monoclonal IgM, clone A4.840, Developmental
Studies Hybridoma Bank (DSHB)), and the other with
MHCIIa antibody (mouse monoclonal IgG, clone A4.74,
DSHB) for 2 h at room temperature. After a quick wash
in TBST, secondary antibody was applied (1:10,000), and
protein signals quantified as described under Whole mus-
cle protein expression (section above).
The remaining part of each denatured fiber segment
(7 lL) was pooled into groups of ST or FTa fibers depend-
ing on MHC expression. The number of segments entailed
in each pool of fibers per biopsy was 15 2 (range: 8–42)
for ST and 18 2 (range 8–39) for FTa fibers before the
intervention, and 19 3 (range: 7–55) and 18 2 (range
7–41), respectively, after the intervention. Hybrid fibers
(expressing multiple MHC isoforms) were excluded from
analysis. Protein expression was determined in pools of ST
and FTa fibers using western blotting as detailed in the sec-
tion above. The primary antibodies used were: (ab. cat
number and company, respectively): CS (ab96600, Abcam),
UCP3 (AB3046; Millipore) SERCA1 (MA3-912; Thermo
Scientific), SERCA2 (N-19 Sc-8095; Santa Cruz Technol-
ogy). Pools of fibers from biopsies obtained before and
after the intervention was loaded on the same gel (stain-
free, 4–15%, precast), along with either a calibration curve
or two loading controls of whole-muscle homogenate. Pro-
tein bands were quantified by normalizing each band to the
total protein content in each lane on the stain-free gel.
Muscle enzyme activity
Muscle
enzyme
activity
was
determined
by
use
of
~2.5 mg dw muscle tissue dissected free from blood, fat,
and connective tissue, which was homogenized (1:400) in
a 0.3 mol/L phosphate buffer (pH 7.7) by 2 rounds of
30-sec using a TissueLyser II (Retch, Germany). Maximal
activity of CS, b-hydroxyacyl-CoA-dehydrogenase (HAD)
and phosphofructokinase (PFK) was determined fluoro-
metrically with NAD-NADH coupled reactions (Lowry
and Passonneau 1972) on a Fluoroskan Ascent apparatus
(Thermo Scientific) using Ascent Software version 2.6.
Blood analysis
A total of ~7 mL blood was drawn in a heparinized 2-mL
syringe and a 5-mL syringe at rest. A part of the 2-mL blood
ª 2018 The Authors. Physiological Reports published by Wiley Periodicals, Inc. on behalf of
The Physiological Society and the American Physiological Society
2018 | Vol. 6 | Iss. 3 | e13601
Page 5
C. Skovgaard et al.
Muscle Fiber Type Adaptations to Speed Endurance Training in Runners
sample (~1.5 mL) and the 5-mL sample (split into
2 9 2 mL tubes containing 30 lL EDTA) were centrifuged
at 20,000 g for ~2 min and the remaining whole blood from
the 2-mL sample (~0.5 mL) was stored on ice for further
analyses. After centrifugation, the plasma was transferred
into tubes that were placed in ice-cold water until they were
stored at 20°C. Plasma samples were subsequently ana-
lyzed for testosterone and cortisol, creatine kinase (CK) and
immunoglobulin A (IgA). CK activity was analyzed by enzy-
matic kinetic assay methods (Roche Diagnostic, Mannheim,
Germany) using a Hitachi 912 (Roche Diagnostic, Indi-
anapolis). IgA was determined using an immunoturbidi-
metric assay method (Horiba, Montpellier, France) on an
automatic analyzer (Pentra C400, Horiba, Montpellier,
France). Testosterone and cortisol was determined using
ELISA kits (R&D Systems, Inc. Minneapolis). Whole blood
was analyzed for hemoglobin, hematocrit and HCO3 at rest
(ABL800 Flex; Radiometer Medical, Copenhagen, Den-
mark).
Running economy
Running Economy (RE) was calculated using the follow-
ing formula:
REðmLO2=kg=kmÞ ¼ VO2ðmL/minÞ 60 min/h=BM (kg)
running speedðkm/hÞ
where VO2 is the average value during the last 2 min of
running for the two intervals at 60% vVO2-max and v10-
km, and BM is body mass.
Statistics
Paired t tests were used to evaluate the effect of the inter-
vention (Pre vs. Post) with two-way ANOVA repeated mea-
sures being used to evaluate the effect of glycogen condition
(normal vs. ST glycogen-depleted) on 10-km running per-
formance and running economy (at 60% vVO2-max). Level
of significance was set at P < 0.05. A Student-Newman
Keuls post-hoc test was applied in case significance was
reached in the ANOVA. Absolute data values was used and
presented as means SE unless otherwise stated.
Results
Pulmonary oxygen uptake and heart rate
during submaximal exercise
Pulmonary VO2 during running at v10-km was 1.9%
lower (P < 0.05) after compared to before the interven-
tion (3.46 0.14 vs. 3.53 0.14 L/min), and running
economy was improved by 2.1% (P < 0.05; 207.6 2.6
vs. 212.1 2.8 mL/kg/km) (Fig. 2). Mean HR at v10-km
was 1.7% lower (P < 0.05) after than before the interven-
tion (162 2 vs. 165 2 bpm).
In the normal condition, pulmonary VO2 at 60%
vVO2-max was the same before and after the intervention
(3.01 0.13 vs. 2.99 0.13 L/min), whereas running
economy was 1.7% better (P < 0.05) after compared to
before the intervention (210.4 2.9 vs. 214.1 3.2 mL/
kg/km) (Fig. 3). In the ST glycogen-depleted condition,
pulmonary VO2 at 60% vVO2-max (3.05 0.15 (Post)
vs. 3.04 0.13 (Pre) L/min) and running economy
(216.5 2.9 (Post) vs. 217.4 2.9 (Pre) mL/kg/km) did
not change with the intervention (Fig. 3).
Before the intervention, pulmonary VO2 at 60% vVO2-
max was the same in normal and ST glycogen-depleted
condition, whereas after the intervention, pulmonary VO2
was 2.0% lower (P < 0.01) in normal than ST glycogen-
depleted condition. Before and after the intervention,
running economy was 1.6% and 2.9% better (P < 0.05),
respectively, in the normal compared to the ST glycogen-
depleted condition (Fig. 3).
HR during running at 60% vVO2-max in normal and
ST glycogen-depleted condition did not change with the
intervention, and there were no differences between
conditions.
Expression of proteins in muscle
homogenate
Expression of SERCA2 in muscle homogenate was 20%
higher (P < 0.05) after compared to before the interven-
tion, whereas expression of muscle SERCA1 was 22%
lower (P < 0.05). Expression of muscle actin and UCP3
did not change with the intervention (Fig. 4).
Expression of proteins in single muscle
fibers
After compared to before the intervention, expression of
muscle CS and UCP3 in ST fibers was 22% and 25%,
respectively, lower (P < 0.05), and expression of muscle
dystrophin in ST fibers was 41% higher (P < 0.05)
(Fig. 5). Expression of muscle SERCA1 was 19% lower
(P < 0.05) in FTa fibers, and expression of MHCIIa was
19% higher (P < 0.05) in FTa fibers after than before the
intervention. Expression of SERCA2 and MHCI in the
single fiber pools was unchanged with the intervention
(Fig. 5).
Muscle enzymatic activity
Maximal activity of CS, HAD, and PFK was 10.7%, 9.1%,
and 23.4%, respectively, higher (P < 0.05) after than
before the intervention (Table 1).
2018 | Vol. 6 | Iss. 3 | e13601
Page 6
ª 2018 The Authors. Physiological Reports published by Wiley Periodicals, Inc. on behalf of
The Physiological Society and the American Physiological Society
Muscle Fiber Type Adaptations to Speed Endurance Training in Runners
C. Skovgaard et al.
10-km run
Compared to before, 10-km performance in the normal
condition improved (P < 0.01) by 3.2% (43.7 1.0 vs.
45.2 1.2 min) after the intervention (Fig. 6). In the ST
glycogen-depleted
condition,
10-km
performance
was
3.9% better (P < 0.001) after compared to before the
intervention (45.8 1.2 vs. 47.7 1.3 min; Fig. 6). Ten
kilometer performance was reduced (P < 0.001) to the
same degree in the ST glycogen-depleted compared to
the normal condition before (5.3%) and after (4.7%) the
intervention (Fig. 6).
Maximum oxygen uptake, body mass, and
heart rate
VO2-max was the same before and after the intervention
(4.06 0.16
vs.
4.13 0.18 L/min;
56.4 1.0
vs.
56.3 1.2 mL/min/kg), but vVO2-max was 2.0% higher
(P < 0.05) after compared to before (19.3 0.3 vs.
18.9 0.3 km/h). Peak heart rate during INC was the
same before and after the intervention (187 2 vs.
188 2 bpm) as well as body mass (72.5 2.4 vs.
72.9 2.3).
Blood variables
Blood hematocrit and concentration of hemoglobin as
well as plasma concentrations of testosterone, cortisol, CK
and HCO3
were the same before and after the interven-
tion. Compared to before the intervention, testosterone to
cortisol ratio was 31.3% higher (P < 0.05) and plasma
IgA level was 4.0% higher (P < 0.05) after (Table 2).
Discussion
The main findings of the present study were that a period
of intense and reduced volume of training in trained
202
204
206
208
210
212
214
216
Pre
Post
Running economy (ml/kg/km) at v10-km
B
***
3350
3400
3450
3500
3550
3600
3650
3700
Pre
Post
Oxygen uptake (L/min) at v10-km
A
***
Figure 2. Oxygen uptake (A) and running economy (B) at v10-km before (Pre) and after (Post) 5 blocks/40 days of speed endurance training
and reduced training volume in trained runners. Values are means SE. ***Post different (P < 0.001) to Pre.
204
206
208
210
212
214
216
218
220
222
Pre
Post
Running Economy (ml/kg/km)
at 60% VO2-max
Depleted
Normal
B
#
#
*
2850
2900
2950
3000
3050
3100
3150
3200
3250
Pre
Post
Oxygen uptake (L/min)
at 60% VO2-max
Depleted
Normal
A
#
Figure 3. Oxygen uptake (A) and running economy (B) at 60% VO2-max in depleted and normal conditions before (Pre) and after (Post) 5
blocks/40 days of speed endurance training and reduced training volume in trained runners. Values are means SE. *Post different (P < 0.05)
to Pre; #difference (P < 0.05) within time-point.
ª 2018 The Authors. Physiological Reports published by Wiley Periodicals, Inc. on behalf of
The Physiological Society and the American Physiological Society
2018 | Vol. 6 | Iss. 3 | e13601
Page 7
C. Skovgaard et al.
Muscle Fiber Type Adaptations to Speed Endurance Training in Runners
runners improved running economy together with higher
expression of dystrophin and lowered expression of UCP3
in ST muscle fibers as well as lower expression of
SERCA1 in FTa muscle fibers. In addition, compared to
the normal condition, 10-km running performance and
running economy was equally reduced after the ST mus-
cle glycogen-depletion protocol before and after the inter-
vention period.
The better running economy at 60% vVO2-max and
v10-km after the intervention period is in accordance
with findings in other studies of intense training and low-
ered training volume in trained runners (Bangsbo et al.
2009; Iaia and Bangsbo 2010; Skovgaard et al. 2014). In
the
ST glycogen-depleted
condition,
where
a higher
recruitment of FT fibers would be expected, the running
economy remained unchanged with training, suggesting
that it was mainly changes in ST fibers that caused the
improvement in running economy in the normal condi-
tion. In accordance, the expression of UCP3 in ST fibers
was lowered by training in the present study. As mechani-
cal energy efficiency is negatively related to UCP3 expres-
sion (Russell et al. 2003a,b; Mogensen et al. 2006), this
suggests that the reduced UCP3 expression in ST fibers
may have improved the mitochondrial efficiency, and
thereby running economy. On the other hand, reduced
energy
expenditure
during
submaximal
exercise
was
–50%
–40%
–30%
–20%
–10%
0%
10%
20%
30%
40%
50%
60%
Actin
UCP3
SERCA1
SERCA2
Muscle expression (Post relative to Pre)
*
*
Figure 4. Protein expression of actin; UCP3, mitochondrial
uncoupling protein 3; SERCA1 and 2, sarcoplasmic reticulum
calcium ATPase, before (Pre) and after (Post) 5 blocks/40 days of
speed endurance training and reduced training volume in trained
runners. Values are geometric means 95% confidence interval
(CI) (Post relative to Pre). *Post different (P < 0.05) from Pre.
40%
60%
80%
100%
120%
140%
160%
180%
MHCIIa
SERCA1
CS
UCP3
Dystrophin
FTa single muscle fiber expression (relative to Pre)
Post FTa
*
*
40%
60%
80%
100%
120%
140%
160%
180%
MHCI
SERCA2
CS
UCP3
Dystrophin
ST single muscle fiber expression (relative to Pre)
Post ST
*
*
*
A
B
Figure 5. ST (A) and FTa (B) single muscle fiber expression of MHCI and II, myosin heavy chain; SERCA1 and 2, sarcoplasmic reticulum calcium
ATPase; CS, citrate synthase; UCP3, mitochondrial uncoupling protein 3; and dystrophin, before (Pre) and after (Post) 5 blocks/40 days of speed
endurance training and reduced training volume in trained runners. Values are means SE (relative to Pre). *Post different (P < 0.05) from Pre.
Table 1. Maximal
activity
of
muscle
citrate
synthase
(CS),
b-hydroxyacyl-CoA-dehydrogenase
(HAD);
phosphofructokinase
(PFK) at rest before (Pre) and after (Post) 5 blocks/40 days of
speed endurance training and reduced training volume in trained
runners.
Pre
Post
CS (lmolg/dw/min)
17.7 2.9
19.6 2.9*
HAD (lmolg/dw/min)
15.6 0.9
17.0 0.7*
PFK (lmolg/dw/min)
72.1 15.3
88.9 13.7*
Data are presented as means SE.
*Post different (P < 0.05) to Pre.
2018 | Vol. 6 | Iss. 3 | e13601
Page 8
ª 2018 The Authors. Physiological Reports published by Wiley Periodicals, Inc. on behalf of
The Physiological Society and the American Physiological Society
Muscle Fiber Type Adaptations to Speed Endurance Training in Runners
C. Skovgaard et al.
reported in a study where trained subjects (VO2-max:
56 1 mL/min/kg) performed 4 weeks of speed endur-
ance training (8–12 9 30-sec at maximum
speed; 3
times/week) with a 65% reduced training volume without
change in the expression of whole muscle UCP3 (Iaia
et al. 2009). It could be speculated that the reduction in
training volume was too large to elicit changes in UCP3
expression or that a potential reduced expression of
UCP3 in ST fibers was undetected by the analysis of
whole muscle tissue (Iaia et al. 2009).
In addition, the expression of dystrophin, a protein
that connects the sarcomere and the extracellular matrix
(Hughes et al. 2015), increased in ST muscle fibers with
the training intervention. An important function of dys-
trophin is to transmit forces generated by the actin-myo-
sin cross-bridge (Chopard et al. 2005), and the higher
expression of dystrophin in ST fibers may have enhanced
the structural integrity of the ST muscle fibers and
thereby influenced running economy.
The intervention period also led to lowered expression
of SERCA1 in FT muscle fibers, which has also been
found in studies of endurance training (Majerczak et al.
2008, 2012; Green et al. 2011). The lower expression of
muscle SERCA1 may have reduced the energy turnover
during exercise, since calcium handling by the ATP
dependent SERCA pumps is reported to be responsible
for up to 50% of total energy usage (Clausen et al. 1991;
Walsh et al. 2006; Smith et al. 2013), and, thus, may have
contributed to the better running economy after the
intervention period.
The finding of improved 10-km running performance
after the intervention period is in agreement with other
studies investigating the effect of intense training and low-
ered training volume in trained runners (Bangsbo et al.
2009; Iaia and Bangsbo 2010; Skovgaard et al. 2014). The
novel finding in the present study was that the magnitude of
the difference between 10-km running in normal versus ST
fiber glycogen-depleted condition was the same before and
after the training period. This observation suggests that any
effect of the intervention on the oxidative capacity of the FT
fibers was small, which is supported by the finding that the
expression of CS in the FT fibers did not change with the
intervention. In agreement, a 7-week intense training period
(12 9 30-sec maximal sprints 2.5 times/week and 5 9 4-
min intervals (at a heart rate (HR) of 89% HRmax) 1.5
times/week) with a 50% reduction in training volume, did
not change expression of muscle CS and COX-4 in segments
of FT fibers in well trained cyclists (VO2-max: 59 4 mL/
min/kg) (Christensen et al. 2015). Collectively, these find-
ings suggest that intense training with a decrease (36–50%)
in training volume does not affect oxidative proteins in FT
muscle fibers in trained subjects. Nevertheless, the mixed
muscle CS activity was elevated with the intervention and
may have contributed to the better 10-km performance.
In agreement with other studies on the effect of speed
endurance training and reduced training volume in run-
ners (Bickham et al. 2006; Iaia et al. 2008; Bangsbo et al.
2009; Iaia and Bangsbo 2010; Skovgaard et al. 2014),
VO2-max did not change with the intervention and can-
not explain the improved 10-km performance. Based on
the performance during the 10-km run, VO2-max and
running economy, the fraction of FVO2-max [FVO2-
max = 10-km velocity (km/hr)*running economy at v10-
km (mL/kg/km)/VO2-max (mL/min/kg)100] during the
10-km run was calculated. It showed that FVO2-max did
not
change
with
the
intervention
period
(Pre:
84.1 1.3% vs. Post: 85.1 1.2%). In agreement, Iaia
et al. (2009) observed a FVO2-max of 84.8% and 81.6%
at v10-km (14.5 km/h) before and after, respectively, a 4-
wk intervention period with speed endurance training.
41
42
43
44
45
46
47
48
49
50
Pre
Post
10-km time (min)
Depleted
Normal
**
**
###
###
Figure 6. Time to complete a 10-km run in depleted and normal
conditions before (Pre) and after (Post) 5 blocks/40 days of speed
endurance training and reduced training volume in trained runners.
Values are means SE. **Post different (P < 0.01) to Pre;
###difference (P < 0.001) within time-point.
Table 2. Plasma testosterone (T), cortisol (C) and T:C ratio, hemo-
globin and hematocrit, creatine kinase, (CK), and Immunoglobulin
A, (IgA) and HCO3
before (Pre) and after (Post) 5 blocks/40 days
of speed endurance training and reduced training volume in
trained runners.
Pre
Post
Testosterone (nmol/L)
24.2 2.9
25.4 3.5
Cortisol (nmol/L)
195.8 21.2
168.3 25.2
t:c ratio
0.16 0.03
0.21 0.03*
Hemoglobin (mmol/L)
8.9 0.2
8.5 0.2
Hematocrit (%)
44.2 1.1
41.8 0.9
CK (U/L)
208 41
155 17
IgA (g/L)
1.98 0.19
2.06 0.19*
HCO3
(mmol/L)
24.3 0.4
25.0 0.7
Data are presented as means SE.
*Post different (P < 0.05) to Pre.
ª 2018 The Authors. Physiological Reports published by Wiley Periodicals, Inc. on behalf of
The Physiological Society and the American Physiological Society
2018 | Vol. 6 | Iss. 3 | e13601
Page 9
C. Skovgaard et al.
Muscle Fiber Type Adaptations to Speed Endurance Training in Runners
And in the study by Bangsbo et al. (2009), FVO2-max
was 85.7% and 83.6% at v10-km (16.0 km/h) before and
after, respectively, a 6–9-week period with speed endur-
ance training and a basic volume of aerobic training in
trained runners. These observations suggest that changes
in FVO2-max do not explain the improved 10-km perfor-
mance with speed endurance training and reduced train-
ing volume. Thus, the improved performance of the 10-
km run appears mainly to be caused by the better run-
ning economy. It should be noted, however, that the
anaerobic energy production during the 10-km run,
which is suggested to amount up to 20% of the energy
provided during a 10-km run (Joyner and Coyle 2008), is
not taken into account in the calculation. In the present
study, anaerobic energy production may have been higher
after the speed endurance training period due to a possi-
ble higher anaerobic capacity reflected by the finding of
unchanged VO2-max and higher maximal speed during
the incremental test. In support, maximal activity of PFK
was higher after the intervention period, which theoreti-
cally may have promoted a higher energy production
from glycolysis during the 10-km run.
In summary, running economy was improved after
40 days of intense and reduced volume of training, which
may have been related to a reduced expression of UCP3
and higher expression of dystrophin in ST muscle fibers
and lower expression of SERCA1 in FT muscle fibers. The
finding that running economy at 60% VO2-max in a ST
muscle fiber glycogen-depleted condition was unchanged,
suggests that the change in running economy was due to
adaptation in ST muscle fibers. The better running econ-
omy may explain the improved 10-km running perfor-
mance together with a possibly higher anaerobic capacity.
Acknowledgments
We thank all the participants for their valuable time and
extraordinary effort during training and testing. We also
gratefully acknowledge Jens Jung Nielsen and Jon Egelund
for excellent technical assistance and Thomas Gunnarsson,
Morten Hostrup, Lars Nybo, Johan Onslev, Sajad Habib,
Steffen Raun, Julian Christoffer Bachmann, Peter Munch
Larsen and Stefan Madsen for invaluable help during
training and experimental days.
Conflict of Interest
None declared.
References
Bangsbo, J. 2015. Performance in sports - with specific
emphasis on the effect of intensified training. Scand. J. Med.
Sci. Sport 25:88–99.
Bangsbo, J., T. P. Gunnarsson, J. Wendell, L. Nybo, and M.
Thomassen. 2009. Reduced volume and increased training
intensity elevate muscle Na+-K+ pump 2-subunit expression
as well as short- and long-term work capacity in humans. J.
Appl. Physiol. 107:1771–1780.
Barclay, C. J., J. K. Constable, and C. L. Gibbs. 1993.
Energetics of fast- and slow-twitch muscles of the mouse. J.
Physiol. 472:61–80.
Berchtold, M. W., H. Brinkmeier, and M. M€untener. 2000.
Calcium ion in skeletal muscle: its crucial role for muscle
function, plasticity, and disease. Physiol. Rev. 80:1215–1265.
Bergstrom, J. 1962. Muscle electrolytes in man. Scand. J. Clin.
Lab. Investig. 68:1–110.
Bickham, D. C., D. J. Bentley, P. F. Le Rossignol, and D.
Cameron-Smith. 2006. The effects of short-term sprint
training on MCT expression in moderately endurance-
trained runners. Eur. J. Appl. Physiol. 96:636–643.
Boss, O., T. Hagen, and B. B. Lowell. 2000. Uncoupling
proteins 2 and 3: potential regulators of mitochondrial
energy metabolism. Diabetes 49:143–156.
Chopard, A., N. Arrighi, A. Carnino, and J. F. Marini. 2005.
Changes in dysferlin, proteins from dystrophin glycoprotein
complex, costameres, and cytoskeleton in human soleus and
vastus lateralis muscles after a long-term bedrest with or
without exercise. FASEB J. 19:1722–1724.
Christensen, P. M., T. P. Gunnarsson, M. Thomassen, D. P.
Wilkerson, J. J. Nielsen, and J. Bangsbo. 2015. Unchanged
content of oxidative enzymes in fast-twitch muscle fibers
and kinetics after intensified training in trained cyclists.
Physiol. Rep. 3:e12428.
Clausen, T., C. Van Hardeveld, M. E. Everts, H. C. Van, and
M. E. Everts. 1991. Significance of cation transport in
control of energy metabolism and thermogenesis. Physiol.
Rev. 71:733–774.
Crow, M. T., and M. J. Kushmerick. 1982. Chemical energetics
of slow- and fast-twitch muscles of the mouse. J. Gen.
Physiol. 79:147–166.
Delbono, O., and G. Meissner. 1996. Sarcoplasmic reticulum
Ca2+ release in rat slow- and fast-twitch muscles. J. Membr.
Biol. 151:123–130.
Egan, B., and J. R. Zierath. 2013. Exercise metabolism and the
molecular regulation of skeletal muscle adaptation. Cell
Metab. 17:162–184.
Egan, B., B. P. Carson, P. M. Garcia-Roves, A. V.
Chibalin, F. M. Sarsfield, N. Barron, et al. 2010. Exercise
intensity-dependent regulation of peroxisome proliferator-
activated receptor coactivator-1 mRNA abundance is
associated with differential activation of upstream
signalling kinases in human skeletal muscle. J. Physiol.
588:1779–1790.
Froemming, G. R., B. E. Murray, S. Harmon, D. Pette, and K.
Ohlendieck. 2000. Comparative analysis of the isoform
expression pattern of Ca2+-regulatory membrane proteins in
fast-twitch, slow-twitch, cardiac, neonatal and chronic low-
2018 | Vol. 6 | Iss. 3 | e13601
Page 10
ª 2018 The Authors. Physiological Reports published by Wiley Periodicals, Inc. on behalf of
The Physiological Society and the American Physiological Society
Muscle Fiber Type Adaptations to Speed Endurance Training in Runners
C. Skovgaard et al.
frequency stimulated muscle fibers. Biochim. Biophys. Acta
– Biomembr. 1466:151–168.
Gollnick, P. D., K. Piehl, and B. Saltin. 1974. Selective
glycogen depletion pattern in human muscle fibres after
exercise of varying intensity and at varying pedalling rates. J.
Physiol. 241:45–57.
Gong, D. W., Y. He, M. Karas, and M. Reitman. 1997.
Uncoupling protein-3 is a mediator of thermogenesis
regulated by thyroid hormone, b3-adrenergic agonists, and
leptin. J. Biol. Chem. 272:24129–24132.
Green, H. J., M. Burnett, H. Kollias, J. Ouyang, I. Smith, and
S. Tupling. 2011. Malleability of human skeletal muscle
sarcoplasmic reticulum to short-term training. Appl.
Physiol. Nutr. Metab. 36:904–912.
Gunnarsson, T. P., P. M. Christensen, K. Holse, D.
Christiansen, and J. Bangsbo. 2012. Effect of additional
speed endurance training on performance and muscle
adaptations. Med. Sci. Sports Exerc. 44:1942–1948.
He, Z. H., R. Bottinelli, M. A. Pellegrino, M. A. Ferenczi, and
C. Reggiani. 2000. ATP consumption and efficiency of
human single muscle fibers with different myosin isoform
composition. Biophys. J . 79:945–961.
Howley, E. T., D. R. Bassett, and H. G. Welch. 1995. Criteria
for maximal oxygen uptake: review and commentary. Med.
Sci. Sports Exerc. 27:1292–1301.
Hughes, D. C., M. A. Wallace, and K. Baar. 2015. Effects of
aging, exercise, and disease on force transfer in skeletal
muscle. Am. J. Physiol. – Endocrinol. Metab. 309:E1–E10.
Iaia, F. M., and J. Bangsbo. 2010. Speed endurance training is
a powerful stimulus for physiological adaptations and
performance improvements of athletes. Scand. J. Med. Sci.
Sport 20:11–23.
Iaia, F. M., M. Thomassen, H. Kolding, T. Gunnarsson, J.
Wendell, T. Rostgaard, et al. 2008. Reduced volume but
increased training intensity elevates muscle Na+-K+ pump
alpha1-subunit and NHE1 expression as well as short-term
work capacity in humans. J. Appl. Physiol. 106:73–80.
Iaia, F. M., Y. Hellsten, J. J. Nielsen, M. Fernstrom, K. Sahlin,
J. Bangsbo, et al. 2009. Four weeks of speed endurance
training reduces energy expenditure during exercise and
maintains muscle oxidative capacity despite a reduction in
training volume. J. Appl. Physiol. 106:73–80.
Jansson, E., and L. Kaijser. 1977. Muscle adaptation to extreme
endurance training in man. Acta Physiol. Scand. 100:315–
324.
Joyner, M. J., and E. F. Coyle. 2008. Endurance exercise
performance: the physiology of champions. J. Physiol.
586:35–44.
Krustrup, P., K. Soderlund, M. Mohr, and J. Bangsbo. 2004.
Slow-twitch fiber glycogen depletion elevates moderate-
exercise fast-twitch fiber activity and O2 uptake. Med. Sci.
Sports Exerc. 36:973–982.
Krustrup, P., N. H. Secher, M. U. Relu, Y. Hellsten, K.
Soderlund, and J. Bangsbo. 2008. Neuromuscular blockade
of slow twitch muscle fibres elevates muscle oxygen uptake
and energy turnover during submaximal exercise in humans.
J. Physiol. 586:6037–6048.
Lowry, O. H., and J. V. Passonneau. 1972. A flexible system of
enzymatic analysis. P. 237. Academic, New York.
Majerczak, J., J. Karasinski, and J. A. Zoladz. 2008.
Training induced decrease in oxygen cost of cycling is
accompanied by down-regulation of serca expression in
human vastus lateralis muscle. J. Physiol. Pharmacol.
59:589–602.
Majerczak, J., M. Korostynski, Z. Nieckarz, Z. Szkutnik, K.
Duda, and J. A. Zoladz. 2012. Endurance training
decreases the non-linearity in the oxygen uptake-
power output relationship in humans. Exp. Physiol.
97:386–399.
Mogensen, M., M. Bagger, P. K. Pedersen, M. Fernstr€om, and
K. Sahlin. 2006. Cycling efficiency in humans is related to
low UCP3 content and to type I fibres but not to
mitochondrial efficiency. J. Physiol. 571:669–681.
Murphy, R. M. 2011. Enhanced technique to measure
proteins in single segments of human skeletal muscle
fibers: fiber-type dependence of AMPK-a1 and -b1.
J Appl Physiol. 110:820–825. http://doi.org/10.1152/
japplphysiol.01082.2010
Nordsborg, N. B., C. Lundby, L. Leick, and H. Pilegaard.
2010. Relative workload determines exercise-induced
increases in PGC-1a mRNA. Med. Sci. Sports Exerc.
42:1477–1484.
Prins, K. W., J. L. Humston, A. Mehta, V. Tate, E. Ralston,
and J. M. Ervasti. 2009. Dystrophin is a microtubule-
associated protein. J. Cell Biol. 186:363–369.
Russell, A. P., E. Somm, M. Praz, A. Crettenand, O. Hartley,
A. Melotti, et al. 2003a. UCP3 protein regulation in human
skeletal muscle fibre types I, IIa and IIx is dependent on
exercise intensity. J. Physiol. 550:855–861.
Russell, A. P., G. Wadley, M. K. C. Hesselink, G. Schaart, S.
Lo, B. Leger, et al. 2003b. UCP3 protein expression is lower
in type I, IIa and IIx muscle fiber types of endurance-
trained compared to untrained subjects. Pflugers Arch. Eur.
J. Physiol. 445:563–569.
Rybakova, I. N., J. R. Patel, and J. M. Ervasti. 2000. The
dystrophin complex forms a mechanically strong link
between the sarcolemma and costameric actin. J. Cell Biol.
150:1209–1214.
Saunders, P. U., D. B. Pyne, R. D. Telford, and J. A. Hawley.
2004. Factors affecting running economy in trained distance
runners. Sport Med. 34:465–485.
Schiaffino, S., and C. Reggiani. 2011. Fiber types in
mammalian skeletal muscles. Physiol. Rev. 91:1447–1531.
Skovgaard, C., P. M. Christensen, S. Larsen, T. Andersen, M.
Thomassen, and J. Bangsbo. 2014. Concurrent speed
endurance and resistance training improves performance,
running economy, and muscle NHE1 in moderately trained
runners. J. Appl. Physiol. 117:1097–1109.
ª 2018 The Authors. Physiological Reports published by Wiley Periodicals, Inc. on behalf of
The Physiological Society and the American Physiological Society
2018 | Vol. 6 | Iss. 3 | e13601
Page 11
C. Skovgaard et al.
Muscle Fiber Type Adaptations to Speed Endurance Training in Runners
Skovgaard, C., N. Brandt, H. Pilegaard, and J. Bangsbo. 2016.
Combined speed endurance and endurance exercise amplify
the exercise-induced PGC-1a and PDK4 mRNA response in
trained human muscle. Physiol. Rep.
4:e12864.
Smith, I. C., E. Bombardier, C. Vigna, and A. R. Tupling. 2013.
ATP consumption by sarcoplasmic reticulum Ca2+ pumps
accounts for 40–50% of resting metabolic rate in mouse fast
and slow twitch skeletal muscle. PLoS ONE 8:1–11.
Vøllestad, N. K., and P. C. Blom. 1985. Effect of varying
exercise intensity on glycogen depletion in human muscle
fibres. Acta Physiol. Scand. 125:395–405.
Walsh, B., R. A. Howlett, C. M. Stary, C. A. Kindig, and
M. C. Hogan. 2006. Measurement of activation energy
and oxidative phosphorylation onset kinetics in isolated
muscle fibers in the absence of cross-bridge cycling. Am.
J. Physiol. Regul. Integr. Comp. Physiol. 290:R1707–
R1713.
2018 | Vol. 6 | Iss. 3 | e13601
Page 12
ª 2018 The Authors. Physiological Reports published by Wiley Periodicals, Inc. on behalf of
The Physiological Society and the American Physiological Society
Muscle Fiber Type Adaptations to Speed Endurance Training in Runners
C. Skovgaard et al.
| Effect of speed endurance training and reduced training volume on running economy and single muscle fiber adaptations in trained runners. | [] | Skovgaard, Casper,Christiansen, Danny,Christensen, Peter M,Almquist, Nicki W,Thomassen, Martin,Bangsbo, Jens | eng |
PMC10453861 | Citation: Poole, G.; Harris, C.;
Greenough, A. Exercise Capacity in
Very Low Birth Weight Adults: A
Systematic Review and
Meta-Analysis. Children 2023, 10,
1427. https://doi.org/10.3390/
children10081427
Academic Editor: Srinivas Bolisetty
Received: 28 June 2023
Revised: 31 July 2023
Accepted: 5 August 2023
Published: 21 August 2023
Copyright:
© 2023 by the authors.
Licensee MDPI, Basel, Switzerland.
This article is an open access article
distributed
under
the
terms
and
conditions of the Creative Commons
Attribution (CC BY) license (https://
creativecommons.org/licenses/by/
4.0/).
children
Systematic Review
Exercise Capacity in Very Low Birth Weight Adults: A
Systematic Review and Meta-Analysis
Grace Poole 1, Christopher Harris 1 and Anne Greenough 2,*
1
Neonatal Intensive Care Centre, King’s College Hospital NHS Foundation Trust, London SE5 9RS, UK;
[email protected] (G.P.); [email protected] (C.H.)
2
Department of Women and Children’s Health, Faculty of Life Sciences and Medicine, King’s College London,
London SE5 9RS, UK
*
Correspondence: [email protected]
Abstract: There is an association between very low birth weight (VLBW) and cardiovascular mor-
bidity and mortality in adulthood. Aerobic fitness, measured as the maximal oxygen consumption
(VO2 max), is a good indicator of cardiopulmonary health and predictor of cardiovascular mortal-
ity. Our aim was to determine the effect of birth weight on aerobic exercise capacity and physical
activity. We systematically identified studies reporting exercise capacity (VO2 max and VO2 peak)
and physical activity levels in participants born at VLBW aged eighteen years or older compared
to term-born controls from six databases (MEDLINE, OVID, EMBASE, CI NAHL, CENTRAL, and
Google Scholar). Meta-analysis of eligible studies was conducted using a random effect model.
We screened 6202 articles and identified 15 relevant studies, 10 of which were eligible for meta-
analysis. VLBW participants had a lower VO2 max compared to their term counterparts (−3.35,
95% CI: −5.23 to −1.47, p = 0.0005), as did VLBW adults who had developed bronchopulmonary
dysplasia (−6.08, 95% CI −11.26 to −0.90, p = 0.02). Five of nine studies reported significantly re-
duced self-reported physical activity levels. Our systematic review and meta-analysis demonstrated
reduced maximal aerobic exercise capacity in adults born at VLBW compared to term-born controls.
Keywords: very low birth weight; VLBW; neonatal intensive care; VO2 max; exercise capacity;
exercise tolerance; physical activity
1. Introduction
According to the World Health Organization (WHO), 15–20% infants worldwide are
born at a low birth weight (LBW, <2500 g) [1]. Since the introduction of neonatal intensive
care units, there has been a dramatic improvement in survival rates of very low birth weight
infants (VLBW, <1500 g) [2,3]. The increased survival has resulted in a focus on morbidity
and mortality of these cohorts in later life.
The impact of preterm birth on multi-organ development can have deleterious effects
on the cardiopulmonary system [4–7]. Cardiac magnetic resonance imaging (CMR) has
demonstrated structural myocardial changes in adolescents and adults born prematurely
which may be associated with reduced functional reserve [4,8]. A study of 102 adults
born prematurely demonstrated they had greater left ventricular (LV) mass, smaller in-
ternal diameters, and poorer LV strain compared to term-born controls [4]. Mohamed
et al. supported those findings reporting smaller LV volumes and reduced LV function in
200 preterm adults [9]. On stress echocardiography, a lower ejection fraction (EF) propor-
tional to exercise intensity was demonstrated [8].
Given the increased prevalence of cardiovascular risk factors described in prematurely
born individuals, it is not surprising that birthweight is inversely proportional to adult
morbidity and mortality from cardiovascular disease [6,10,11]. In 1991, Barker et al. re-
ported diminished airway function in adults born of reduced birthweight and speculated
Children 2023, 10, 1427. https://doi.org/10.3390/children10081427
https://www.mdpi.com/journal/children
Children 2023, 10, 1427
2 of 15
that this may be secondary to poor prenatal nutrition [12]. Subsequent evidence has shown
that LBW is associated with excess respiratory morbidity, independent of or secondary
to premature birth or in utero growth retardation (small for gestational age—SGA) [7].
Indeed, prematurity and being born SGA are associated with different risk factors, but
many prematurely born infants are SGA. Individuals who developed bronchopulmonary
dysplasia (BPD) are particularly at increased risk of chronic respiratory morbidity includ-
ing an increased requirement for supplementary oxygen following neonatal discharge,
more hospital readmissions particularly for respiratory viral infections, and lung function
abnormalities persisting even into adulthood. Sadly, such infants may also suffer hear-
ing and visual impairment, feeding difficulties, growth restriction, and chronic kidney
disease [13,14].
The impact of preterm birth on multi-organ development can have deleterious effects
on the cardiopulmonary system [4–7]. Furthermore, birthweight has been shown to be
inversely proportional to adult morbidity and mortality from cardiovascular disease [8,9].
Measurement of the maximal oxygen consumption (VO2 max) is considered the gold
standard assessment of cardiorespiratory fitness [15]. It has been shown to be a strong
predictor of cardiovascular health, morbidity, and mortality [16,17]. A systematic review
of studies of maximal aerobic exercise capacity found a 13% reduction in VO2 max in
children and adults born prematurely, compared to their term-born counterparts. While
the pooled results were significantly different, it was highlighted that the majority of
included observational studies showed no significant difference in VO2 max between the
two groups [18–20]. To our knowledge, no systematic review has exclusively focused
on determining whether there was an association between VLBW and maximal aerobic
exercise capacity in adults.
As a strong predictor of cardiovascular health, an improvement in VO2 max may re-
duce the risk of cardiovascular disease and associated mortality [21]. It is well-established
that regular exercise is an effective means of increasing VO2 max [22]. In addition to its
important association with VO2 max, physical activity levels are an important independent
protective factor for cardiovascular health [23]. Several studies have suggested that prema-
turely born individuals are less physically active than their term-born peers, independent
of whether they had developed bronchopulmonary dysplasia (BPD) and socio-economic
confounders during childhood [24–26]. However, the evidence is conflicting [27]; the
Epicure study did not reveal any significant differences in physical activity levels between
school-age children born prematurely and term-born controls when measured using ac-
celerometers. Therefore, the relationship between birth weight and physical activity levels
in adulthood would benefit from further clarification.
The evidence, however, is conflicting [19]. A study of 61 children found no significant
differences in physical activity levels assessed using an accelerometer [19]. In addition to
its important association with VO2 max, physical activity levels are also an independent
risk factor that may contribute to cardiovascular morbidity and mortality.
Our primary aim was to undertake a systematic review to evaluate the impact of
VLBW on exercise capacity in adults as assessed by VO2 max. Our secondary outcome was
to compare self-reported physical activity levels between VLBW and term-born adults and
assess whether this impacted on exercise capacity.
2. Materials and Methods
2.1. Methods
This systematic review and meta-analysis was prospectively registered on PROSPERO
at https://www.crd.york.ac.uk/prospero/ (accessed on 25 May 2023) as CRD42023429309 [28].
The literature search was conducted according to the Meta-analysis of Observational
Studies in Epidemiology (MOOSE) guidelines [29]. The Preferred Reporting Items for
Systematic Reviews and Meta-analyses (PRIMSA) guidelines were used to prepare the
manuscript [30].
Children 2023, 10, 1427
3 of 15
2.2. Search Strategy
Relevant studies were identified through searching six electronic databases (EMBASE,
OVID MEDLINE(R.), Scopus, CENTRAL, CINAHL, and Google Scholar) between the
1 March 2023 and 15 March 2023. A repeat search was conducted on the 1 June 2023 to
identify any further articles that met inclusion criteria. We also hand-searched references
from included articles. Search strategies were based on the Cochrane library Neonatal
Search Terms [31].
2.3. Eligibility Criteria
Studies on exercise capacity in adults (defined as greater than 18 years old) born at
VLBW compared to term controls with results of VO2 max (mL/kg/min) or VO2 Peak
(mL/kg/min) using a treadmill or cycle ergometer were eligible. VO2 max refers to the
maximum rate of oxygen consumption attainable during physical exertion. VO2 peak,
directly reflective of VO2 max, is the highest value VO2 attained upon incremental or
other high-intensity exercise testing [32]. The recruitment of subjects in some papers was
based on gestational age, but only those which reported birthweight were included in the
review. BPD was defined as either dependence on supplementary oxygen at 28 days of life
or dependence on supplementary oxygen at 36 weeks postmenstrual age (PMA). Other
measures of cardiorespiratory exercise capacity were reviewed (such as anaerobic threshold
and minute ventilation), but there were insufficient data for a meta-analysis. Given the
authors’ capabilities, studies were restricted to those reported in the English language.
Using pre-agreed inclusion criteria, two independent authors (GP and CH) removed
duplicates, screened titles and abstracts of retrieved articles, and obtained full-text articles.
Any disagreements were resolved through discussion between the two reviewers until a
consensus was achieved.
2.4. Data Analysis
Data extraction was performed by a single reviewer (GP) using a pre-specified data
extraction form. A second reviewer (CH) independently checked the accuracy of the first
extraction. Study characteristics, sample size, the method of assessing exercise capacity,
and reference values were summarised for each study. For each study, VO2 max, VO2 peak,
and activity levels were extracted for adults born at VLBW and term-born controls.
To be eligible for meta-analysis, a study had to fulfil the following criteria, defined a
priori: an original report on the relation between exercise capacity in adults that were born
at VLBW, odds ratios (OR), and 95% confidence intervals (95% CI) for exercise tolerance
in at least two strata of birth weight. To assess exercise capacity, we analysed results for
VO2 max and VO2 peak. Meta-analyses were conducted using Review Manager (RevMan)
version 5.4 [33].
2.5. Quality Assessment
The risk of bias for each study was assessed by two independent reviewers (GP
and CH) using the Newcastle–Ottawa Scale for cohort and cross-sectional studies [34,35].
Studies were scored across three domains: case selection, comparability, and outcome.
Scores across three domains were tabulated to give an overall rating of good, fair, or poor
quality. The data extraction for quality was performed by a single reviewer (GP) and three
randomly chosen papers were checked for consistency by a second reviewer (CH), with
no discrepancies being identified. For cohort studies, we considered that participants lost
to follow-up were unlikely to introduce bias if follow-up rates were greater than 80% or
between 70% and 80% with an accompanying statement describing those lost to follow up.
3. Results
3.1. Identified Studies and Characteristics
The course of the systematic review is outlined in a PRIMSA 2020 flow diagram
(Figure 1). Seven thousand, eight hundred and seven studies were identified through
Children 2023, 10, 1427
4 of 15
database searching. A total of 1605 duplicates were removed, and 6202 abstracts were
screened. Eighty-one full-text articles were screened for eligibility, and the quality of
fifteen studies was evaluated [8,24,36–47]. The characteristics of the studies included are
summarised in Table 1. From all included studies, there were 1132 VLBW participants,
914 controls, and 75 VLBW who had had BPD. Individuals were born between 1984 and
1998. Participants were assessed at ages 18 to 30 years old [8,24,36–47].
Figure 1. Preferred Reporting Items for Systematic Reviews and Meta-analyses (PRIMSA) flow
diagram outlining the course of our systematic literature search for articles evaluating maximal
aerobic exercise capacity in adults born at very low birth weight. Six databases (SCOPUS, EMBASE,
MEDLINE, CINAHL, CENTRAL, and Google Scholar) were searched. A total of 6202 articles were
screened and 81 articles assessed for eligibility. Fifteen studies were included in the review and ten in
the meta-analysis.
Children 2023, 10, 1427
5 of 15
Table 1. Study characteristics of included studies.
Author
Country
Study
Design
Number of
VLBW
Infants
Number of
Control
Subjects
Age at
Follow Up
(Years)
Outcome Measures
Vrijlandt et al.,
2006 [36]
Netherlands
Prospective
cohort study
42
48
18–22
•
VO2 max
•
Physical activity level
Evensen et al.,
2009 [37]
Norway
Prospective
cohort study
32
51
18
•
VO2 max
Narang et al.,
2009 [38]
UK
Prospective
cohort study
57
50
20–25
•
Physical activity level
Sipola-
Leppanen et al.,
2011 [39]
Finland
Prospective
cohort study
116
118
20–28
•
Physical activity level
Lovering et al.,
2013 [40]
USA
Prospective
cohort study
12
12
18–27
•
VO2 peak
Clemm et al.,
2014 [24]
Norway
Prospective
cohort study
34
33
24–25
•
VO2 peak
•
VO2 peak at anaerobic
threshold
•
Physical activity level
Duke et al.,
2014 [48]
USA
Prospective
cohort study
13
14
20–25
•
VO2 peak
Saarenpaa et al.,
2015 [42]
Finland
Prospective
cohort study
160
162
20–25
•
Physical activity level
Farrell et al.,
2015 [43]
USA
Prospective
cohort study
14
16
20–23
•
VO2 max
Caskey et al.,
2016 [44]
UK
Prospective
cohort study
20
24
23–30
•
VO2 peak
Kasaeva et al.,
2012 [45]
Finland
Prospective
cohort study
94
101
21–27
•
Physical activity level
Haraldsdottir
et al., 2020 [46]
USA
Prospective
cohort study
12
12
24–28
•
VO2 max at normoxia
and hypoxia
Huckstep et al.,
2018 [8]
UK
Prospective
cohort study
47
54
20–26
•
VO2 max
•
Physical activity level
Yang et al., 2022
[47]
New
Zealand
Prospective
cohort study
202
93
26–30
•
VO2 peak
•
Physical activity level
Cheong et al.,
2023 [49]
Australia
Prospective
cohort study
128
126
25
•
Six-minute walk test
•
Maximum beep test
level
3.2. VO2 Max in VLBW Infants
Ten studies assessed VO2
max in VLBW adults compared to term-born
controls [24,36,37,40,41,43,44].
Three studies undertook a subgroup analysis for VLBW individuals who had BPD,
including 75 participants. Table 2 summarises the characteristics and results of the studies
included in the analysis. The average weight of VLBW adults at follow-up was 68.85 kg,
compared to 73.75 kg in the control group (Student’s t test, p = 0.034). Four studies assessed
VO2 max, three of which used a cycle ergometer [8,36,37,46]. Six studies assessed VO2 peak
using a combination of cycle ergometry and treadmill exercise protocols [24,40–43,47].
Children 2023, 10, 1427
6 of 15
Table 2. Summary of physical maximum aerobic exercise capacity in adults born at very low birth weight.
Study
Birth
Weight
VLBW
Infants (g)
Birth
Weight
Control
Subjects
(g)
Age at
Follow
Up
(years)
Weight
VLBW
Adults
(kg)
Weight
Control
Group
(kg)
VO2 Max/Peak
Measurement
VO2 Mea-
surement
in VLBW
Group
(mL/kg/min)
VO2 Mea-
surement
in Control
Group
(mL/kg/min)
VO2 Mea-
surement
in BPD
Group
(mL/kg/min)
Vrijlandt
et al., 2006
[36]
1246 ± 232
-
18–22
65 ± 10
72 ± 10
•
VO2 Max
•
Cycle
Ergometer
35.3 ± 6.9
20.8 ± 1.2
-
Evensen
et al., 2009
[37]
1245
(800–1500)
3700 (2670–
5140)
18
64.2 ± 1.7
69.8 ± 1.3
•
VO2 Max
•
Treadmill
48.8 ± 1.4
48.5 ± 1.1
-
Lovering
et al., 2013
[40]
1160 ± 450
-
21–24
64.7 ± 9.3
75.7 ±
10.4
•
VO2 Peak
•
Cycle
Ergometer
40.6 ± 9.4
48.8 ± 7.6
40.7 ± 14.3
Clemm
et al., 2014
[24]
1173 ± 163
-
24–25
71.5 ± 4.3
72.3 ± 5.9
•
VO2 Peak
•
Treadmill
40.7 ± 2.8
44.2 ± 3.2
-
Duke et al.,
2014 [48]
1080 ± 430
-
20–25
65 ± 10
72 ± 12
•
VO2 Peak
•
Cycle
Ergometer
35.0 ± 9.0
48.0 ± 9.0
-
Farrell
et al., 2015
[43]
1027 ± 296
>1500
20–23
76.3 ± 5.0
71.8 ± 5.4
•
VO2 Peak
•
Cycle
Ergometer
39.5 ± 1.7
38.9 ± 1.6
-
Caskey
et al., 2016
[44]
1234 ± 205
3569 ± 297
21–30
-
-
•
VO2 Peak
•
Cycle
Ergometer
45.2 ±11.3
39.3 ± 8.8
35.6 ± 7.5
Haraldsdottir
et al., 2020
[46]
<1500 g
-
24–28
70.1 ±
13.3
75.6 ± 0.7
•
VO2 Max
•
Cycle
Ergometer
34.88 ±
9.26
45.79 + 8.71
-
Huckstep
et al., 2021
[50]
1916 ± 806
3390 ± 424
22–27
-
-
•
VO2 Max
•
Cycle
Ergometer
33.6 ± 8.6
40.1 ± 9.0
-
Yang et al.,
2022 [47]
1131 ± 233
3362 ± 529
28–29
74.1 ±
18.8
80.8 ±
16.3
•
VO2 Peak
•
Cycle
Ergometer
30.46 ±
8.06
31.45 ±
7.86
28.3 ± 1.1
Meta-analysis indicated that adults born at VLBW had a significantly lower VO2
max/VO2 peak compared to controls (mean difference: −3.35 [95% CI −5.23 to −1.47]
mL/kg/min, p = 0.0005) (Figure 2). However, there was high heterogeneity between
studies included in the analysis (I2 = 87%). Meta-analysis of studies solely reporting on
those who had BPD found they were more likely to have a lower VO2 max (mean difference:
−6.08 [95% CI −11.26 to −0.90] mL/kg/min, p = 0.02,) than term-born controls. There
were no significant differences in VO2 max between VLBW participants who had BPD and
those that did not (p = 0.33).
Children 2023, 10, 1427
7 of 15
Children 2023, 10, x. https://doi.org/10.3390/xxxxx
www.mdpi.com/journal/children
Figure 2. Forest plot of aerobic exercise capacity (VO2 max/VO2 peak) in adults born at very low
birth weight compared to their term-born counterparts [24,36,37,40,43,44,46–48,50].
3.3. Levels of Physical Activity in VLBW Infants
Nine studies from five different countries reported on physical activity levels in VLBW
adults compared to controls (Table 3) [8,26,37–39,42,44,47]. Three studies followed up
individuals from the Helsinki Study of VLBW Adults [39,41,44]. Three studies used the
European Community Respiratory Health Survey II to identify adult’s physical activity
levels [36,44,47,51]. Five studies did not report on questionnaires or tools utilised to assess
physical activity levels [8,24,37,39,42].
Five studies found significant differences in self-reported activity levels in VLBW
adults compared to controls [36,39,44,45,47]. In four studies, VLBW adults were less likely
to engage in weekly vigorous physical activity [36,44,45,47]. In one study, despite no
significant difference in the frequency of exercise, VLBW adults were more likely to engage
in less intense physical activity for a shorter duration of time [39]. Four studies found
no significant difference in the frequency of physical activity between VLBW and control
groups [8,24,37,42]. Due to a difference in measurable outcomes and assessment tools, the
results were unsuitable for meta-analysis.
Children 2023, 10, 1427
8 of 15
Table 3. Summary of physical activity levels in adults born at very low birth weight.
Study
Weight
VLBW
Infants (g)
Weight
Control
Subjects (g)
Age at
Follow Up
VLBW
Infants
(Years)
Age at Follow Up
Control Subjects
(Years)
PA Assessment Measure
Results
Summary of
Impact
Vrijlandt et al.,
2006 [36]
1246 ± 232
-
19 ± 0.3
20.8 ± 1.2
•
European Community Respiratory
Health Survey II
•
Mean hours of vigorous exercise per
week
•
Preterm born infants undertake
significantly less vigorous exercise per
week (1.9 h ± 2) compared to term
born controls (2.9 h ± 2)
Narang et al., 2009
[38]
1440 ± 550
3410 ± 2390
21.7 ± 1.2
23.1 ± 2.0
•
No formal questionnaire reported
•
Mean days engaged with physical
activity per week
•
There was no statistical difference in
time spent being active per week
between VLBW infants (3.0 ± 2.42) and
the control group (3.0 ± 1.79)
Sippola-Leppanen
et al., 2011 [39]
1125 ± 223
3606 ± 469
22.3 ± 2.2
22.6 ± 2.2
•
No formal questionnaire reported
•
Frequency, duration, and intensity of
exercise
•
There was significant difference
between VLBW and control subjects in
the intensity and duration of physical
activity during a typical week
•
There was no difference in the
frequency of activity between groups
Kaseva et al., 2012
[45]
1157 ± 208.7
3608 ± 492
24.9 ± 2.1
25.1 ± 2.2
•
Modified Kuopio Ischaemic Heart
Disease Risk Factor Study
•
Frequency, time, and intensity of
conditioning exercise
•
Frequency, time, and intensity of
leisure-time physical activity
•
No significant difference in commuting,
leisure-time, or conditioning physical
activity between groups
•
VLBW infants were more likely to do
less vigorous activity compared to their
counterparts
Clemm et al., 2014
[24]
1173 ± 163
-
24.7 ± 1.2
25.1 ± 1.2
•
No formal questionnaire reported
•
Categorical hours spent exercising per
week
•
No statistically significant difference in
leisure time spent doing physical
activity between EP and term-born
individuals.
Caskey et al., 2016
[44]
1234 ± 205
3569 ± 297
26.4 ± 3.7
28.3 ± 3.3
•
European Community Respiratory
Health Survey II
•
Frequency exercised 2–3 h per week
•
Statistically significant difference in the
frequency individuals exercised 2–3 h
per week between BPD, non-BPD
adults, and term-born controls
Saarenpaa et al.,
2015 [42]
1126 ± 218
3599 ± 466
22.4 ± 2.1
22.5 ± 2.5
•
No formal questionnaire reported
•
Frequency of exercise per week
•
No statistically significant difference in
the frequency of exercise per week
between VLBW and non VLBW
individuals
Children 2023, 10, 1427
9 of 15
Table 3. Cont.
Study
Weight
VLBW
Infants (g)
Weight
Control
Subjects (g)
Age at
Follow Up
VLBW
Infants
(Years)
Age at Follow Up
Control Subjects
(Years)
PA Assessment Measure
Results
Summary of
Impact
Huckstep et al.,
2018 [8]
1916 ±806
3390 ± 424
22.7 ± 3.04
23.6 ± 3.8
•
No formal questionnaire reported
•
Hours spent doing moderate and
vigorous physical activity per week
•
No statistical difference in hours spent
doing moderate or vigorous activity
between groups
Yang et al.,
2022 [47]
1131 ±233
3362 ± 529
28.3 ± 1.1
28.2 ± 0.9
•
European Community Respiratory
Health Survey II
•
Mean days engaged with physical
activity per week
•
VLBW exercised significantly less (2.9
h ± 2.6) per week compared to
term-born controls (37 ± 2.4).
Children 2023, 10, 1427
10 of 15
4. Discussion
We have demonstrated that exercise capacity is significantly reduced in adults born
at VLBW, independent of whether they had BPD, compared to term born controls. It is
important to consider the origin of differences in VO2 max between VLBW adults and TB
term-born controls. Maximal aerobic exercise capacity is impacted by age, sex, weight,
size, body composition, and physical activity levels. Due to a lack of data reported in
individual papers, we were unable to analyse results to determine if there were differences
related to sex. A follow-up study of 150 adults born prematurely recruited into the United
Kingdom Oscillation Study (UKOS) found males compared to females completed signifi-
cantly greater distances during shuttle sprint testing and reported exercising more each
week [52]. While VO2 max was not assessed in that study, sex differences in the amount
of exercise undertaken could potentially impact on VO2 max. Furthermore, in a study of
elite endurance athletes, women were found to have a VO2 max 10% lower than their male
counterparts [53].
Given the absolute value is highly impacted by body weight, VO2 max and VO2
peak are typically expressed as milliliter/kg/minute. While this enables results to be
adjusted for body weight, body composition remains a likely confounder. Eight out
of ten included studies reported the participants’ weight [8,24,36,37,40,41,43,46,47]. On
pooled analysis, there was a significant difference in the mean weight of adults born at
VLBW compared to term-born controls. While impossible to predict based exclusively
on weight, BMI is generally well-correlated to percentage body fat [54–56]. One study of
25 female athletes aged between 17 and 22 years found a non-significant negative correlation
between percentage body fat and VO2 max [57]. Goren et al. demonstrated a strong
correlation between fat-free mass (FFM) and VO2 max [58]. This may explain the greater
effect size observed in our meta-analysis which focused exclusively on adults compared
to the results of Edwards et al. which also included children [18]. Dual energy X-ray
absorptiometry (DXA) of 433 healthy subjects demonstrated a decline in FFM with age [59].
This emphasises the importance of including anthropometric measurements in studies
investigating maximal aerobic exercise capacity in future.
There is a well-established association between physical activity levels and maximal
aerobic exercise capacity. Meta-regression and analysis of 28 articles highlighted an increase
in VO2 max with physical activity training, independent of the volume and intensity of
exercise sessions [60]. Crowley et al. reported in their systematic review that both high and
low intensity training when undertaken frequently increased VO2 max [61]. Interestingly,
despite adjusting for physical activity levels, Gostelow and Stohr found a significantly
lower VO2 max in individuals born at VLBW [19]. In our review, five studies reported
that adults born at VLBW exercised less than their term-born counterparts [36,39,44,45,47],
whereas four studies found no such association [4,8,24,38]. This highlights the need for
further research assessing the relationship of physical activity levels and VO2 max in
relation to birth weight. If frequent exercise improves maximal aerobic exercise capacity, an
important predictor of cardiovascular morbidity and mortality, a pertinent public health
strategy would be to target educational interventions and physical activity programs at
VLBW adults.
Interestingly, despite adjusting for PA levels, Gostelow and Stohr found a significantly
lower VO2 max in individuals born at VLBW [52,53]. There is improvement in maximal
aerobic exercise capacity with regular exercise, that PA may be used as a potential interven-
tion [52,53]. This raises a question as to whether VLBW infants should be recommended
targeted exercise regimens as a preventative cardiovascular strategy. The physiological
mechanisms resulting in reduced maximal aerobic exercise capacity in adults born at VLBW
remain poorly understood. Aerobic exercise capacity is determined by the integrative
responses of the cardiovascular and respiratory systems, in addition to oxygen uptake by
skeletal muscles [62]. Several studies have demonstrated that [54]. Our findings support
previous research demonstrating a reduced maximal aerobic exercise capacity, independent
of prematurity-related perinatal factors such as BPD [40]. Pulmonary gas exchange during
Children 2023, 10, 1427
11 of 15
exercise, assessed by the alveolar-to-arterial oxygen difference (A-aDO2), is comparable
between prematurely born and term-born TB adults during exercise [40,43,63]. It has
however been hypothesised that adults born at VLBW may have higher airway resistance
and smaller peripheral airways, requiring a greater concentration of oxygen to maintain
ventilation respiration during exercise [64]. Follow-up of the UKOS cohort found males
born prematurely were more likely to have poorer smaller airway function, however, they
performed better on exercise testing compared to their female counterparts, possibly in-
dicating that other factors’ physiological mechanisms may have a greater influence [52].
Adults born prematurely have been shown to have a significantly increased pulmonary ar-
terial pressure during exercise, which may reduce pulmonary blood flow and subsequently
VO2 max [64,65]. Physiological mechanisms contributing to a reduced maximal aerobic
exercise capacity are multi-factorial and complex, where further research is required to
fully understand the impact of birth weight and prematurity.
In addition to a reduced VO2 max, a predictor of increased cardiopulmonary mortality,
adults born at VLBW are also at higher risk due to the increased prevalence of hyperten-
sion [66], heart failure [67], diabetes [68], and cardiometabolic syndromes [66]. Given this,
a more detailed cardiovascular risk assessment in adults known to be born at VLBW may
be of benefit. One suggestion is to screen adults born at VLBW in general practice using
a risk scoring system, such as the widely utilised QRISK2, to predict individuals 10-year
risk of cardiovascular disease [69]. It would be important however to evaluate the financial
cost, resource implications, and the most appropriate and effective age-range to target such
an intervention.
Furthermore, adults born preterm have been shown to have a significant increase in
pulmonary arterial pressure during exercise, which may reduce pulmonary blood flow and
subsequently VO2 max [59,60]. Physiological mechanisms contributing to the observed
reduction in maximal aerobic exercise capacity are multi-factorial and it is clear further
research is required to fully understand the impact of birth weight and prematurity.
Adults born at VLBW are at a higher risk of hypertension [61], heart failure [62], dia-
betes [63], and cardiometabolic syndromes [61]. Interestingly, despite a greater prevalence
of cardiovascular risk factors, an association with ischemic heart disease remains inconclu-
sive [64,65]. An increased relative-risk of all-cause mortality in adults born prematurely,
however, is well-established [66]. A more detailed cardiovascular health assessment in
adults known to be born at VLBW may be of benefit, but evaluation of the efficacy and
resource implications of adult-targeted interventions would be important. Studies such as
the trial of exercise to prevent hypertension in young adults are therefore very welcome,
even though only 38.7% of those born prematurely were born at less than 32 weeks of
gestation [70].
Despite our efforts to generate a precise effect of being born at VLBW on maximal
aerobic exercise capacity, our review has some limitations. On analysis, there was a high
proportion of heterogeneity between studies reporting maximal aerobic exercise capacity.
In part, this may be secondary to our decision to include studies reporting VO2 peak and
VO2 max, however, prior studies have shown that VO2 peak is reflective of VO2 max [32].
The heterogeneity is possibly attributable to different methodologies and protocols used
between studies to estimate maximal aerobic exercise capacity. Eight studies utilised
cycle ergometry [8,36,37,39,41,44,46,47] whereas two studies utilised a treadmill [24,27].
While studies utilising a treadmill demonstrated comparable results to those using a cycle
ergometer in this review, prior studies have commented on lower values of VO2 max
using cycle ergometry when intra-subject comparisons of both methods were utilised in
the same study [71,72]. While challenging, this perhaps highlights a need to standardise
methodology and protocols utilised to measure maximal aerobic exercise capacity.
All participants included in the meta-analysis were between their second and third
decade of life, a period well-established to correlate to peak maximal aerobic exercise
capacity. Generally, it is estimated that VO2 max declines 10% per decade after the age
of 25 years and 15% between the ages of 50 and 75 [73–75]. Most studies included in our
Children 2023, 10, 1427
12 of 15
meta-analysis followed up participants in their third decade of life, with latter follow-up. It
will be interesting to observe the impact of age on differences in VO2 max between adults
born at VLBW and at term.
Due to differences in outcome measures between studies, a meta-analysis could not be
performed to assess self-reported physical activity levels in adults born at VLBW. Given the
correlation between physical activity levels and cardiovascular morbidity and mortality, in
addition to the possibility of reduced activity levels in adults born at VLBW, standardisation
of outcome measures between studies is of vital importance. In studies evaluating maximal
aerobic exercise capacity and physical activity levels, it is important to critically evaluate
participant recruitment given the high risk of recruitment bias associated with exercise-
based studies [76].
5. Conclusions
In conclusion, maximal aerobic exercise capacity was significantly reduced in adults
born at VLBW compared to term-born controls. Given the relationship between exercise
capacity and cardiovascular morbidity and mortality, this could have significant impli-
cations for individuals’ long-term health. The variability in outcome measures assessing
physical activity meant it was difficult to accurately assess the association with birthweight.
We recommend a standardised approach of assessing physical activity levels for future
studies, such as the European Respiratory Health Community Questionnaire II or Metabolic
Equivalent of a Task levels.
Author Contributions: Conceptualization, G.P. and A.G.; methodology, G.P.; validation, C.H.; formal
analysis, G.P.; writing—original draft preparation, G.P.; writing—review and editing, A.G.; visual-
ization, G.P.; supervision, A.G. All authors have read and agreed to the published version of the
manuscript.
Funding: This research received no external funding.
Institutional Review Board Statement: Not applicable.
Informed Consent Statement: Not applicable.
Acknowledgments: The authors would like to acknowledge the work of Deidre Gibbons in the
preparation of this manuscript.
Conflicts of Interest: The authors declare no conflict of interest.
References
1.
World Health Organization. Global Nutrition Targets 2025. Low Birth Weight Policy Brief. Available online: https://www.who.
int/publications/i/item/WHO-NMH-NHD-14.5 (accessed on 17 October 2014).
2.
Jeschke, E.; Biermann, A.; Günster, C.; Böhler, T.; Heller, G.; Hummler, H.D.; Bührer, C. Mortality and major morbidity of
very-low-birth-weight infants in Germany 2008-2012: A report based on administrative data. Front. Pediatr. 2016, 4, 23. [CrossRef]
3.
Cartlidge, P.H.; Stewart, J.H. Survival of very low birthweight and very preterm infants in a geographically defined population.
Acta Paediatr. 1997, 86, 105–110. [CrossRef] [PubMed]
4.
Lewandowski, A.J.; Augustine, D.; Lamata, P.; Davis, E.F.; Lazdam, M.; Francis, J.; McCormick, K.; Wilkinson, A.R.; Singhal, A.;
Lucas, A.; et al. Preterm heart in adult life cardiovascular magnetic resonance reveals distinct differences in left ventricular mass,
geometry, and function. Circulation 2013, 127, 197. [CrossRef] [PubMed]
5.
Lewandowski, A.J.; Bradlow, W.M.; Augustine, D.; Davis, E.F.; Francis, J.; Singhal, A.; Lucas, A.; Neubauer, S.; McCormick,
K.; Leeson, P. Right ventricular systolic dysfunction in young adults born preterm. Circulation 2013, 128, 713–720. [CrossRef]
[PubMed]
6.
Raisi-Estabragh, Z.; Cooper, J.; Bethell, M.S.; McCracken, C.; Lewandowski, A.J.; Leeson, P.; Neubauer, S.; Harvey, N.C.; Petersen,
S.E. Lower birth weight is linked to poorer cardiovascular health in middle-aged population-based adults. Heart 2023, 109,
535–551. [CrossRef]
7.
Greenough, A. Does low birth weight confer a lifelong respiratory disadvantage? Am. J. Respir. Crit. Care Med. 2009, 180, 107–108.
[CrossRef] [PubMed]
8.
Huckstep, O.; Williamson, W.; Telles, F.; Burchert, H.; Bertagnolli, M.; Herdman, C.; Arnold, L.; Smillie, R.; Mohamed, A.;
Boardman, H.; et al. Physiological stress elicits impaired left ventricular function in preterm-born adults. J. Am. Coll. Cardiol.
2018, 71, 1347–1356. [CrossRef]
Children 2023, 10, 1427
13 of 15
9.
Mohamed, A.; Marciniak, M.; Williamson, W.; Huckstep, O.J.; Lapidaire, W.; McCance, A.; Neubauer, S.; Leeson, P.; Lewandowski,
A.J. Association of systolic blood pressure elevation with disproportionate left ventricular remodeling in very preterm-born
young adults: The preterm heart and elevated blood pressure. JAMA Cardiol. 2021, 6, 821–829. [CrossRef]
10.
Risnes, K.R.; Vatten, L.J.; Baker, J.L.; Jameson, K.; Sovio, U.; Kajantie, E.; Osler, M.; Morley, R.; Jokela, M.; Painter, R.C.; et al.
Birthweight and mortality in adulthood: A systematic review and meta-analysis. Int. J. Epidemiol. 2011, 40, 647–661. [CrossRef]
11.
Wang, Y.X.; Li, Y.; Rich-Edwards, J.W.; Florio, A.A.; Shan, Z.; Wang, S.; Manson, J.A.; Mukamal, K.J.; Rimm, E.B.; Chavarro, J.E.
Associations of birth weight and later life lifestyle factors with risk of cardiovascular disease in the USA: A prospective cohort
study. EClinicalMedicine 2022, 51, 101570. [CrossRef]
12.
Barker, D.J.P.; Godfrey, K.M.; Fall, C.; Osmond, C.; Winter, P.D.; Shaheen, S.O. Relation of birth weight and childhood respiratory
infection to adult lung function and death from chronic obstructive airways disease. Br. Med. J. 1991, 303, 671–675. [CrossRef]
13.
Dyson, A.; Kent, A.L. The effect of preterm birth on renal development and renal health outcome. Neoreviews 2019, 20, e725–e736.
[CrossRef] [PubMed]
14.
Bonadies, L.; Cavicchiolo, M.E.; Priante, E.; Moschino, L.; Baraldi, E. Prematurity and BPD: What general practitioners should
know. Eur. J. Pediatr. 2023, 182, 1505–1516. [CrossRef] [PubMed]
15.
Hill, A.V.; Lupton, H. Muscular exercise, lactic acid, and the supply and utilization of oxygen. QJM 1923, 16, 62. [CrossRef]
16.
Bye, A.; Røsjø, H.; Aspenes, S.T.; Condorelli, G.; Omland, T.; Wisløff, U. Circulating MicroRNAs and Aerobic Fitness—The
HUNT-Study. PLoS ONE 2013, 8, e57496. [CrossRef] [PubMed]
17.
Harber, M.P.; Kaminsky, L.A.; Arena, R.; Blair, S.N.; Franklin, B.A.; Myers, J.; Ross, R. Impact of cardiorespiratory fitness on
all-cause and disease-specific mortality: Advances since 2009. Prog. Cardiovasc. Dis. 2017, 60, 11–20. [CrossRef]
18.
Edwards, M.O.; Kotecha, S.J.; Lowe, J.; Watkins, W.J.; Henderson, A.J.; Kotecha, S. Effect of preterm birth on exercise capacity: A
systematic review and meta-analysis. Pediatr. Pulmonol. 2015, 50, 293–301. [CrossRef]
19.
Gostelow, T.; Stöhr, E.J. The effect of preterm birth on maximal aerobic exercise capacity and lung function in healthy adults: A
systematic review and meta-analysis. Sports Med. 2022, 52, 2627–2635. [CrossRef]
20.
Agusti, A.; Faner, R. Lung function trajectories in health and disease. Lancet Respir. Med. 2019, 7, 358–364. [CrossRef]
21.
Myers, J.; Prakash, M.; Froelicher, V.; Do, D.; Partington, S.; Atwood, J.E. Exercise capacity and mortality among men referred for
exercise testing. N. Engl. J. Med. 2002, 346, 793–801. [CrossRef]
22.
di Prampero, P.E. Factors limiting maximal performance in humans. Eur. J. Appl. Physiol. 2003, 90, 420–429. [CrossRef] [PubMed]
23.
Nystoriak, M.A.; Bhatnagar, A. Cardiovascular effects and benefits of exercise. Front. Cardiovasc. Med. 2018, 5, 135. [CrossRef]
24.
Clemm, H.H.; Vollsæter, M.; Røksund, O.D.; Eide, G.E.; Markestad, T.; Halvorsen, T. Exercise capacity after extremely preterm
birth: Development from adolescence to adulthood. Ann. Am. Thorac. Soc. 2014, 11, 537–545. [CrossRef] [PubMed]
25.
Kajantie, E.; Strang-Karlsson, S.; Hovi, P.; Räikkönen, K.; Pesonen, A.K.; Heinonen, K.; Järvenpää, A.L.; Eriksson, J.G.; Andersson,
S. Adults born at very low birth weight exercise less than their peers born at term. J. Pediatr. 2010, 157, 610–616. [CrossRef]
[PubMed]
26.
Engan, M.; Vollsæter, M.; Øymar, K.; Markestad, T.; Eide, G.E.; Halvorsen, T.; Juliusson, P.; Clemm, H. Comparison of physical
activity and body composition in a cohort of children born extremely preterm or with extremely low birth weight to matched
term-born controls: A follow-up study. BMJ Paediatr. Open 2019, 3, e000481. [CrossRef] [PubMed]
27.
Welsh, L.; Kirkby, J.; Lum, S.; Odendaal, D.; Marlow, N.; Derrick, G.; Stocks, J. The EPICure study: Maximal exercise and physical
activity in school children born extremely preterm. Thorax 2010, 65, 165–172. [CrossRef]
28.
University of York, National Institute for Health and Care Research (2023). PROSPERO. Available online: https://www.crd.york.
ac.uk/prospero/ (accessed on 22 June 2023).
29.
Brooke, B.S.; Schwartz, T.A.; Pawlik, T.M. MOOSE reporting guidelines for meta-analyses of observational studies. JAMA Surg.
2021, 156, 787–788. [CrossRef]
30.
Stewart, L.A.; Clarke, M.; Rovers, M.; Riley, R.D.; Simmonds, M.; Stewart, G.; Tierney, J.F. PRISMA-IPD Development Group.
Preferred reporting items for systematic review and meta-analyses of individual participant data: The PRISMA-IPD Statement.
JAMA 2015, 313, 1657–1665. [CrossRef]
31.
The Cochrane Collaberation. Literature Search Filters for Neonatal Reviews. 2023. Available online: https://neonatal.cochrane.
org/Literature-Search-Filters-for-Neonatal-Reviews (accessed on 26 June 2023).
32.
Whipp, B.J.; Ward, S.A. Physiological determinants of pulmonary gas exchange kinetics during exercise. Med. Sci. Sports Exerc.
1990, 22, 62–71. [CrossRef]
33.
Review Manager Web (RevMan Web). Version (5.4). The Cochrane Collaboration. Available online: Revman.cochrane.org
(accessed on 1 August 2023).
34.
Wells, G.; Shea, B.; O’Connell, D.; Peterson, J.; Welch, V.; Losos, M.; Tugwell, P. Newcastle-Ottawa Quality Assessment form for Cohort
Studies; Ottawa Hospital Research Institute: Ottawa, ON, Canada, 2014.
35.
Sidwell, K. Newcastle—Ottawa Quality Assessment Scale Case Control Studies. Available online: https://www.ohri.ca/
programs/clinical_epidemiology/nosgen.pdf (accessed on 27 June 2023).
36.
Vrijlandt, E.J.; Gerritsen, J.; Boezen, H.M.; Grevink, R.G.; Duiverman, E.J. Lung function and exercise capacity in young adults
born prematurely. Am. J. Respir. Crit. Care Med. 2006, 173, 890–896. [CrossRef]
Children 2023, 10, 1427
14 of 15
37.
Evensen, K.A.; Steinshamn, S.; Tjønna, A.E.; Stølen, T.; Høydal, M.A.; Wisløff, U.; Brubakk, A.M.; Vik, T. Effects of preterm birth
and fetal growth retardation on cardiovascular risk factors in young adulthood. Early Hum. Dev. 2009, 85, 239–245. [CrossRef]
[PubMed]
38.
Narang, I.; Bush, A.; Rosenthal, M. Gas transfer and pulmonary blood flow at rest and during exercise in adults 21 years after
preterm birth. Am. J. Respir. Crit. Care Med. 2009, 180, 339–345. [CrossRef] [PubMed]
39.
Sipola-Leppänen, M.; Hovi, P.; Andersson, S.; Wehkalampi, K.; Vääräsmäki, M.; Strang-Karlsson, S.; Järvenpää, A.L.; Mäkitie, O.;
Eriksson, J.G.; Kajantie, E. Resting energy expenditure in young adults born preterm--the Helsinki study of very low birth weight
adults. PLoS ONE 2011, 6, e17700. [CrossRef] [PubMed]
40.
Lovering, A.; Laurie, S.; Elliott, J.; Beasley, K.; Yang, X.; Gust, C.; Mangum, T.S.; Goodman, R.D.; Hawn, J.A.; Gladstone, I.M.
Normal pulmonary gas exchange efficiency and absence of exercise-induced arterial hypoxemia in adults with bronchopulmonary
dysplasia. J. Appl. Physiol. 2013, 115, 1050–1056. [CrossRef] [PubMed]
41.
Duke, J.W.; Lewandowski, A.J.; Abman, S.H.; Lovering, A.T. Physiological aspects of cardiopulmonary dysanapsis on exercise in
adults born preterm. J. Physiol. 2022, 600, 463–482. [CrossRef] [PubMed]
42.
Saarenpää, H.K.; Tikanmäki, M.; Sipola-Leppänen, M.; Hovi, P.; Wehkalampi, K.; Siltanen, M.; Vääräsmäki, M.; Järvenpää, A.L.;
Eriksson, J.G.; Andersson, S.; et al. Lung function in very low birth weight adults. Pediatrics 2015, 136, 642–650. [CrossRef]
[PubMed]
43.
Farrell, E.T.; Bates, M.L.; Pegelow, D.F.; Palta, M.; Eickhoff, J.C.; O’Brien, M.J.; Eldridge, M.W. Pulmonary gas exchange and
exercise capacity in adults born preterm. Ann. Am. Thorac. Soc. 2015, 12, 1130–1137. [CrossRef] [PubMed]
44.
Caskey, S.; Gough, A.; Rowan, S.; Gillespie, S.; Clarke, J.; Riley, M.; Megamy, J.; Nicholls, P.; Patterson, C.; Halliday, H.L.; et al.
Structural and functional lung impairment in adult survivors of bronchopulmonary dysplasia. Ann. Am. Thorac. Soc. 2016, 13,
1262–1270. [CrossRef]
45.
Kaseva, N.; Wehkalampi, K.; Strang-Karlsson, S.; Salonen, M.; Pesonen, A.K.; Räikkönen, K.; Tammelin, T.; Hovi, P.; Lahti, J.;
Heinonen, K.; et al. Lower conditioning leisure-time physical activity in young adults born preterm at very low birth weight.
PLoS ONE 2012, 7, e32430. [CrossRef]
46.
Haraldsdottir, K.; Watson, A.M.; Pegelow, D.F.; Palta, M.; Tetri, L.H.; Levin, T.; Brix, M.D.; Centanni, R.M.; Goss, K.N.; Eldridge,
M.M. Blunted cardiac output response to exercise in adolescents born preterm. Eur. J. Appl. Physiol. 2020, 120, 2547–2554.
[CrossRef]
47.
Yang, J.; Epton, M.J.; Harris, S.L.; Horwood, J.; Kingsford, R.A.; Troughton, R.; Greer, C.; Darlow, B.A. Reduced exercise capacity
in adults born at very low birth weight a population-based cohort study. Am. J. Respir. Crit. Care Med. 2022, 205, 88–98. [CrossRef]
[PubMed]
48.
Duke, J.W.; Elliott, J.E.; Laurie, S.S.; Beasley, K.M.; Mangum, T.S.; Hawn, J.A.; Gladstone, I.M.; Lovering, A.T. Pulmonary gas
exchange efficiency during exercise breathing normoxic and hypoxic gas in adults born very preterm with low diffusion capacity.
J. Appl. Physiol. 2014, 117, 381–473. [CrossRef] [PubMed]
49.
Cheong, J.L.; Olsen, J.E.; Konstan, T.; Mainzer, R.M.; Hickey, L.M.; Spittle, A.J.; Wark, J.D.; Cheung, M.M.; Garland, S.M.; Duff, J.;
et al. Growth from infancy to adulthood and associations with cardiometabolic health in individuals born extremely preterm.
Lancet Reg. Health Wet. Pac. 2023, 34, 100717. [CrossRef]
50.
Huckstep, O.J.; Burchert, H.; Williamson, W.; Telles, F.; Tan, C.M.; Bertagnolli, M.; Arnold, L.; Mohamed, A.; McCormick, K.;
Hanssen, H.; et al. Impaired myocardial reserve underlies reduced exercise capacity and heart rate recovery in preterm-born
young adults. Eur. Heart J. Cardiovasc. Imaging 2021, 22, 572–580. [CrossRef]
51.
Jarvis, D. The European Community Respiratory Health Survey II. Eur. Respir. J. 2002, 20, 1071–1079.
52.
Harris, C.; Lunt, A.; Peacock, J.; Greenough, A. Lung function at 16–19 years in males and females born very prematurely.
Pediatr. Pulmonol. 2023, 58, 2035–2041. [CrossRef]
53.
Bassett, D.R. Scientific contributions of A. V. Hill: Exercise physiology pioneer. J. Appl. Physiol. 2002, 93, 1567–1582. [CrossRef]
54.
Flegal, K.M.; Shephard, J.A.; Looker, A.C.; Graubard, B.I.; Borrud, L.G.; Ogden, C.L.; Harris, T.B.; Everhart, J.E.; Schenker, N.
Comparisons of percentage body fat, body mass index, waist circumference, and waist-stature ratio in adults. Am. J. Clin. Nutr.
2009, 89, 500–508. [CrossRef]
55.
Akindele, M.O.; Phillips, J.S.; Igumbor, E.U. The relationship between body fat percentage and body mass index in overweight
and obese individuals in an urban African setting. J. Public Health Africa 2016, 7, 515. [CrossRef]
56.
Ranasinghe, C.; Gamage, P.; Katulanda, P.; Andraweera, N.; Thilakarathne, S.; Tharanga, P. Relationship between body mass
index (bmi) and body fat percentage, estimated by bioelectrical impedance, in a group of Sri Lankan adults: A cross sectional
study. BMC Public Health 2013, 13, 797. [CrossRef]
57.
Shete, A.N.; Bute, S.S.; Deshmukh, P.R. A study of VO2 max and body fat percentage in female athletes. J. Clin. Diagn. Res. 2014,
8, bc01–bc03. [PubMed]
58.
Goran, M.I.; Fields, D.A.; Hunter, G.R.; Herd, S.L.; Weinsier, R.L. Total body fat does not influence maximal aerobic capacity. Int. J.
Obes. Relat. Metab. Disord. 2000, 24, 841–848. [CrossRef] [PubMed]
59.
Kyle, U.G.; Genton, L.; Hans, D.; Karsegard, L.; Slosman, D.O.; Pichard, C. Age-related differences in fat-free mass, skeletal
muscle, body cell mass and fat mass between 18 and 94 years. Eur. J. Clin. Nutr. 2001, 55, 663–672. [CrossRef] [PubMed]
60.
Scribbans, T.D.; Vecsey, S.; Hankinson, P.B.; Foster, W.S.; Gurd, B.J. The effect of training intensity on VO2max in young healthy
adults: A meta-regression and meta-analysis. Int. J. Exerc. Sci. 2016, 9, 230–247. [PubMed]
Children 2023, 10, 1427
15 of 15
61.
Crowley, E.; Powell, C.; Carson, B.P.; Davies, R.W. The effect of exercise training intensity on VO2 max in healthy adults: An
overview of systematic reviews and meta-analyses. Trans. Sports Med. 2022, 2022, 9310710.
62.
Duke, J.W.; Lovering, A.T. Respiratory and cardiopulmonary limitations to aerobic exercise capacity in adults born preterm.
J. Appl. Physiol. 2020, 129, 718–724. [CrossRef] [PubMed]
63.
Duke, J.W.; Gladstone, I.M.; Sheel, A.W.; Lovering, A.T. Premature birth affects the degree of airway dysanapsis and mechanical
ventilatory constraints. Exp. Physiol. 2018, 103, 261–275. [CrossRef] [PubMed]
64.
Goss, K.N.; Beshish, A.G.; Barton, G.P.; Haraldsdottir, K.; Levin, T.S.; Tetri, L.H.; Battiola, T.J.; Mulchrone, A.M.; Pegelow, D.F.;
Palta, M.; et al. Early pulmonary vascular disease in young adults born preterm. Am. J. Respir. Crit. Care Med. 2018, 198,
1549–1558. [CrossRef]
65.
Laurie, S.S.; Elliott, J.E.; Beasley, K.M.; Mangum, T.S.; Goodman, R.D.; Duke, J.W.; Gladstone, I.M.; Lovering, A.T. Exaggerated
increase in pulmonary artery pressure during exercise in adults born preterm. Am. J. Respir. Crit. Care Med. 2018, 197, 821–823.
[CrossRef]
66.
Parkinson, J.R.; Hyde, M.J.; Gale, C.; Santhakumaran, S.; Modi, N. Preterm birth and the metabolic syndrome in adult life: A
systematic review and meta-analysis. Pediatrics 2013, 131, e1240–e1263. [CrossRef]
67.
Carr, H.; Cnattingius, S.; Granath, F.; Ludvigsson, J.F.; Edstedt Bonamy, A.K. Preterm birth and risk of heart failure up to early
adulthood. J. Am. Coll. Cardiol. 2017, 69, 2634–2642. [CrossRef] [PubMed]
68.
Li, S.; Zhang, M.; Tian, H.; Liu, Z.; Yin, X.; Xi, B. Preterm birth and risk of type 1 and type 2 diabetes: Systematic review and
meta-analysis. Obes. Rev. 2014, 15, 804–811. [CrossRef] [PubMed]
69.
Hippisley-Cox, J.; Coupland, C.; Vinogradova, Y.; Robson, J.; Brindle, P. Performance of the QRISK cardiovascular risk prediction
algorithm in an independent UK sample of patients from general practice: A validation study. Heart 2008, 94, 34–39. [CrossRef]
[PubMed]
70.
Burchert, H.; Lapidaire, W.; Williamson, W.; McCourt, A.; Dockerill, C.; Woodward, W.; Tan, C.M.J.; Bertagnolli, M.; Mohamed,
A.; Alsharqi, M.; et al. Aerobic exercise training response in preterm born young adults with elevated blood pressure and stage 1
hypertension: A randomised clinical trial. Am. J. Respir. Crit. Care Med. 2023, 207, 1227–1236. [CrossRef] [PubMed]
71.
Abrantes, C.; Sampaio, J.; Reis, V.; Sousa, N.; Duarte, J. Physiological responses to treadmill and cycle exercise. Int. J. Sports Med.
2012, 33, 26–30. [CrossRef]
72.
Hermansen, L.; Saltin, B. Oxygen uptake during maximal treadmill and bicycle exercise. J. Appl. Physiol. 1969, 26, 31–37.
[CrossRef] [PubMed]
73.
Robinson, S. Experimental studies of physical fitness in relation to age. Arbeitsphysiologie 1938, 10, 251–323. [CrossRef]
74.
Astrand, I. Aerobic work capacity in men and women with special reference to age. Acta Physiol. Scand Suppl. 1960, 49, 1–92.
75.
Hawkins, S.; Wiswell, R. Rate and mechanism of maximal oxygen consumption decline with aging: Implications for exercise
training. Sports Med. 2003, 33, 877–888. [CrossRef]
76.
Hoover, J.C.; Alenazi, A.M.; Alothman, S.; Alshehri, M.M.; Rucker, J.; Kluding, P. Recruitment for exercise or physical activity
interventions: A protocol for systematic review. BMJ Open 2018, 8, e019546. [CrossRef]
Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual
author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to
people or property resulting from any ideas, methods, instructions or products referred to in the content.
| Exercise Capacity in Very Low Birth Weight Adults: A Systematic Review and Meta-Analysis. | 08-21-2023 | Poole, Grace,Harris, Christopher,Greenough, Anne | eng |
PMC9206639 | 1
Vol.:(0123456789)
Scientific Reports | (2022) 12:10295
| https://doi.org/10.1038/s41598-022-13844-4
www.nature.com/scientificreports
Health status of recreational
runners over 10‑km
up to ultra‑marathon distance
based on data of the NURMI Study
Step 2
Katharina Wirnitzer1,2,3, Patrick Boldt4, Gerold Wirnitzer5, Claus Leitzmann6,
Derrick Tanous1,2, Mohamad Motevalli1,2, Thomas Rosemann7 & Beat Knechtle7,8*
Endurance running is well‑documented to affect health beneficially. However, data are still conflicting
in terms of which race distance is associated with the maximum health effects to be obtained.
Therefore, the aim of this study was to compare the health status of endurance runners over different
race distances. A total of 245 recreational runners (141 females, 104 males) completed an online
survey. Health status was assessed by measuring eight dimensions in two clusters of health‑related
indicators (e.g., body weight, mental health, chronic diseases and hypersensitivity reactions,
medication intake) and health‑related behaviors (e.g., smoking habits, supplement intake, food
choice, healthcare utilization). Each dimension consisted of analytical parameters derived to a general
domain score between 0 and 1. Data analysis was performed by using non‑parametric ANOVA and
MANOVA. There were 89 half‑marathon (HM), 65 marathon/ultra‑marathon (M/UM), and 91 10‑km
runners. 10‑km runners were leaner than both the HM and M/UM runners (p ≤ 0.05). HM runners
had higher health scores for six dimensions (body weight, mental health, chronic diseases and
hypersensitivity reactions, medication intake, smoking habits, and health care utilization), which
contributed to an average score of 77.1% (score range 62–88%) for their overall state of health.
Whereas 10‑km and M/UM runners had lesser but similar average scores in the overall state of health
(71.7% and 72%, respectively). Race distance had a significant association with the dimension
“chronic diseases and hypersensitivity reactions” (p ≤ 0.05). Despite the null significant associations
between race distance and seven (out of eight) multi‑item health dimensions, a tendency towards
better health status (assessed by domain scores of health) among HM runners was found compared to
other distance runners. However, the optimal state of health across all race distances supported the
notion that endurance running contributed to overall health and well‑being.
Trial registration number: ISRCTN73074080. Retrospectively registered 12th June 2015.
As the basic form of human movement, running is the most popular leisure-time physical activity1. This low-cost
and convenient activity can be practiced at any age with little effort and a lower level of expertise and mastery2.
Over the past decades, the number of recreational and professional runners has increased across various dis-
tances, marathons in particular3, and various reasons for actively following a running routine have been reported
by runners. While health-oriented purposes have been shown to be the most significant motive for running1,4,
literature indicates that several motives including but not limited to leisure, hobby, weight control, winning,
OPEN
1Department of Research and Development in Teacher Education, University College of Teacher Education Tyrol,
Innsbruck, Austria. 2Department of Sport Science, University of Innsbruck, Innsbruck, Austria. 3Research Center
Medical Humanities, Leopold-Franzens University of Innsbruck, Innsbruck, Austria. 4Department of Child and
Adolescent Psychiatry and Psychotherapy, LVR-Klinik Viersen, Viersen, Germany. 5adventureV & change2V, Stans,
Austria. 6Institute of Nutrition, University of Gießen, Gießen, Germany. 7Institute of Primary Care, University of
Zurich, Zurich, Switzerland. 8Medbase St. Gallen Am Vadianplatz, St. Gallen, Switzerland. *email: beat.knechtle@
hispeed.ch
2
Vol:.(1234567890)
Scientific Reports | (2022) 12:10295 |
https://doi.org/10.1038/s41598-022-13844-4
www.nature.com/scientificreports/
and social reasons encourage runners to engage in running activities/events3,5. Motivations for running could
potentially influence the intensity, duration, and frequency of training routines as well as lifestyle behaviors
in endurance runners, which together might affect short- and long-term health status4,5. Despite the fact that
distance runners are depicted as the healthiest fraction of the general population, it has been reported that a
“faster and further” dosage fails6.
Research indicated that among 26 different kinds of sport, endurance running provides the most favorable
health implications9. Regular participation in recreational running was found to positively affect body weight
(BW), body fat, blood pressure, blood glucose levels, insulin sensitivity, blood-lipid profile, and musculoskeletal
health10–12. Additionally, running could favorably influence mood, well-being, and mental status13,14. Other
mental feelings, including fear, depression, worries, anxiety, and anger within the context of an adjustment
disorder, might be positively affected following regular endurance running14. Distance running contributes to
the prevention of chronic diseases by lowering the risks, such as cardiovascular disease (e.g., coronary artery
disease, stroke)15,16 and different types of cancer17,18. As a potential link between running and overall mortality,
cardiorespiratory fitness is a strong predictor for morbidity and mortality, and further reduces total mortality
from cardiovascular disease, cancer, infections, and other causes15,19. Substantial health-related advantages fol-
lowing endurance running are correlated with running exposure in a dose–response association, as the larger
effects on health are achieved with increased loads of running8. Moreover, evidence supports more beneficial
health effects of regular endurance running on cardiovascular risk factors, particularly artery carotid diameter
thickness20 and low-grade inflammation21 compared to irregular endurance running. In addition to the well-
established fact that endurance running is an effective tool to improve individual health7, regular and long-term
involvement in running activities could be a powerful tool to affect public health positively and thus tackle global
health problems8,9.
It has been shown that marathon runners benefit from a greater metabolic fitness (e.g., insulin response, fast-
ing lipids, fasting insulin), aerobic performance (e.g., velocity at VO2max, running economy), exercise metabo-
lism (e.g., lactate threshold), and skeletal muscle levels of mitochondrial proteins compared to sedentary subjects
with matched cardiovascular fitness, age, gender, and body mass index (BMI)22. Marathon running was found to
significantly diminish the risk of coronary plaque prevalence as a result of reducing the relevant risk factors (e.g.,
hypertension and hyperlipidemia)23. In addition to a low incidence of cardiovascular disease, marathoners are
shown to have an extended longevity compared to the general population24. The favorable health consequences
of distance running are not limited to marathoners, as distance runners in lower and higher mileages were also
shown to have comparable outcomes. Evidence indicates that ultra-marathoners were healthier and less often
sick compared to the general population25. Half marathon running was found to positively affect immune cell
proportions, pro-inflammatory cytokine levels, and recovery behavior on a short-term basis as a midterm anti-
inflammatory effect26. Research on 10-km running demonstrates a positive relationship between running and
cardio-metabolic health, independent of exercise volume and cardiorespiratory fitness27. Furthermore, compared
to longer-distance runners, 10-km runners also appear to be at a lower risk of injuries; however, weekly mileage
and race distance were identified as risk factors for injuries in endurance runners7,28.
Despite the aforementioned advantageous influences, there have been reports of some adverse effects of
distance running on health status (e.g., musculoskeletal injuries, unintended weight reduction, cardiovascular
abnormalities) that potentially increase with age28–30. The increased exercise-induced stress during an ultra-
marathon run leads to several pathophysiological changes, such as an increase in acute phase proteins, a decrease
in testosterone, an increase in liver values, hemolysis, skeletal muscle cell damage, micro-hematuria, and a loss of
bone mass31. Ultra-marathoners also tend to suffer from more knee pain, stress fractures, allergies, and asthma
than the general population25. In addition, intensive and long-lasting endurance running was found to lead
arterial changes toward constricting the coronary, cerebral, and peripheral arteries1,32,33, which not only affects
performance but could also be associated with an increased risk for acute cardiac disorders (e.g., cardiac death,
clinical arrhythmias, angina, myocardial infarcts)34,35. While older individuals are at a higher risk36, the incidence
of race-related cardiac arrest was found to be significantly higher in males than female marathoners—although
the overall risk is low34. It was found that a half marathon running could also significantly increase post-exercise
levels of biomarkers related to cardiovascular damage and dysfunction37, which is associated with an increased
risk for race-related cardiac arrest34. Moreover, activation of the inflammatory response and the detoxification
process was shown by proteomic profile changes after a half marathon race, and additional pathways associated
with immune response, lipid transport, and coagulation were involved38. Distance running is also associated
with a high risk of running-induced injuries, as approximately half of the active runners reported having more
than one injury per year, with excess BW, the weekly mileage, and the race distance recognized as relevant
risk factors7,28. Furthermore, gastrointestinal complaints (due to decreased exercise-induced mesenteric blood
flow)39, symptomatic hyponatremia35, and exercise-induced asthma, as well as hay fever, are reported in distance
runners39,40.
In spite of the well-recognized effects of endurance running on different health parameters, there is a paucity
of research comparing the health status among different groups of endurance runners. The available health-
associated reports did not distinguish different race distances and instead have focused on 10-km runners41,
half marathoners26,37,38, marathoners12,22,24, or ultra-marathoners31,32. Therefore, the aim of the present study was
to investigate the health-related indicators and behaviors of recreational endurance runners and compare their
health status across different race distances. It was hypothesized that the health status differs between endurance
runners over 10-km, half-marathon, and marathon/ultra-marathon race distances.
3
Vol.:(0123456789)
Scientific Reports | (2022) 12:10295 |
https://doi.org/10.1038/s41598-022-13844-4
www.nature.com/scientificreports/
Methods
Study design and ethical approval.
The present study is a part of the NURMI (Nutrition and Run-
ning High Mileage) Study and has been conducted following a cross-sectional design42. The NURMI study was
designed by an interdisciplinary team of scientists and aims to assess and compare recreational endurance run-
ners by sex, race distance, diet type, etc. Data collection was conducted via a series of self-reported online sur-
veys in three separate but subsequent steps. The NURMI Study Step 1 will therefore examine epidemiological
aspects (e.g., age, sex, and prevalence of diet type at running events), Step 2 focuses on behaviors considering
running training and racing, nutrition, health, etc., and Step 3 investigates running performance linked to diet
and sports-psychological parameters.
The subsequent method was introduced in detail elsewhere10,42,43, to which the interested readers are kindly
referred. The study protocol was approved by the ethics board of St. Gallen, Switzerland, on May 6, 2015 (EKSG
14/145). The trial registration number is ISRCTN73074080.
Experimental approach and inclusion criteria.
Endurance runners in the NURMI study were mostly
engaged from German-speaking countries, including Germany, Austria, and Switzerland. Runners were con-
tacted and recruited mainly via social media, websites of the organizers of marathon events, online running
communities, email-lists and runners’ magazines, as well as via magazines for health, nutrition and lifestyle,
trade fairs on sports, plant-based nutrition and lifestyle, as well as through personal contacts.
Participants completed an online survey within the NURMI Study Step 2, which was available in German and
English at www. nurmi- study. com. Prior to completion of the questionnaire, participants were provided a writ-
ten description of the procedures and gave their informed consent to take part in the study. In parallel, physical
and psychological information—including the assignment to one of three basic areas of sports (as participants
are mainly active in running due to either health, leisure, or performance foci)—motivation and aim of run-
ning activities, and details regarding other sports activities to balance for running were obtained to differentiate
between a health, leisure, or predominantly performance-orientated approach. For successful participation in
the study, the following inclusion criteria were determined initially: (1) written informed consent; (2) at least
18 years of age; (3) questionnaire Step 2 completed; (4) having a BMI < 30 kg/m2; and (5) successful participation
in a running event of at least a half-marathon distance in the past two years. However, to avoid an irreversible
loss of valuable data sets, those who met the inclusion criteria 1–4 but stated being 10-km runners were included
as additional participants and were assigned to a further race distance group.
To control for a minimal status of health linked to a minimum level of fitness and to further enhance the
reliability of data sets, BMI-associated criteria were implemented in the present study. With a BMI ≥ 30 kg/m2,
however, other health-protective and/or weight loss strategies other than running are necessary to reduce body
weight safely, and could thus potentially affect health-related data. Therefore, participants with a BMI ≥ 30 kg/
m2 (n = 3) were excluded from data analysis.
Data clearance and classification of participants.
Control questions were included throughout differ-
ent sections of the survey to control for self-reported information of running-related variables (history, training,
racing, etc.), and consequently, to identify inconsistent or conflicting data. In general, from the initial number
of 317 endurance runners, 72 participants who did not meet the inclusion criteria or did not provide consistent
or complete answers to essential questions (e.g., sex, age, race distance, health-related questions) were excluded
from the study. As a result, a total of 245 runners with complete data sets were included for descriptive statistical
analysis after data clearance (Fig. 1).
Participants were initially categorized according to race distance: half-marathon and marathon/ultra-mara-
thon (data were pooled since the marathon distance is included in an ultra-marathon). The shortest distance for
ultra-marathon was 50 km, and the longest distance was 160 km in the present study. In addition, a total of 91
highly-motivated 10-km runners provided accurate and complete answers; however, they had not successfully
participated in either a half-marathon or marathon. In general, the most frequently stated race distance was
considered the main criterion to assign runners to the respective study groups.
It is well-established that the BMI of active runners is lower than the general population44, and people with a
higher BMI might have a different health status, as their main goal to engage in running activities is to achieve
and maintain a healthy BW. The World Health Organization45,46 recommends maintaining a BMI in the range of
18.5–24.9 kg/m2 (BMINORM) for individuals, while at the same time pointing to an increased risk of co-morbidities
for a BMI 25.0–29.9 kg/m2 and moderate to severe risk of co-morbidities for a BMI > 30 kg/m2. Therefore, calcu-
lated BMI was classified into three categories, under 18.49, BMINORM, and over 25, to differentiate health-related
findings based on BMI subgroups. In addition, given the importance of diet types in endurance runners’ health
status10,20, participants were assigned into three dietary subgroups of omnivores, vegetarians, and vegans47.
Health‑related dimensions.
As a latent variable, health status was derived by using both the two clusters
of health-related indicators and health-related behaviors10,48. Each cluster pooled four dimensions defined by
specific items based on manifest measures. The following dimensions described health-related indicators: (1)
BW and BMI; (2) mental health (stress perception); (3) chronic diseases and hypersensitivity reactions: preva-
lence of chronic diseases (incl. heart disease, state after heart attack, cancer), prevalence of metabolic diseases
(incl. diabetes mellitus 1, diabetes mellitus 2, hyperthyroidism, hypothyroidism), prevalence of hypersensitiv-
ity reactions (incl. allergies, intolerances); and (4) medication intake (for thyroid disease, for hypertension, for
cholesterol level, for contraception). The following dimensions described health-related behaviors: (1) smoking
habits (current and history of smoking); (2) supplement intake (supplements prescribed by a doctor, supple-
ments for performance enhancement, supplements to cope with stress); (3) food choice (motivation, desired
4
Vol:.(1234567890)
Scientific Reports | (2022) 12:10295 |
https://doi.org/10.1038/s41598-022-13844-4
www.nature.com/scientificreports/
ingredients, avoided ingredients); and (4) healthcare utilization and regular check-ups. Together, these eight
dimensions described health outcomes. Resulting from this, eight domain scores were derived, which generated
scores between 0 and 1. Low scores indicate detrimental health associations, while higher scores indicate benefi-
cial health associations [given as mean scores plus standard deviation and percentage (%)].
Statistical analysis.
The statistical software R version 3.5.0 Core Team 2018 (R Foundation for Statistical
Computing, Vienna, Austria) was used to perform all statistical analyses. Exploratory analysis was performed
by descriptive statistics (median and interquartile range (IQR)). Significant differences between race distance
subgroups and domain scores to describe health status were calculated by using a non-parametric ANOVA.
Chi-square test and Kruskal–Wallis test were used to examine the association between race distance subgroups
and domain scores with nominal scale variables, and Wilcoxon test and Kruskal–Wallis test (ordinal and metric
scale) approximated by using the F distributions. State of health was statistically modeled as a latent variable and
was derived by manifest variables (e.g., BW, cancer, smoking). In order to scale the state of health described by
the respective dimensions of health, a heuristic index between 0 and 1 was defined (equivalence in all items).
In order to test the statistical hypothesis considering significant differences between subgroups of race distance,
sex, age, academic qualification, and weekly mileage of running for each dimension, a MANOVA was performed
Figure 1. Enrollment and categorization of participants.
5
Vol.:(0123456789)
Scientific Reports | (2022) 12:10295 |
https://doi.org/10.1038/s41598-022-13844-4
www.nature.com/scientificreports/
to define health status. The assumptions of the ANOVA were verified by residual analysis. The level of statistical
significance was set at p < 0.05 (statistical trend: 0.05 ≥ p < 0.10).
Ethics approval.
The study protocol was approved by the ethics board of St. Gallen, Switzerland on May
6, 2015 (EKSG 14/145). The study was conducted in accordance with the ethical standards of the institutional
review board, medical professional codex and the with the 1964 Helsinki declaration and its later amendments
as of 1996 as well as Data Security Laws and good clinical practice guidelines.
Consent to participate.
All participants gave written informed consent prior to the testing procedure.
Results
Sociodemographic data.
A total of 245 endurance runners (141 women and 104 men) with a mean age of
39 (IQR 17) years and a BMI of 21.72 (IQR 3.50) kg/m2 were included for final data analysis. Germany (n = 177),
Austria (n = 44), and Switzerland (n = 13) had the majority of endurance runners, but 4.5% of participants
(n = 11) were from other countries, including Belgium, Brazil, Canada, Italy, Luxemburg, Netherlands, Poland,
Spain, and the UK. There were 154 NURMI-Runners (89 half-marathoners, 65 marathoners/ultra-marathoners)
and 91 runners over the 10-km distance. The participants reported following an omnivorous diet (44%), vegetar-
ian diet (18%), or vegan diet (37%). Moreover, with regard to the level of academic qualification, 34% of endur-
ance runners (n = 83) had upper secondary/technical education or a university (or higher) degree. In addition,
67% of endurance runners were married or living with partner (Table 1). The characteristics of the subjects are
presented in Tables 1 and 2.
The basic assignment of endurance runners to sports areas was 54% for leisure activity, 36% for sports
achievement, and 10% for health concerns. The main motivation of endurance runners to start running was for
hobby (35%), health (19%), or BW loss (18%). The major goal for participation in running events reported was
to achieve a specific runtime (51%) followed by the pleasure of running (39%). As a supplementary physical
activity, summer sports (53% cycling, 31% respectively swimming, hiking/rambling and trail/uphill running)
were reported to be more prevalent than winter sports.
Table 1. Anthropometric and sociodemographic characteristics of the endurance runners. Data are presented
as “percentage of prevalence (n)” or “median (IQR)”. BMI body mass index, BW body weight, HM half-
marathon, IQR interquartile range, km kilometers, M/UM marathon/ultra-marathon.
Total
HM
M/UM
10 km
Number of Subjects
100% (245)
36% (89)
27% (65)
37% (91)
Sex
Female
58% (141)
55% (49)
38% (25)
74% (67)
Male
42% (104)
45% (40)
62% (40)
26% (24)
Age (years) (median)
39 (IQR 17)
37 (IQR 18)
44 (IQR 17)
37 (IQR 18)
BW (kg) (median)
65.0 (IQR 14.2)
65.0 (IQR 13.0)
67.5 (IQR 17.5)
62 (IQR 11.0)
BMI (kg/m2) (median)
21.72 (IQR 3.50)
21.97 (IQR 3.28)
22.15 (IQR 3.25)
21.30 (IQR 3.94)
Diet
Omnivorous
44% (109)
44% (39)
51% (33)
41% (37)
Vegetarian
18% (45)
22% (20)
15% (10)
16% (15)
Vegan
37% (91)
34% (30)
34% (22)
43% (39)
Academic qualification
No Qualification
< 1% (1)
1% (1)
–
–
Upper Secondary Education/Technical Qualification/GCSE
or Equivalent
34% (83)
37% (33)
40% (26)
26% (24)
A Levels or Equivalent
22% (53)
17% (15)
23% (15)
25% (23)
University Degree/Higher Degree (i.e., doctorate)
34% (83)
30% (27)
28% (18)
42% (38)
No Answer
10% (25)
15% (13)
9% (6)
7% (6)
Marital status
Divorced/Separated
6% (15)
6% (5)
6% (4)
7% (6)
Married/Living with Partner
67% (164)
63% (56)
72% (47)
67% (61)
Single
27% (66)
31% (28)
22% (14)
26% (24)
Country of residence
Austria
18% (44)
17% (15)
20% (13)
18% (16)
Germany
72% (177)
73% (65)
69% (45)
74% (67)
Switzerland
5% (13)
7% (6)
8% (5)
2% (2)
Other
4% (11)
3% (3)
3% (2)
7% (6)
6
Vol:.(1234567890)
Scientific Reports | (2022) 12:10295 |
https://doi.org/10.1038/s41598-022-13844-4
www.nature.com/scientificreports/
The median number of events completed in our sample was eight races, and the marathoners/ultra-maratho-
ners finished the highest number of races. Depending on the stage of preparation for the main event and/or sea-
son within the course of the year, 70% of runners reported their weekly mileage at a medium volume (19–36 km),
while 17% and 13% of runners reported low (< 19 km) and high (> 36 km) volumes, respectively (Table 2).
Health‑related indicators.
Dimension of BW and BMI. There was a significant difference in BW between
race distance subgroups (F(2, 242) = 5.05, p = 0.007), with 10-km runners weighing less than half-marathoners and
Table 2. Characteristics of running activity of the subjects. Data are presented as “percentage of prevalence
(n)” or “median (IQR)”. HM half-marathon, IQR interquartile range, km kilometers, M/UM marathon/ultra-
marathon. a Sport for health: Those who take part in sports for health reasons and train 2–3 times a week
for 30–60 min at a low to moderate intensity with the aim of maintaining or improving their health. b Sport
for leisure: Those who take part for leisure reasons and train 2–5 times a week for 60–90 min at a moderate
intensity with the aim of enjoying their free time actively. c Sport for performance: Performance athletes train
3–6 times a week, at moderate to high intensities and assiduously follow a long-term training plan, including
assessing their performance, with the aim of ascertaining and improving it and measuring it against that of
other athletes in competitions.
Total
HM
M/UM
10 km
Number of subjects
100% (245)
36% (89)
27% (65)
37% (91)
Basic assignment to areas of sport
Sport for Healtha
10% (23)
8% (7)
5% (3)
14% (13)
Sport for Leisureb
54% (133)
64% (57)
37% (24)
57% (52)
Sport for Performancec
36% (89)
28% (25)
58% (38)
29% (26)
Motive for running
Initial Motivation for Running
Counteraction to Job
9% (22)
10% (9)
11% (7)
7% (6)
Leisure Activity
4% (11)
7% (6)
5% (3)
2% (2)
Hobby
35% (85)
33% (29)
38% (25)
34% (31)
Weight Maintenance
7% (17)
9% (8)
6% (4)
5% (5)
Weight Loss
18% (45)
17% (15)
15% (10)
22% (20)
Health
19% (46)
19% (17)
18% (12)
19% (17)
Other
8% (19)
6% (5)
6% (4)
11% (10)
Aim for running events
For the Pleasure of Running
39% (90)
40% (35)
47% (27)
32% (28)
Specific Placing
3% (8)
2% (2)
5% (3)
3% (3)
Specific Time
51% (117)
51% (44)
44% (25)
55% (48)
Taking Part is All that Matters
7% (16)
7% (6)
4% (2)
9% (8)
Completion of running events
Total Races Completed Before the NURMI Study (median)
8 (IQR 11)
7 (IQR 11)
10 (IQR 10)
7 (IQR 12)
Races Completed in the Past 2 Years Over Distances (median)
8 (IQR 11)
6 (IQR 11)
10 (IQR 11)
7 (IQR 11)
Half-Marathon
2 (IQR 3)
3 (IQR 4)
2 (IQR 3)
1 (IQR 2)
Marathon/Ultra-Marathon
1 (IQR 2)
0 (IQR 1)
2 (IQR 3)
0 (IQR 1)
Running Training per week (Mean mileage, km)
Low Mileage (≤ 1 km)
17% (41)
26% (23)
5% (3)
16% (5)
Medium Mileage (> 19–36 km)
70% (172)
65% (58)
63% (41)
80% (73)
High Mileage (> 36–100 km)
13% (32)
9% (8)
32% (21)
3% (3)
Other sports to balance for running
Summer Sports
Cycling
53% (130)
55% (49)
57% (36)
49% (45)
Swimming
31% (75)
35% (31)
22% (14)
33% (30)
Hiking/Rambling
31% (75)
33% (29)
32% (20)
29% (26)
Trail/Uphill Running
31% (75)
33% (29)
46% (29)
19% (17)
Triathlon
19% (46)
21% (19)
17% (11)
18% (16)
Winter Sports
Skiing (alpine)
14% (34)
15% (13)
16% (10)
12% (11)
Cross Country Skiing
11% (26)
12% (11)
13% (8)
8% (7)
Snowboarding
7% (16)
9% (8)
5% (3)
5% (5)
Ski Touring
4% (9)
7% (6)
5% (3)
–
7
Vol.:(0123456789)
Scientific Reports | (2022) 12:10295 |
https://doi.org/10.1038/s41598-022-13844-4
www.nature.com/scientificreports/
marathoners/ultra-marathoners. However, there was no difference in the health-related item BMI between the
subgroups (χ2
(4) = 1.35, p = 0.854) (Table 3). In addition, 10-km runners showed the lowest calculated BMI, while
half-marathoners contributed the largest fraction of BMINORM (85%). Although no significant between-group
difference was observed in the dimension of “BW and BMI” (F(2, 242) = 0.84, p = 0.433), comparative data showed
that half-marathoners had the highest score for the health-related indicator “BW and BMI” (0.69 ± 0.39), and
were followed closely by marathon/ultra-marathon runners (0.67 ± 0.39) (Table 4).
Dimension of mental health. There was no significant association between race distance and mental health
(χ2
(2) = 5.83, p = 0.054) (Table 3). However, half-marathoners reported least often to suffer from perceived stress
(27%, n = 23). Although no significant between-group difference was observed in the dimension of “mental
health” (F(2, 219) = 2.95, p = 0.054), comparative data showed that half-marathoners had the highest score with
regard to mental health (0.73 ± 0.45) (Table 4).
Dimension of chronic diseases and hypersensitivity reactions. There was no significant association between
race distance and the prevalence of (1) cardiovascular diseases and cancer (χ2
(4) = 4.76, p = 0.313), (2) metabolic
diseases (χ2
(10) = 13.25, p = 0.210), and (3) hypersensitivity reactions (χ2
(4) = 8.90, p = 0.064). However, none of
the half-marathoners reported having chronic diseases, and half-marathoners most often reported having no
metabolic diseases (92%, n = 78) and no hypersensitivity reactions (73%, n = 62) while having allergies the least
often (22%, n = 19), (Table 3). Overall, half-marathoners scored highest significantly with regard to the health-
related indicator chronic diseases and hypersensitivity reactions, and it was the only dimension with significant
between-group differences (0.88 ± 0.18, F(2, 219) = 3.31, p = 0.038) (Table 4).
Dimension of medication intake. There was no significant association between medication intake and race dis-
tance (χ2
(6) = 2.64, p = 0.852). Furthermore, there was no significant association between race distance and the
intake of contraceptives (χ2
(2) = 5.93, p = 0.051) (Table 3). However, half-marathoners most often reported hav-
ing no regular medication intake (87%, n = 74). Although no significant between-group difference was observed
in the dimension of “medication intake” (F(2, 219) = 0.20, p = 0.817), comparative data showed that half-maratho-
ners had the highest score with regard to medication intake (0.87 ± 0.34) but were closely followed by two other
groups (Table 4).
Health‑related behaviors.
Dimension of smoking habits. Race distance and current or former smoking
were not significantly associated (χ2
(4) = 4.00, p = 0.406) (Table 3). In addition, half-marathoners showed the
highest fraction of non-smokers (67%, n = 57). Although no significant between-group difference was observed
in the dimension of “smoking habits” (F(2, 219) = 2.00, p = 0.138), comparative data showed that half-marathoners
showed the best health-related behaviors with regard to smoking habits (0.83 ± 0.25) (Table 4).
Dimension of supplement intake. There was no significant association between race distance and (1) supple-
ment intake prescribed by a doctor (χ2
(2) = 0.07, p = 0.968), (2) the consumption of performance-enhancing
substances (χ2
(4) = 3.52, p = 0.476), or (3) the intake of substances to cope with stress (χ2
(4) = 6.66, p = 0.155)
(Table 3). Although no significant between-group difference was observed in the dimension of “supplement
intake” (F(2, 219) = 0.92, p = 0.400), comparative data showed that 10-km runners had the highest health scores
with regard to supplement intake (0.92 ± 0.17) but were closely followed by two other groups (Table 4).
Dimension of food choice. There was no significant association between race distance and motives for food
choice (1) because it is healthy (χ2
(2) = 0.74, p = 0.690), health-promoting (χ2
(2) = 1.00, p = 0.607), and good for
maintaining health (χ2
(2) = 2.15, p = 0.341); (2) in order to obtain vitamins (χ2
(2) = 0.15, p = 0.928), minerals/trace
elements (χ2
(2) = 0.10, p = 0.953), antioxidants (χ2
(2) = 1.06, p = 0.587), phytochemicals (χ2
(2) = 2.92, p = 0.232),
and fiber (χ2
(2) = 2.58, p = 0.276); or (3) with regard to the avoidance of the following ingredients (Table 3):
refined sugar (χ2
(2) = 1.89, p = 0.390), sweeteners (χ2
(2) = 5.63, p = 0.060), fat in general (χ2
(2) = 3.13, p = 0.210),
saturated fats (χ2
(2) = 0.21, p = 0.899), cholesterol (χ2
(2) = 0.46, p = 0.794), alcohol (χ2
(2) = 1.22, p = 0.542), and caf-
feine (χ2
(2) = 3.04, p = 0.219). However, there was a significant association between race distance and food choice
with regard to the avoidance of the following ingredients (Table 3): white flour (χ2
(2) = 8.70, p = 0.013), sweets
(χ2
(2) = 6.29, p = 0.043), and nibbles (χ2
(2) = 6.11, p = 0.047), with 10-km runners reporting doing so more often
(all three food items) than the other distance runners. Although no significant between-group difference was
observed in the dimension of “food choice” (F(2, 219) = 1.32, p = 0.270), comparative data showed that 10-km run-
ners had the best health-related behaviors with regard to food choice (0.72 ± 0.20) (Table 4).
Dimension of healthcare utilization. There was no significant association between the use of regular health
check-ups and race distance (χ2
(2) = 2.64, p = 0.268) (Table 3). Although no significant between-group difference
was observed in the dimension of “healthcare utilization” (F(2, 219) = 1.32, p = 0.270), comparative data showed
that half-marathoners had the highest scores with regard to healthcare utilization (0.62 ± 0.49) while maratho-
ners/ultra-marathoners scored lowest (0.49 ± 0.50) (Table 4).
Results of the MANOVA.
The findings of the MANOVA considering the health status of endurance run-
ners are presented in Table 5, indicating significant differences for the following results: (1) education (academic
qualification) had an association with BW and BMI (p = 0.004), smoking habits (p = 0.005), and supplement
intake (p = 0.022); (2) race distance had a significant association with the dimension “chronic diseases and hyper-
8
Vol:.(1234567890)
Scientific Reports | (2022) 12:10295 |
https://doi.org/10.1038/s41598-022-13844-4
www.nature.com/scientificreports/
Cluster and respective Dimensions
HM
M/UM
10 km
Statistics
‘Health-Related Indicators’
BW and BMI
BW (kg) (median)
65.0 (IQR 13.0)
67.5 (IQR 17.5)
62 (IQR 11.0)
F(2, 242) = 5.05, p = 0.007
BMI (median)
21.97 (IQR 3.28)
22.15 (IQR 3.25)
21.30 (IQR 3.94)
χ2
(4) = 1.35, p = 0.854
≤ 18.49
4% (4)
6% (4)
8% (7)
18.50–24.99
85% (76)
82% (53)
79% (72)
≥ 25–29.99
10% (9)
12% (8)
13% (12)
Mental health
χ2
(2) = 5.83, p = 0.054
Stress Perception
Yes
27% (23)
42% (23)
44% (36)
No
73% (62)
58% (32)
56% (46)
Chronic diseases/hypersensitivity reactions
Prevalence of Chronic Diseases
χ2
(4) = 4.76, p = 0.313
Heart Disease
–
2% (1)
–
Heart Attack
–
–
–
Cancer
–
–
1% (1)
No Diseases
100% (85)
98% (54)
99% (81)
Prevalence of Metabolic Diseases
χ2
(10) = 13.25, p = 0.210
Diabetes Mellitus 1
–
4% (2)
–
Diabetes Mellitus 2
1% (1)
–
1% (1)
Hyperthyroidism
–
2% (1)
2% (2)
Hypothyroidism
7% (6)
7% (4)
4% (3)
Other Diseases
–
–
2% (2)
No Diseases
92% (78)
87% (48)
90% (74)
Prevalence of Hypersensitivity Reactions
χ2
(4) = 8.90, p = 0.064
Allergies
22% (19)
25% (14)
35% (29)
Intolerances
5% (4)
4% (2)
11% (9)
No Reactions
73% (62)
71% (39)
54% (44)
Medication intake (regularly)
χ2
(6) = 2.64, p = 0.852
Thyroid Disease
7% (6)
11% (6)
7% (6)
Hypertension
4% (3)
2% (1)
2% (2)
Cholesterol Level
–
–
–
Other Medication
2% (2)
4% (2)
6% (5)
No Medication
87% (74)
84% (46)
84% (69)
Contraceptives (females only)
12% (10)
5% (3)
20% (16)
χ2
(2) = 5.93, p = 0.051
‘Health-Related Behaviors’
Smoking habits
χ2
(4) = 4.00, p = 0.406
Non-Smoker
67% (57)
56% (31)
52% (43)
Ex-Smoker
32% (27)
42% (23)
45% (37)
Smoker
1% (1)
2% (1)
2% (2)
Supplement intake
Prescribed by doctor
8% (7)
7% (4)
7% (6)
χ2
(2) = 0.07, p = 0.968
To boost your performance
χ2
(4) = 3.52, p = 0.476
Occasionally
16% (14)
11% (6)
9% (7)
Regularly/every day
2% (2)
4% (2)
1% (1)
To cope wit stress
χ2
(4) = 6.66, p = 0.155
Occasionally
6% (5)
7% (4)
6% (5)
Regularly/every day
5% (4)
–
–
Food Choice
Motivation
Because it is healthy
74% (63)
73% (40)
68% (56)
χ2
(2) = 0.74, p = 0.690
Because it is health-promoting
81% (69)
82% (45)
87% (71)
χ2(2) = 1.00, p = 0.607
Because it is good for maintaining health
88% (75)
87% (48)
94% (77)
χ2(2) = 2.15, p = 0.341
Avoided ingredients
Refined Sugar
66% (56)
58% (32)
70% (57)
χ2(2) = 1.89, p = 0.390
Sweetener
82% (73)
64% (35)
82% (67)
χ2(2) = 5.63, p = 0.060
Continued
9
Vol.:(0123456789)
Scientific Reports | (2022) 12:10295 |
https://doi.org/10.1038/s41598-022-13844-4
www.nature.com/scientificreports/
sensitivity reactions” (p = 0.038); (3) there was an association between sex and smoking habits (p = 0.048); (4)
training (weekly mileage) had an association with food choice (p = 0.003); and (5) there was an association
between age and healthcare utilization (p = 0.002). However, no significant associations were found considering
the dimensions of mental health and medication intake.
Discussion
This study aimed to investigate the potential differences in the health status of recreational half-marathoners,
marathoners/ultra-marathoners, and 10-km runners. Mental health, BW and BMI, the prevalence of chronic
diseases and hypersensitivity reactions, medication and supplement intake, smoking habits, food choice from
ingredients to be avoided or desired, and regular or routine health checkups were measured and compared
between the study groups. The main findings were (1) that while no association between race distance and seven
health dimensions were found, “chronic diseases and hypersensitivity reactions” had a significant association
with race distance, and (2) compared to 10-km and marathon/ultra-marathon runners, half-marathoners showed
a tendency towards better scores in six out of eight dimensions of health (BW/BMI, mental health, chronic
diseases and hypersensitivity reactions, medication intake, smoking habits, and health care utilization) with
an average score of 77.1%; the half-marathon distance was found to contribute best to the overall health status
among endurance runners.
Interestingly, only 8% of half-marathon runners and 10% of the overall sample reported “sport for health” as
the basic assignment to a sports area, while “sport for leisure” (54% of total participants, 64% of half-marathoners)
and “sport for performance” (36% of total participants, 28% of half-marathon runners) were ranked higher.
Cluster and respective Dimensions
HM
M/UM
10 km
Statistics
Fat in General
38% (32)
44% (24)
51% (42)
χ2(2) = 3.13, p = 0.210
Saturated Fats
58% (49)
58% (32)
61% (50)
χ2(2) = 0.21, p = 0.899
Cholesterol
42% (36)
45% (25)
48% (39)
χ2(2) = 0.46, p = 0.794
White Flour
60% (51)
60% (33)
79% (65)
χ2
(2) = 8.70, p = 0.013
Sweets
62% (53)
51% (28)
72% (59)
χ2
(2) = 6.29, p = 0.043
Nibbles
58% (59)
53% (29)
72% (59)
χ2
(2) = 6.11, p = 0.047
Alcohol
52% (44)
53% (29)
60% (49)
χ2
(2) = 1.22, p = 0.542
Caffeine
38% (32)
25% (14)
39% (32)
χ2
(2) = 3.04, p = 0.219
Desired ingredients
Vitamins
82% (70)
80% (44)
80% (66)
χ2
(2) = 0.15, p = 0.928
Minerals/trace elements
73% (62)
71% (39)
73% (60)
χ2
(2) = 0.10, p = 0.953
Antioxidants
54% (46)
45% (25)
52% (43)
χ2
(2) = 1.06, p = 0.587
Phytochemicals
44% (37)
40% (22)
54% (44)
χ2
(2) = 2.92, p = 0.232
Fiber
71% (60)
69% (38)
70% (57)
χ2
(2) = 0.04, p = 0.980
Health care utilization
Regular check-ups or routine health checks
62% (53)
49% (27)
54% (44)
χ2
(2) = 2.64, p = 0.268
Table 3. Descriptive and ANOVA results for the eight dimensions of health status displayed by race distance.
Data are presented as “percentage of prevalence (n)” or “median (IQR)”. BMI body mass index, BW body
weight, HM half-marathon, IQR interquartile range, km kilometers, M/UM marathon/ultra-marathon.
Table 4. Domain scores of ‘health-related indicators’ and ‘health-related behaviors’ of endurance runners,
displayed by race distance groups. Data are presented as Domain Scores and (SD): Low scores indicate
detrimental health effects; high scores indicate beneficial health effects (scales: 0–1). BMI body mass index, BW
body weight, HM half-marathon, km kilometers, M/UM marathon/ultra-marathon.
Total
HM
M/UM
10 km
Statistics
Health-Related Indicators
BW and BMI
0.65 (0.40)
0.69 (0.39)
0.67 (0.41)
0.60 (0.42)
F(2, 242) = 0.84, p = 0.433
Mental Health
0.63 (0.48)
0.73 (0.45)
0.58 (0.50)
0.56 (0.50)
F(2, 219) = 2.95, p = 0.054
Chronic Diseases/Hypersensitivity Reactions
0.85 (0.19)
0.88 (0.18)
0.85 (0.19)
0.81 (0.20)
F(2, 219) = 3.31, p = 0.038
Medication Intake
0.85 (0.36)
0.87 (0.34)
0.84 (0.37)
0.84 (0.37)
F(2, 219) = 0.20, p = 0.817
Health-Related Behaviors
Smoking
0.79 (0.27)
0.83 (0.25)
0.77 (0.27)
0.75 (0.27)
F(2, 219) = 2.00, p = 0.138
Supplement Intake
0.90 (0.20)
0.88 (0.23)
0.91 (0.21)
0.92 (0.17)
F(2, 219) = 0.92, p = 0.400
Food Choice
0.68 (0.22)
0.67 (0.21)
0.65 (0.26)
0.72 (0.20)
F(2, 219) = 1.32, p = 0.270
Healthcare Utilization
0.56 (0.50)
0.62 (0.49)
0.49 (0.50)
0.54 (0.50)
F(2, 219) = 1.32, p = 0.270
10
Vol:.(1234567890)
Scientific Reports | (2022) 12:10295 |
https://doi.org/10.1038/s41598-022-13844-4
www.nature.com/scientificreports/
“Hobby” and “health” with 34% and 19% of total participants, respectively, were ranked highest among other
initial motives for running, with no considerable difference between the study groups. The number of completed
races shows that endurance athletes in the present study are not novices but rather active in recreational (not
professional) running. It has been shown that recreational participation in running activities could affect some
health-related findings49, which could be linked to the participants’ slight emphasis on specific personal achieve-
ments versus the joy of running (53% vs. 47%) as the main goal to participate in running events. Consistent with
the present findings, it has been reported that “the joy of running races” was a top reason, and “winning” was
identified as an unimportant reason to participate in running events4. Although “health” was the second-highest
ranked reason among the seven motivations for running, it could be considered as the 1st rank (by 44%) when
pooled with two other health-related motivations (BW loss and maintenance). This finding is consistent with the
literature available, with the main underlying intention probably being to achieve the advantageous effects and
pronounced benefits associated with health1,4, especially for long-term adherence to running activity4,50. Run-
ning is expected to be a powerful strategy in the prevention of diseases, promotion of health, and maintenance
of a good state of health, especially in elderly populations with an age of ≥ 50 years50.
Table 5. MANOVA results for the eight dimensions of health status. BMI body mass index, BW body weight,
df degrees of freedom. F F-value, η2 partial effect (small: 0.01; medium: 0.059; large: 0.138), p p value for
between-group differences.
Cluster
Dimension
Subgroup
F
df
η2
p
Health-Related Indicators
BW and BMI
Race Distance
0.39
2
0.00
0.677
Sex
1.17
1
0.01
0.281
Age
0.00
1
0.00
0.999
Education (academic qualification)
5.66
2
0.05
0.004
Training (weekly mileage)
0.23
2
0.00
0.797
Mental health
Race Distance
2.97
2
0.03
0.053
Sex
3.43
1
0.02
0.065
Age
1.04
1
0.00
0.310
Education (academic qualification)
0.48
2
0.00
0.619
Training (weekly mileage)
0.95
2
0.01
0.390
Chronic diseases/hypersensi-
tivity reactions
Race Distance
3.04
2
0.03
0.050
Sex
0.61
1
0.00
0.435
Age
0.24
1
0.00
0.623
Education (academic qualification)
0.65
2
0.01
0.525
Training (weekly mileage)
0.71
2
0.01
0.492
Medication Intake
Race Distance
0.20
2
0.00
0.815
Sex
0.92
1
0.00
0.340
Age
3.05
1
0.01
0.082
Education (academic qualification)
1.43
2
0.01
0.241
Training (weekly mileage)
0.56
2
0.01
0.573
Health-Related Behaviors
Smoking habits
Race Distance
2.08
2
0.02
0.128
Sex
3.96
1
0.02
0.048
Age
1.97
1
0.01
0.161
Education (academic qualification)
5.35
2
0.05
0.005
Training (weekly mileage)
0.25
2
0.00
0.776
Supplement intake
Race Distance
1.04
2
0.01
0.356
Sex
1.74
1
0.01
0.189
Age
3.05
1
0.01
0.082
Education (academic qualification)
3.88
2
0.04
0.022
Training (weekly mileage)
0.37
2
0.00
0.686
Food choice
Race Distance
1.62
2
0.02
0.200
Sex
0.20
1
0.00
0.655
Age
0.55
1
0.00
0.459
Education (academic qualification)
0.29
2
0.00
0.749
Training (weekly mileage)
6.06
2
0.06
0.003
Healthcare utilization
Race Distance
1.37
2
0.01
0.256
Sex
2.86
1
0.01
0.092
Age
9.62
1
0.05
0.002
Education (academic qualification)
1.40
2
0.01
0.249
Training (weekly mileage)
0.11
2
0.00
0.899
11
Vol.:(0123456789)
Scientific Reports | (2022) 12:10295 |
https://doi.org/10.1038/s41598-022-13844-4
www.nature.com/scientificreports/
BW and BMI.
Four out of five endurance runners in this study were found to have a BW that corresponds
to a healthy BMINORM. Half-marathoners most often matched the BMINORM and consequently had higher health
scores compared to marathoners/ultra-marathoners and 10-km runners. However, 10-km runners were found
to have lower BW than half- to ultra-marathoners, nicely matching their reports where BW loss was ranked 2nd
highest motivation to start running. In addition, the higher score of 10-km runners in food choices compared
to runners over longer distances could be partially associated with the existing findings regarding their trend
toward having a lower BW. Another justification could be the higher number of vegan runners in 10-km com-
pared to half-marathon and marathon/ultra-marathon groups in the present study.
About 25% of runners in the present study stated BW management (loss: 18%, and maintenance: 7%) as the
reason to start running. However, the half-marathoners seem to established a good balance between running-
induced energy required and dietary intake, as they reported least often a decrease in BW due to a change in their
diet. These findings emphasize the significance of BW control strategies for endurance runners as dietary changes
potentially cause unintended BW loss29,51, and adherence to appropriate nutrition strategies for sustainable BW
management is highly advised to endurance runners29. Although the lower BMI and being leaner were found
to be associated with increased endurance running performance52, and training/competing in longer race dis-
tances correlates with a decrease in BW and body fat53, evidence excludes marathon runners or ultra-endurance
athletes from this fact54,55. This is consistent with the present findings where marathon/ultra-marathon runners
had a slight but non-significant higher BMI. The higher BMI of ultra-marathon runners compared to shorter
distance endurance runners might be due to the lower importance of running speed in long-distance compared
to shorter distance runs. In general, however, reports from the successful runners over 10-km and marathon
distance indicate that an optimal BMI for health and performance was found to be between 19 and 20 kg/m256.
The vegan diet was shown to effectively reduce BW and particularly body fat57,58, with favorable effects on run-
ning performance, if planed appropriately59. Consistently, previous data from our laboratory show that vegan
endurance runners are significantly leaner than omnivores (64 kg vs. 68 kg), contributing to their overall state
of health with the highest health score (69%)10.
Mental health.
While most participants were not suffering from mental stress, half-marathoners reported
lower perception of pressure and stress compared to 10-km runners and marathoners/ultra-marathoners. In line
with the present findings, it has been found that endurance running leads to stress reduction, a better mood, and
higher resilience to psychological pressure and anxiety43,60. However, data in terms of the appropriate amount
of physical activity in order to maximize these positive effects while avoiding negative effects is sparse. Too little
exercise does not evoke beneficial effects, but too much exercise (defined as overtraining) can cause the percep-
tion of stress60. Half-marathon allows performance to increase within a short period of time, which provides the
feeling of success38. These characteristics are supposed to lead to a certain degree of life satisfaction and thus a
resilience to stress and pressure perception43.
Chronic disease and hypersensitivity reactions.
The present study revealed a significant difference
between the race distance groups and the dimension, “chronic diseases and hypersensitive reactions”, most ben-
eficially contributing to the half-marathoners’ state of health. Recreational endurance running is well accepted,
having various health effects with robust evidence for regular running to add benefits in aerobic, metabolic,
and cardiovascular function at rest. Consistent with the study findings, running has beneficial influences on the
prevention of chronic and cardio-metabolic diseases, including but not limited to coronary heart disease, stroke,
hypertension, diabetes mellitus type 2, and hypercholesterolemia, mainly via increasing cardiorespiratory fitness
as a strong predictor for morbidity and mortality8,9,12,15. This is in line with another finding from the present
study, where race distance was found to have a significant association with chronic diseases and hypersensitivity
reactions. These exercise-induced advantageous effects are based on various mechanisms, such as adaptations to
the cardiorespiratory and cardio-metabolic system (e.g., changes in the musculoskeletal system and heart muscle
cells, increased maximal oxygen uptake), modifications in hormonal response and enzymatic activity, the acti-
vation of both inflammatory response and detoxification processes, the involvement of pathways associated to
immune response, lipid transport and coagulation, and further genetic adaptions38,61.
The present findings could be influenced by the distribution of diet types, particularly vegetarians and vegans,
among the endurance runners. It has been reported that appropriately planned vegetarian and vegan diets are
healthful and nutritionally adequate even for athletes and provide health benefits for the prevention and treat-
ment of cardio-metabolic disorders and certain diseases such as ischemic heart disease, type 2 diabetes, hyper-
tension, inflammatory problems, and some types of cancer47,62. More specifically, the higher prevalence of plant
diets together with the null association between race distance and the incidence of allergies in the present study
is in line with the available data on the protective effects of fruits and vegetables on the incidence of food aller-
gies, including allergic asthma18 as well as the lower prevalence of allergies in vegan endurance runners (20%)
compared to omnivores (32%) and vegetarians (36%)10. Despite the null association between the occurrence of
food intolerances and race distance in the present study, gastrointestinal complaints due to food intolerances
are common among endurance runners63, probably caused by subclinical food sensitivities that occur during
vigorous exercise64.
Medication intake.
Medication intake in the form of contraceptives was lower with a statistical trend
(p = 0.051) in marathoners/ultra-marathoners compared to half-marathoners and 10-km runners. This finding,
however, could be explained by a sex-based bias as there were fewer females (38%) among marathoners/ultra-
marathoners than in half-marathoners (55%) and 10-km runners (74%). Indeed, 85% of those who reported an
intake of thyroid hormones were women, and 100% of those who reported an intake of other hormones than
12
Vol:.(1234567890)
Scientific Reports | (2022) 12:10295 |
https://doi.org/10.1038/s41598-022-13844-4
www.nature.com/scientificreports/
thyroid medication were women who reported the intake of contraceptives. However, there was no associa-
tion between sex and the dimension “medication intake” when runners were pooled for the MANOVA. As a
well-established fact associated with the present findings, women suffer more often from hypothyroidism than
men65, and importantly, more than 100 million women worldwide use contraceptive pills to avoid undesired
pregnancies66. Although there were no associations between race distance and the intake of any medication, race
distance had a considerable association (score range 0.82–0.86) with medication intake. However, as the major-
ity of distance runners (84–87%) reported no medication intake, caution must be considered when interpreting
the present limited data concerning the intake of non-contraceptives medications across different subgroups of
distance runners.
Smoking habits.
A low rate of smoking (< 2%) was found in endurance runners across all race distances.
Consistently, data indicate that smoking prevalence is usually quite low among endurance runners67. This can
be justified by undesired performance limitations due to smoking68 and the health-consciousness of athletes in
general69. On the other hand, adhering to regular physical exercise, particularly endurance running, can be an
effective way to prevent people from smoking or even help in smoking cessation by reducing cessation-related
mood symptoms, cigarette cravings, and withdrawal symptoms among temporarily abstinent smokers68. In the
present study, there was no association between smoking habits and race distance, but half-marathoners showed
a better score in this dimension. While no comparable data are available in the literature, evidence has found a
positive association between smoking quitters and running activity in terms of weekly training mileage67.
Supplement intake and performance‑enhancing substances.
The most commonly reported sup-
plement by the runners was vitamin D. Several studies have detected a huge difference between required and real
vitamin D intake in athletes worldwide70,71. In addition to dietary intake, athletes’ vitamin D level depends on
skin color, training day-time, indoor/outdoor training, and geographic location71. Although supplement intake
was not associated with race distance, it was found to have high scores (score range 0.88–0.92) among race dis-
tance groups, with a slight predominance in 10-km runners. However, the prevalence of intake was generally
low, reflected by high health scores across all race distance subgroups. Compared with the highest rate of supple-
ment intake reported by half-marathoners (16%), a recent study reported that 30% of female and 40.2% of male
endurance runners consume supplements in order to enhance performance72. Although few studies have yet
compared different groups of endurance runners regarding the patterns of supplement intake73, it has been well-
documented that endurance athletes use supplements to a greater extent than non-endurance athletes74, proba-
bly due to the higher exercise-induced nutritional requirements associated with long-time training, competition,
and recovery75. Reports from a recent study on elite track and field athletes indicated that distance runners have
a significantly higher prevalence in supplemental micronutrient but not macronutrient intake when compared
to runners in other track and field disciplines76. Moreover, there is some evidence for an increasing problem of
doping among elite endurance runners77. However, as the participants in the present study were mostly recrea-
tional runners, they may have different choices of dietary supplements, which could be associated with their
different goals for engaging in training and competition compared to elite athletes49. In addition, findings from
the present study regarding the participants’ attitudes towards food choices characterize them as being health-
conscious, so they might have been aware of potential detrimental effects of risky performance-enhancing sub-
stances. In general, despite the fact that the beneficial effects of many supplements on the promotion of health,
prevention of chronic disease, and enhancement of athletic performance remain unclear78, it is well-established
that these products significantly contribute to the nutrient requirements of athletes78–80.
Food choice.
The present study showed that food choice was not associated with race distance, but the run-
ners over the 10-km distance reported choosing food in order to avoid white flour, sweets, and nibbles more
often than half to ultra-marathoners. This is even reflected by their higher score for food choice (72% vs. 67%
and 65%) along with their motivation for choosing food based on health-promoting and health-maintaining
reasons. However, caution must be warranted while interpreting the findings, as the higher score of 10-km run-
ners in food choice could be potentially associated with their lower BMI among the study groups. Although the
majority of the runners in this study reported following a mixed diet, 59% of 10-km and 56% of half-marathon
runners reported following vegetarian/vegan diets, which were recently found to add most advantageous ben-
efits to the runners’ state of health mainly due to maximizing favorable food choice behaviors in endurance
runners10. The imbalanced distribution of vegans in the 10-km group (compared to the overall groups) might
explain, in part, the highest scores for both supplement intake and food choice, as vegans are known to be more
health-conscious and thus take special care and compensate for potential deficiencies considering critical nutri-
ents such as vitamin B12
10,59,81. Considering a health-related food choice to get desired ingredients by a specific
choice of healthy and health-maintaining items, most participants reported health-conscious behavior across all
race distance subgroups. This finding was in line with available literature2,69, where athletes were characterized
as being health-conscious, particularly with regard to food choice10.
Healthcare utilization.
Overall, most athletes reported seeing a doctor at least once a year and making
use of regular health checkups. These findings were consistent with the previous literature82 and emphasize the
fact that regular and sustainable physical activity can diminish morbidity rates and thus the necessity for doctor
consultations83. The endurance runners of the present study were found to have a good balance between healthy
physical activity and vigorous exercise, which could be advantageous for gaining the desired health effects2, and
importantly for the avoidance of the detrimental consequences of overtraining following excessive running or
training activities. In the present study, there was a statistically significant association between race distance and
13
Vol.:(0123456789)
Scientific Reports | (2022) 12:10295 |
https://doi.org/10.1038/s41598-022-13844-4
www.nature.com/scientificreports/
age. Interestingly, and although being older than runners over other distances, marathoners/ultra-marathoners
had a low score for regular and routine health checkups, indicating disadvantageous contribution to overall
health from weak healthcare utilization.
Limitations, strengths, and future perspectives.
There are limitations worth mentioning. The pre-
sent study shares with others the limitations of the cross-sectional design. The fact that the findings relied on
self-reported records should be considered as the primary limitation since under- and over-reporting are poten-
tially prevalent in self-reported data. However, this effect was compensated by using control questions. Also, the
high intrinsic motivation of the participants could be consequential to increase the accuracy of their answers to
provide a high quality of the data set. The operationalization of state of health as a latent variable (domain scores)
should also be considered as a statistical limitation. Nonetheless, the health score was identified as a meaningful
tool to assess the health status. In this regard, however, retrospective rating of the cross-sectional design might
raise misunderstandings about the associations between health-related variables and race distance, and thus,
caution must be warranted in the representativeness of the present findings. Moreover, the sex-based imbalance
in the study groups (particularly the higher number of males in the marathon/ultra-marathon group and females
in the 10-km group) could be influential on the health-related findings, as females are well-known to be more
health-conscious than males considering favorable habits and healthy lifestyles (e.g., physical activity, alcohol/
nicotine, plant-based diets). Nevertheless, the data contribute to the growing scientific interest and knowledge in
health-related consequences of endurance exercise for distance running in particular, and can be taken as a step
towards broadening the body of evidence in the field.
Although it is well-established that endurance running offers various health benefits, the body of science is
still contradictory considering both quantity and quality of running activity that enables obtaining the maximum
beneficial health effects and preventing the minimum undesired or adverse effects. Therefore, specific knowledge
about the interconnectedness of running distance (in training and racing) and health can provide a better basis
for athletes, coaches, physicians, and specialists to optimize health-related training and racing strategies. Thus, the
results might be useful for different populations by providing such knowledge to aid the decision of an active and
healthy lifestyle, with regular involvement in running training, and also to advise individuals to run for sustain-
able health outcomes. Even at community and public health levels, health authorities can use this information
to support policies towards investing in running programs that promote sustainable running training strategies.
Conclusions
Regardless of the race distance, endurance runners in the present study showed an optimal state of health. This
finding supports the notion that endurance running contributes beneficially to an increased level of health. Half-
marathon running was found to contribute to 62–88% of their overall state of health; in addition, the higher
score of half-marathon runners in overall state of health (77.1% vs. 72.0% in marathon/ultra-marathon runners
and 71.7% in 10-km runners), along with the predominance of half-marathoners in six out of eight dimensions,
might suggest that recreational runners over the half-marathon distance have a tendency toward a better health
status compared to runners over shorter and longer distances. However, among eight health-related dimensions
investigated in the present study, only the “chronic diseases and hypersensitivity reactions” dimension was found
to have a significant association with race distance, with a significantly better status for half-marathon runners
compared to marathoners/ultra-marathoners and 10-km runners.
Data availability
The datasets generated during and/or analysed during the current study are not publicly available but may be
made available upon reasonable request.
Received: 10 September 2021; Accepted: 16 May 2022
References
1. Predel, H. G. Marathon run: Cardiovascular adaptation and cardiovascular risk. Eur. Heart J. 35, 3091–3098. https:// doi. org/ 10.
1093/ eurhe artj/ eht502 (2014).
2. Wirnitzer, K. C. Vegan nutrition: latest boom in health and exercise. In Therapeutic, Probiotic, and Unconventional Foods (eds
Grumezescu, A. M. & Holban, A. M.) 387–453 (Academic Press, Elsevier Inc., New York, 2018).
3. Waśkiewicz, Z., Nikolaidis, P. T., Chalabaev, A., Rosemann, T. & Knechtle, B. Motivation in ultra-marathon runners. Psychol. Res.
Behav. Manag. 12, 31–37. https:// doi. org/ 10. 2147/ PRBM. S1890 61 (2018).
4. Malchrowicz-Mòsko, E. & Poczta, J. Running as a form of therapy socio-psychological functions of mass running events for men
and women. Int. J. Environ. Res. Public Health 15, 2262. https:// doi. org/ 10. 3390/ ijerp h1510 2262 (2018).
5. Waśkiewicz, Z. et al. What motivates successful marathon runners? The role of sex, age, education, and training experience in
Polish runners. Front. Psychol. 10, 1671. https:// doi. org/ 10. 3389/ fpsyg. 2019. 01671 (2019).
6. Hagan, J. C. III. Pheidippides’ last words: “My feet are killing me!”. Mo Med. 109, 256–258 (2012).
7. Tschopp, M. & Brunner, F. Diseases and overuse injuries of the lower extremities in long distance runners. Z. Rheumatol. 76,
443–450. https:// doi. org/ 10. 1007/ s00393- 017- 0276-6 (2017).
8. Hespanhol Junior, L. C., Pillay, J. D., van Mechelen, W. & Verhagen, E. Meta-analyses of the effects of habitual running on indices
of health in physically inactive adults. Sports Med. 45, 1455–1468. https:// doi. org/ 10. 1007/ s40279- 015- 0359-y (2015).
9. Oja, P. et al. Health benefits of different sport disciplines for adults: Systematic review of observational and intervention studies
with meta-analysis. Br. J. Sports Med. 49, 434–440. https:// doi. org/ 10. 1136/ bjspo rts- 2014- 093885 (2015).
10. Ainsworth, B. E. et al. Compendium of physical activities: An update of activity codes and MET intensities. Med. Sci. Sports Exerc.
32, S498-504. https:// doi. org/ 10. 1097/ 00005 768- 20000 9001- 00009 (2000).
11. Williams, P. T. Lower prevalence of hypertension, hypercholesterolemia, and diabetes in marathoners. Med. Sci. Sports Exerc. 41,
523–529. https:// doi. org/ 10. 1249/ MSS. 0b013 e3181 8c1752 (2009).
14
Vol:.(1234567890)
Scientific Reports | (2022) 12:10295 |
https://doi.org/10.1038/s41598-022-13844-4
www.nature.com/scientificreports/
12. Nikolaidis, P. T., Clemente-Suárez, V. J., Chlíbková, D. & Knechtle, B. Training, anthropometric, and physiological characteristics
in men recreational marathon runners: the role of sport experience. Front. Physiol. 12, 666201. https:// doi. org/ 10. 3389/ fphys. 2021.
666201 (2021).
13. Wirnitzer, K. et al. Health status of female and male vegetarian and vegan endurance runners compared to omnivores-results from
the NURMI Study (Step 2). Nutrients 11, 29. https:// doi. org/ 10. 3390/ nu110 10029 (2018).
14. Knechtle, B. & Quarella, A. Laufen tut gut—Oder wie man ohne Psychiater aus der Depression zum Marathon kommt! [Running
helps—or how you escape depression without a psychiatrist and end up running a marathon!]. Praxis (Bern 1994) 96, 1351–1356
(2007).
15. Lee, D. et al. Leisure-time running reduces all-cause and cardiovascular mortality risk. J. Am. Coll. Cardiol. 64, 472–481. https://
doi. org/ 10. 1016/j. jacc. 2014. 04. 058 (2014).
16. Williams, P. T. Walking and running produce similar reductions in cause-specific disease mortality in hypertensives. Hypertension
62, 485–491. https:// doi. org/ 10. 1161/ HYPER TENSI ONAHA. 113. 01608 (2013).
17. Chomistek, A. K., Cook, N. R., Flint, A. J. & Rimm, E. B. Vigorous-intensity leisure-time physical activity and risk of major chronic
disease in men. Med. Sci. Sports Exerc. 44, 1898–1905. https:// doi. org/ 10. 1249/ MSS. 0b013 e3182 5a68f3 (2012).
18. Zimmer, P. et al. Exercise-induced natural killer cell activation is driven by epigenetic modifications. Int. J. Sports Med. 36, 510–515.
https:// doi. org/ 10. 1055/s- 0034- 13985 31 (2015).
19. Schnohr, P., O’Keefe, J., Marott, J., Lange, P. & Jensen, G. Dose of jogging and long-term mortality. The Copenhagen City Heart
Study. J. Am. Coll. Cardiol. 65, 411–419. https:// doi. org/ 10. 1016/j. jacc. 2014. 11. 023 (2015).
20. Fontana, L., Meyer, T. E., Klein, S. & Holloszy, J. O. Long-term low-calorie low-protein vegan diet and endurance exercise are
associated with low cardiometabolic risk. Rejuvenation Res. 10, 225–234. https:// doi. org/ 10. 1089/ rej. 2006. 0529 (2007).
21. Bernecker, C. et al. Evidence for an exercise induced increase of TNF-α and IL-6 in marathon runners. Scand. J. Med. Sci. Sports
23, 207–214. https:// doi. org/ 10. 1111/j. 1600- 0838. 2011. 01372.x (2013).
22. Laye, M. J., Nielsen, M. B., Hansen, L. S., Knudsen, T. & Pedersen, B. K. Physical activity enhances metabolic fitness independently
of cardiorespiratory fitness in marathon runners. Dis Mark. https:// doi. org/ 10. 1155/ 2015/ 806418 (2015).
23. Roberts, W. O. et al. Long-term marathon running is associated with low coronary plaque formation in women. Med. Sci. Sports
Exerc. 49, 641–645. https:// doi. org/ 10. 1249/ MSS. 00000 00000 001154 (2017).
24. Rosin, B. Is marathon running toxic? An observational study of cardiovascular disease prevalence and longevity in 54 male mara-
thon runners. Phys. Sportsmed. 45, 105–109. https:// doi. org/ 10. 1080/ 00913 847. 2017. 12885 45 (2017).
25. Van der Wall, E. E. Long-distance running: running for a long life?. Neth. Heart J. 22, 89–90. https:// doi. org/ 10. 1007/ s12471- 014-
0521-4 (2014).
26. Zimmer, P. et al. Impact of a half marathon on cellular immune system, pro-inflammatory cytokine levels, and recovery behavior
of breast cancer patients in the aftercare compared to healthy controls. Eur. J. Haematol. 96, 152–159. https:// doi. org/ 10. 1111/ ejh.
12561 (2016).
27. Williams, P. T. Relationship of running intensity to hypertension, hypercholesterolemia, and diabetes. Med. Sci. Sports Exerc. 40,
1740–1748. https:// doi. org/ 10. 1249/ MSS. 0b013 e3181 7b8ed1 (2008).
28. Gosling, C. M., Forbes, A. B., McGivern, J. & Gabbe, B. J. A profile of injuries in athletes seeking treatment during a triathlon race
series. Am. J. Sports Med. 38, 1007–1014. https:// doi. org/ 10. 1177/ 03635 46509 356979 (2010).
29. Manore, M. M. Weight management for athletes and active individuals: a brief review. Sports Med. 45, 83–92. https:// doi. org/ 10.
1007/ s40279- 015- 0401-0 (2015).
30. Waite, O., Smith, A., Madge, L., Spring, H. & Noret, N. Sudden cardiac death in marathons: a systematic review. Phys Sportsmed.
44, 79–84. https:// doi. org/ 10. 1080/ 00913 847. 2016. 11350 36 (2016).
31. Knechtle, B. & Nikolaidis, P. T. Physiology and pathophysiology in ultra-marathon running. Front. Physiol. 9, 634. https:// doi. org/
10. 3389/ fphys. 2018. 00634 (2018).
32. Merghani, A. et al. Prevalence of subclinical coronary artery disease in masters endurance athletes with a low atherosclerotic risk
profile. Circulation 136, 126–137. https:// doi. org/ 10. 1161/ CIRCU LATIO NAHA. 116. 026964 (2017).
33. Schwartz, R. S. et al. Increased coronary artery plaque volume among male marathon runners. Mo Med. 111, 89–94 (2014).
34. Kim, J. H. et al. Cardiac arrest during long-distance running races. N Engl J Med. 366, 130–140. https:// doi. org/ 10. 1056/ NEJMo
a1106 468 (2012).
35. Sanchez, L. D., Corwell, B. & Berkoff, D. Medical problems of marathon runners. Am. J. Emerg. Med. 24, 608–615. https:// doi. org/
10. 1016/j. ajem. 2006. 01. 023 (2006).
36. Chugh, S. S. & Weiss, J. B. Sudden cardiac death in the older athlete. J. Am. Coll. Cardiol. 65, 493–502. https:// doi. org/ 10. 1016/j.
jacc. 2014. 10. 064 (2015).
37. Lippi, G. et al. Influence of a half-marathon run on NT-proBNP and troponin T. Clin. Lab. 54, 251–4 (2008).
38. Dalle Carbonare, L. et al. Can half-marathon affect overall health? The yin-yang of sport. J. Proteom. 170, 80–87. https:// doi. org/
10. 1016/j. jprot. 2017. 09. 004 (2018).
39. Jaworski, C. A. Medical concerns of marathons. Curr. Sports Med. Rep. 4, 137–143. https:// doi. org/ 10. 1097/ 01. csmr. 00003 06196.
51994. 5f (2005).
40. Hoffman, M. D. & Krishnan, E. Health and exercise-related medical issues among 1,212 ultramarathon runners: baseline findings
from the Ultrarunners Longitudinal TRAcking (ULTRA) Study. PLoS ONE 9, e83867. https:// doi. org/ 10. 1371/ journ al. pone. 00838
67 (2014).
41. Breslow, R. G. et al. Medical tent utilization at 10-km road races: Injury, illness, and influencing factors. Med. Sci. Sports Exerc. 51,
2451–2457. https:// doi. org/ 10. 1249/ MSS. 00000 00000 002068 (2019).
42. Wirnitzer, K. et al. Prevalence in running events and running performance of endurance runners following a vegetarian or vegan
diet compared to non-vegetarian endurance runners: The NURMI Study. Springerplus 5, 458. https:// doi. org/ 10. 1186/ s40064- 016-
2126-4 (2016).
43. Boldt, P. et al. Quality of life of female and male vegetarian and vegan endurance runners compared to omnivores - results from
the NURMI Study (Step 2). J. Int. Soc. Sports Nutr. 15, 33. https:// doi. org/ 10. 1186/ s12970- 018- 0237-8 (2018).
44. Marc, A. et al. Marathon progress: demography, morphology and environment. J. Sports Sci. 32, 524–532. https:// doi. org/ 10. 1080/
02640 414. 2013. 835436 (2014).
45. World Health Organization (WHO). Body Mass Index—BMI. Table 1. Nutritional Status. In WHO Regional Office for Europe.
2018. http:// www. euro. who. int/ en/ health- topics/ disea se- preve ntion/ nutri tion/a- healt hy- lifes tyle/ body- mass- index- bmi. Accessed
18 Sept 2018.
46. World Health Organization (WHO). Mean Body Mass Index (BMI). Situation and Trends. In Global Health Observatory (GHO)
Data. 2018. http:// www. who. int/ gho/ ncd/ risk_ facto rs/ bmi_ text/ en/. Accessed 18 Sept 2018.
47. Melina, V., Craig, W., Levin, S. & Dietetics, AoNa. Position of the academy of nutrition and dietetics: vegetarian diets. J. Acad.
Nutr. Diet. 115, 1970–1980. https:// doi. org/ 10. 1016/j. jand. 2016. 09. 025 (2016).
48. Boldt, P. et al. Sex differences in the health status of endurance runners: Results from the NURMI Study (Step 2). J. Strength Cond.
Res. 33, 1929–1940. https:// doi. org/ 10. 1519/ JSC. 00000 00000 003010 (2019).
49. Desbrow, B., Slater, G. & Cox, G. R. Sports nutrition for the recreational athlete. Aust. J. Gen. Pract. 49, 17–22. https:// doi. org/ 10.
31128/ AJGP- 10- 19- 5108 (2020).
15
Vol.:(0123456789)
Scientific Reports | (2022) 12:10295 |
https://doi.org/10.1038/s41598-022-13844-4
www.nature.com/scientificreports/
50. Allender, S., Cowburn, G. & Foster, C. Understanding participation in sport and physical activity among children and adults: A
review of qualitative studies. Health Educ. Res. 21, 826–835. https:// doi. org/ 10. 1093/ her/ cyl063 (2006).
51. Turner-McGrievy, G. M., Moore, W. J. & Barr-Anderson, D. The Interconnectedness of diet choice and distance running: Results of
the Research Understanding the NutritioN of Endurance Runners (RUNNER) Study. Int. J. Sport Nutr. Exerc. Metab. 26, 205–211.
https:// doi. org/ 10. 1123/ ijsnem. 2015- 0085 (2015).
52. Larsen, H. B. Kenyan dominance in distance running. Comp. Biochem. Physiol. A Mol. Integr. Physiol. 136, 161–170. https:// doi.
org/ 10. 1016/ s1095- 6433(03) 00227-7 (2003).
53. Knechtle, B., Stiefel, M., Rosemann, T., Rüst, C. & Zingg, M. Running and the association with anthropometric and training
characteristics. Ther Umsch. 72, 343–355. https:// doi. org/ 10. 1024/ 0040- 5930/ a0006 85 (2015).
54. Knechtle, B., Knechtle, P., Andonie, J. L. & Kohler, G. Influence of anthropometry on race performance in extreme endurance
triathletes: World Challenge Deca Iron Triathlon 2006. Br. J. Sports Med. 41, 644–648. https:// doi. org/ 10. 1136/ bjsm. 2006. 035014
(2007).
55. Hagan, R. D., Upton, S. J., Duncan, J. J. & Gettman, L. R. Marathon performance in relation to maximal aerobic power and training
indices in female distance runners. Br. J. Sports Med. 21, 3–7. https:// doi. org/ 10. 1136/ bjsm. 21.1.3 (1987).
56. Sedeaud, A. et al. BMI, a performance parameter for speed improvement. PLoS ONE 9, e90183. https:// doi. org/ 10. 1371/ journ al.
pone. 00901 83 (2014).
57. Berkow, S. E. & Barnard, N. Vegetarian diets and weight status. Nutr. Rev. 64, 175–188. https:// doi. org/ 10. 1111/j. 1753- 4887. 2006.
tb002 00.x (2006).
58. Barnard, N. D., Levin, S. M. & Yokoyama, Y. A systematic review and meta-analysis of changes in body weight in clinical trials of
vegetarian diets. J. Acad Nutr. Diet. 115, 954–969. https:// doi. org/ 10. 1016/j. jand. 2014. 11. 016 (2015).
59. Wirnitzer, K. C. Vegan diet in sports and exercise. Health benefits and advantages to athletes and physically active people. A nar-
rative review. Int. J. Sports Exerc. Med. 6, 165 (2020).
60. Gerber, M. & Pühse, U. Do exercise and fitness protect against stress-induced health complaints? A review of the literature. Scand.
J. Public Health 37, 801–819. https:// doi. org/ 10. 1177/ 14034 94809 350522 (2009).
61. Bishop-Bailey, D. Mechanisms governing the health and performance benefits of exercise. Br. J. Pharmacol. 170, 1153–1166. https://
doi. org/ 10. 1111/ bph. 12399 (2013).
62. Kahleova, H., Levin, S. & Barnard, N. Cardio-metabolic benefits of plant-based diets. Nutrients 9, 848. https:// doi. org/ 10. 3390/
nu908 0848 (2017).
63. Simons, S. M. & Kennedy, R. G. Gastrointestinal problems in runners. Curr. Sports Med. Rep. 3, 112–116. https:// doi. org/ 10. 1249/
00149 619- 20040 4000- 00011 (2004).
64. Costa, R. J. S., Knechtle, B., Tarnopolsky, M. & Hoffman, M. D. Nutrition for ultramarathon running: trail, track, and road. Int. J.
Sport Nutr. Exerc. Metab. 29, 130–140. https:// doi. org/ 10. 1123/ ijsnem. 2018- 0255 (2019).
65. Dunn, D. & Turner, C. Hypothyroidism in women. Nurs. Womens Health 20, 93–98. https:// doi. org/ 10. 1016/j. nwh. 2015. 12. 002
(2016).
66. Christin-Maitre, S. History of oral contraceptive drugs and their use worldwide. Best Pract. Res. Clin. Endocrinol. Metab. 27, 3–12.
https:// doi. org/ 10. 1016/j. beem. 2012. 11. 004 (2013).
67. Marti, B., Abelin, T., Minder, C. E. & Vader, J. P. Smoking, alcohol consumption, and endurance capacity: An analysis of 6,500
19-year-old conscripts and 4,100 joggers. Prev. Med. 17, 79–92. https:// doi. org/ 10. 1016/ 0091- 7435(88) 90074-6 (1988).
68. Mündel, T. Nicotine: sporting friend or foe? A review of athlete use, performance consequences and other considerations. Sports
Med. 47, 2497–2506. https:// doi. org/ 10. 1007/ s40279- 017- 0764-5 (2017).
69. Breslin, G., Shannon, S., Haughey, T., Donnelly, P. & Leavey, G. A systematic review of interventions to increase awareness of mental
health and well-being in athletes, coaches and officials. Syst. Rev. 6, 177. https:// doi. org/ 10. 1186/ s13643- 017- 0568-6 (2017).
70. Knez, W. L. & Peake, J. M. The prevalence of vitamin supplementation in ultraendurance triathletes. Int. J. Sport Nutr. Exerc. Metab.
20, 507–514. https:// doi. org/ 10. 1123/ ijsnem. 20.6. 507 (2010).
71. Larson-Meyer, E. Vitamin D supplementation in athletes. Nestle Nutr. Inst. Workshop Ser. 75, 109–121. https:// doi. org/ 10. 1159/
00034 5827 (2013).
72. Wilson, P. B. Nutrition behaviors, perceptions, and beliefs of recent marathon finishers. Phys. Sportsmed. 44, 242–251. https:// doi.
org/ 10. 1080/ 00913 847. 2016. 11774 77 (2016).
73. Wirnitzer, K. et al. Sex differences in supplement intake in recreational endurance runners—results from the NURMI Study (Step
2). Nutrients 13, 2776. https:// doi. org/ 10. 3390/ nu130 82776 (2021).
74. Maughan, R. J., Depiesse, F. & Geyer, H., International Association of Athletics Federations. The use of dietary supplements by
athletes. J. Sports Sci. 25, S103–S113 (2007). https:// doi. org/ 10. 1080/ 02640 41070 16073 95.
75. Tiller, N. B. et al. International Society of Sports Nutrition Position Stand: Nutritional considerations for single-stage ultra-
marathon training and racing. J. Int. Soc. Sports Nutr. 16, 50. https:// doi. org/ 10. 1186/ s12970- 019- 0312-9 (2019).
76. Tabata, S. et al. Use of nutritional supplements by elite Japanese track and field athletes. J. Int. Soc. Sports Nutr. 17, 38. https:// doi.
org/ 10. 1186/ s12970- 020- 00370-9 (2020).
77. Hoberman, J. History and prevalence of doping in the marathon. Sports Med. 37, 386–388. https:// doi. org/ 10. 2165/ 00007 256- 20073
7040- 00029 (2007).
78. Maughan, R. J. et al. IOC consensus statement: Dietary supplements and the high-performance athlete. Int. J. Sport Nutr. Exerc.
Metab. 28, 104–125. https:// doi. org/ 10. 1123/ ijsnem. 2018- 0020 (2018).
79. Kerksick, C. M. et al. ISSN exercise & sports nutrition review update: research & recommendations. J. Int. Soc. Sports Nutr. 15, 38.
https:// doi. org/ 10. 1186/ s12970- 018- 0242-y (2018).
80. Wirnitzer, K. et al. Supplement intake in vegan, vegetarian, and omnivorous endurance runners—results from the NURMI Study
(Step 2). Nutrients 13, 2741. https:// doi. org/ 10. 3390/ nu130 82741 (2021).
81. Glick-Bauer, M. & Yeh, M. C. The health advantage of a vegan diet: Exploring the gut microbiota connection. Nutrients 6, 4822–
4838. https:// doi. org/ 10. 3390/ nu611 4822 (2014).
82. Shapero, K. et al. Cardiovascular risk and disease among Masters endurance athletes: Insights from the Boston MASTER (Masters
Athletes Survey to Evaluate Risk) Initiative. Sports Med. Open 2, 29. https:// doi. org/ 10. 1186/ s40798- 016- 0053-0 (2016).
83. Persson, G., Brorsson, A., Ekvall Hansson, E., Troein, M. & Strandberg, E. L. Physical activity on prescription (PAP) from the
general practitioner’s perspective—a qualitative study. BMC Fam. Pract. 14, 128. https:// doi. org/ 10. 1186/ 1471- 2296- 14- 128 (2013).
Acknowledgements
There are no professional relationships with companies or manufacturers who will benefit from the results of
the present study.
Author contributions
K.W. conceptualized, designed and developed the study design and the questionnaires together with B.K. and
C.L. K.W. performed data analysis. P.B. and K.W. drafted the manuscript, M.M., D.T. and T.R. helped in drafting
16
Vol:.(1234567890)
Scientific Reports | (2022) 12:10295 |
https://doi.org/10.1038/s41598-022-13844-4
www.nature.com/scientificreports/
the manuscript, and B.K. and K.W. critically reviewed it. Technical support was provided by G.W. All authors
read and approved the final manuscript.
Funding
This research did not receive any specific grant or funding from funding agencies in the public, commercial, or
not-for-profit sectors.
Competing interests
The authors declare no competing interests.
Additional information
Correspondence and requests for materials should be addressed to B.K.
Reprints and permissions information is available at www.nature.com/reprints.
Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and
institutional affiliations.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International
License, which permits use, sharing, adaptation, distribution and reproduction in any medium or
format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the
Creative Commons licence, and indicate if changes were made. The images or other third party material in this
article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the
material. If material is not included in the article’s Creative Commons licence and your intended use is not
permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from
the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/.
© The Author(s) 2022
| Health status of recreational runners over 10-km up to ultra-marathon distance based on data of the NURMI Study Step 2. | 06-18-2022 | Wirnitzer, Katharina,Boldt, Patrick,Wirnitzer, Gerold,Leitzmann, Claus,Tanous, Derrick,Motevalli, Mohamad,Rosemann, Thomas,Knechtle, Beat | eng |
PMC9956911 | Citation: Manresa-Rocamora, A.;
Fuertes-Kenneally, L.; Blasco-Peris,
C.; Sempere-Ruiz, N.; Sarabia, J.M.;
Climent-Paya, V. Is the Verification
Phase a Suitable Criterion for the
Determination of Maximum Oxygen
Uptake in Patients with Heart Failure
and Reduced Ejection Fraction? A
Validation Study. Int. J. Environ. Res.
Public Health 2023, 20, 2764. https://
doi.org/10.3390/ijerph20042764
Academic Editor: Cristian Álvarez
Received: 12 January 2023
Revised: 31 January 2023
Accepted: 2 February 2023
Published: 4 February 2023
Copyright:
© 2023 by the authors.
Licensee MDPI, Basel, Switzerland.
This article is an open access article
distributed
under
the
terms
and
conditions of the Creative Commons
Attribution (CC BY) license (https://
creativecommons.org/licenses/by/
4.0/).
International Journal of
Environmental Research
and Public Health
Article
Is the Verification Phase a Suitable Criterion for the
Determination of Maximum Oxygen Uptake in Patients with
Heart Failure and Reduced Ejection Fraction?
A Validation Study
Agustín Manresa-Rocamora 1,2
, Laura Fuertes-Kenneally 1,3
, Carles Blasco-Peris 1,4
,
Noemí Sempere-Ruiz 1,2
, José Manuel Sarabia 1,2,*
and Vicente Climent-Paya 1,3
1
Institute for Health and Biomedical Research of Alicante (ISABIAL), 03010 Alicante, Spain
2
Department of Sport Sciences, Sports Research Centre, Miguel Hernández University of Elche,
03202 Elche, Spain
3
Cardiology Department, Dr. Balmis General University Hospital, 03010 Alicante, Spain
4
Department of Physical Education and Sport, University of Valencia, 46010 Valencia, Spain
*
Correspondence: [email protected]; Tel.: +34-96-522-25-68
Abstract: The verification phase (VP) has been proposed as an alternative to the traditional criteria
used for the determination of the maximum oxygen uptake (VO2 max) in several populations.
Nonetheless, its validity in patients with heart failure with reduced ejection fraction (HFrEF) remains
unclear. Therefore, the aim of this study was to analyse whether the VP is a safe and suitable
method to determine the VO2 max in patients with HFrEF. Adult male and female patients with
HFrEF performed a ramp-incremental phase (IP), followed by a submaximal constant VP (i.e., 95%
of the maximal workload during the IP) on a cycle ergometer. A 5-min active recovery period (i.e.,
10 W) was performed between the two exercise phases. Group (i.e., median values) and individual
comparisons were performed. VO2 max was confirmed when there was a difference of ≤ 3% in peak
oxygen uptake (VO2 peak) values between the two exercise phases. Twenty-one patients (13 males)
were finally included. There were no adverse events during the VP. Group comparisons showed no
differences in the absolute and relative VO2 peak values between both exercise phases (p = 0.557
and p = 0.400, respectively). The results did not change when only male or female patients were
included. In contrast, individual comparisons showed that the VO2 max was confirmed in 11 patients
(52.4%) and not confirmed in 10 (47.6%). The submaximal VP is a safe and suitable method for the
determination of the VO2 max in patients with HFrEF. In addition, an individual approach should be
used because group comparisons could mask individual differences.
Keywords: cardiorespiratory fitness; VO2 max; HFrEF; exercise testing; respiratory exchange ratio;
gradual exercise test; VO2 peak
1. Introduction
Heart failure with reduced ejection fraction (HFrEF) is a cardiovascular disorder
characterised by symptoms of breathlessness, fluid retention, and exercise intolerance [1–3].
The maximal ramp or step incremental exercise test, coupled with breath-by-breath and
gas exchange measurements, is widely used in patients with HFrEF to measure maximum
oxygen uptake (VO2 max) and for risk stratification [4–6]. VO2 max is defined as the
physiological limit of oxygen utilisation [7] and is considered a strong predictor of mortality
in patients with HFrEF [8,9]. In fact, VO2 max is considered a prognostic factor in advanced
heart failure and is currently used as a key criterion for the selection of candidates for heart
transplantation (i.e., ≤14 mL·kg−1·min−1) [4].
The measurement of VO2 max requires the patient to perform a maximal exercise effort
(i.e., volitional exhaustion) and thus might be substantially underestimated due to muscle
Int. J. Environ. Res. Public Health 2023, 20, 2764. https://doi.org/10.3390/ijerph20042764
https://www.mdpi.com/journal/ijerph
Int. J. Environ. Res. Public Health 2023, 20, 2764
2 of 10
fatigue, breathlessness, and reduced motivation (i.e., submaximal exercise test). In these
circumstances, the peak oxygen uptake (VO2 peak) instead of the VO2 max is obtained.
Consequently, it is important to determine the criteria to accurately categorise an effort
as maximal. The primary criterion used to verify maximal exercise relies on the presence
of the oxygen uptake (VO2) plateau, which is defined as no increase in VO2 despite an
increase in workload rate [10,11]. Nonetheless, data indicate that only a small percentage
of VO2 assessments actually exhibit a VO2 plateau [12–15], supporting the argument that
this physiological phenomenon is not necessary to acutely determine the VO2 max. Thus,
secondary criteria such as the value of the respiratory exchange ratio (RER)—which is
the most frequently used variable in cardiac patients—age-predicted maximal heart rate
(HR), or blood lactate concentrations are commonly used to verify that a maximal exercise
effort has been achieved [13,16,17]. However, evidence suggests that these criteria lack
validity since they can be met with either a maximal or submaximal exercise effort or even,
not reached at all despite a maximal effort [11,18,19]. In summary, traditional criteria (i.e.,
including both primary and secondary) are not reliable methods to ensure that the VO2
max is reached at the end of an incremental exercise test [20].
In an attempt to overcome the shortcomings of traditional criteria, a new criterion for
the determination of the VO2 max has emerged, known as the verification phase (VP) [13].
The VP is a constant-load phase performed following the incremental phase (IP) and a
short recovery period (e.g., 3–5 min). Other protocols have also been used previously [20].
Regarding its intensity, it can be performed either above (i.e., supramaximal verification
phase) or below (i.e., submaximal verification phase) the peak work rate attained in the
previous IP [10,15]. There is evidence demonstrating that the VP is an adequate standard
for validating the VO2 max in healthy individuals [21] and patients with a wide range of
pathologies [22].
In this regard, Bowen et al. [23] investigated whether the submaximal VP was a valid
method to determine the VO2 max in patients with HFrEF. According to the authors, the
VP was well tolerated by patients with HFrEF, and its precision was greater than that of
secondary criteria (i.e., RER). Nonetheless, only male patients were included, and further
research is needed to determine whether VP is suitable and well tolerated by female patients
with HFrEF. Furthermore, although both group and individual comparisons can be used
to validate the VO2 max, group comparisons could mask individual differences between
the VO2 peak values attained in each exercise phase [19,20]. Also, the clinical utility of the
exercise test is its application to the individual rather than the group. For these reasons,
individual comparisons are more useful than group comparisons [24]. In order to perform
these individual comparisons and assess whether or not the VO2 max was reached, a
standard cut-off point should be established, preferentially using relative differences (e.g.,
≤3%) [20]. In contrast, Bowen et al. [23], who included group and individual comparisons,
carried out statistical comparisons. In this study, the VO2 max was confirmed when
statistical significance was not reached (p > 0.050). Nonetheless, the use of statistical
comparisons is a flawed approach because it is designed to detect differences and depends
on the sample size [25]. Thus, the use of different approaches to conduct individual
comparisons and confirm the VO2 max warrants future studies in patients with HFrEF.
Therefore, the main purpose of the current study was to investigate the utility of the
submaximal VP for validating the VO2 max in male and female patients with HFrEF. In
addition, we compared the level of agreement between the VP and traditional (i.e., RER)
criteria for the determination of the VO2 max. Based on previous evidence, we hypothesised
that the submaximal VP would be an adequate criterion to verify the VO2 max in male
and female patients with HFrEF when an individual approach is used, and no agreement
would be found between both criteria.
Int. J. Environ. Res. Public Health 2023, 20, 2764
3 of 10
2. Materials and Methods
2.1. Patients
Participants needed to fulfil the following inclusion criteria to be eligible: (a) male
or female aged between 50 and 70 years old; (b) diagnosed with HFrEF (left ventricular
ejection fraction < 50%); (c) stable phase of the disease with no recent hospitalisation or
visit to the emergency department due to heart failure (within the last six months before
the beginning of the study); (d) New York Heart Association (NYHA) functional class I, II,
or III; (e) under treatment with B-blockers; and (f) sedentary (i.e., not involved in exercise
training for six months). The exclusion criteria were: (a) use of intravenous diuretics in the
last six months; (b) unstable angina or evidence of severe ventricular arrhythmia; (c) atrial
fibrillation; (d) supraventricular arrhythmias; (e) chronic obstructive pulmonary disease;
(f) recent of haemoglobin concentrations outside optimal parameters (13–16.5 g·dL−1);
(g) physical limitations that impeded the completion of the ergometry; and (h) the presence
of ischaemia, arrhythmias, or high frequency of ectopic heartbeats. All patients were fully
informed and signed the informed consent before any procedure related to the study was
performed. The protocol of this study was approved by the competent ethics committee of
the host institution (PI2021-177).
2.2. Measurements
Participants performed a symptom-limited exercise test which comprised two phases;
(a) the ramp-incremental exercise phase (i.e., IP); and (b) the steady-state exercise phase (i.e.,
VP). The test was carried out on an electromagnetically braked cycle ergometer (SanaBike
500 easy, Truchtelfinger, Germany). Before the start of the IP, a 3 min warm-up at 10 W and a
cadence of 50 revolutions per minute (rpm) was performed. The IP ended when the patient
reached volitional exhaustion or was unable to maintain a cadence of at least 45 rpm. The
exercise test was terminated, and the VP was not carried out in the presence of symptoms
of ischaemia or multifocal ectopic heartbeats (symptom-limited). Otherwise, a free-cadence
recovery period of 5 min at 10 W was performed after finishing the IP. Subsequently, the
VP was carried out at 95% of the maximum power reached during the IP. Throughout
the protocol, gas exchange was recorded with the Metalyzer 3B gas analyser (CORTEX
Biophysik, Leipzig, Germany), and HR was monitored with a 12-lead electrocardiograph.
Patients were asked to fast (at least three hours prior to the test), as well as to refrain from
strenuous physical activity (24 h), alcohol, and smoking (three hours prior).
2.3. Data and Statistical Analyses
Gas exchange and ventilatory variables were analysed to remove atypical breaths
(four standard deviations from the local mean) due to swallows, coughs, and so on [26].
VO2 peak was defined as the highest VO2 occurring during each exercise phase (i.e., IP and
VP). VO2 peak, as well as the remaining ventilatory variables obtained at exercise peak (i.e.,
VCO2, oxygen pulse, RER, VE, VE/VCO2, and VE/VO2), were identified using a 12-breath
rolling average [23]. Breathing frequency and HR were averaged over 10 s.
Data are displayed as median (25th and 75th percentiles) and frequency (percentage)
for continuous and categorical variables, respectively, unless stated otherwise. Overall, the
Fisher-Pitman permutation test [27] and the non-parametric 95% confidence interval (CI)
of the difference [28] were used to conduct between-phase comparisons (i.e., VP vs. IP).
The Bland-Altman plot was used to test the agreement between VO2 peak values measured
during the IP and VP.
Individual comparisons between VO2 peak values reached during the two exercise
phases were also conducted. In this regard, the IP-derived VO2 peak was confirmed
(i.e., VO2 max) if the difference with the VP-derived VO2 peak value was ≤ 3%. After-
wards, patients were classified into two groups, depending on whether IP-derived VO2
max values were confirmed or not. Fisher’s exact test, Mann-Whitney test, and Bonett-
Price 95% CI were used to conduct between-group comparisons (i.e., confirmed vs. not
confirmed groups).
Int. J. Environ. Res. Public Health 2023, 20, 2764
4 of 10
The traditional criterion (i.e., RER peak ≥ 1.10) was also used to verify VO2 max [17].
Compared to the VP, the results were classified as follows: (a) agreement, RER ≥ 1.10
and VO2 max confirmed by VP or RER < 1.10 and VO2 max not confirmed by VP; (b)
false positive, RER ≥ 1.10 and VO2 max not confirmed by VP; and (c) false negative,
RER < 1.10 and VO2 max confirmed by VP. The Kappa index was used to analyse the
degree of agreement between the two criteria (i.e., RER vs. VP). All tests were two-sided,
and p values ≤ 0.050 were considered significant. All analyses were performed using
STATA software (version 16.0; Stata Corp LLC, College Station, TX, USA).
3. Results
3.1. Patients
Thirty patients with HFrEF (22 males; 73.3%) fulfilled the inclusion criteria to be
eligible to participate in the current study. Nonetheless, we excluded a total of eight patients
(26.6%) because the VP was considered contraindicated (i.e., symptoms of ischaemia,
arrhythmias, or high frequency of ectopic heartbeats during the IP). Moreover, one patient
(3.3%) did not complete the VP due to knee pain and was also excluded from the analysis.
All excluded patients were male. Therefore, 21 patients (13 males; 61.9%) were finally
included. The characteristics of these patients are shown in Table 1. No adverse events
occurred during the exercise tests. The median age was 64.0 years (57.5; 68.5), and the
median left ventricular ejection fraction was 39.1% (33.0; 43.1). Ischemic etiology was
the cause of HFrEF in almost half of the patients. Most of the included patients were
smokers (81%).
Table 1. Baseline participant characteristics.
Variable
n = 21
Confirmed Group
(n = 11)
Not Confirmed Group
(n = 10)
p
Age, years
64.0 (57.5; 68.5)
64.0 (56.0; 66.0)
65.0 (58.0; 69.3)
0.717
Height, cm
164 (158; 171)
164 (157; 168)
164 (159; 174)
0.617
Weight, kg
70.0 (66.5; 87.3)
72.0 (64.0; 88.5)
69.6 (67.8; 80.6)
0.850
Body mass index, kg/m2
27.8 (24.8; 31.6)
27.8 (25.1; 32.0)
27.7 (24.1; 31.4)
0.557
LVEF, %
39.1 (33.0; 43.1)
39.0 (33.0; 45.0)
37.5 (32.3; 43.0)
0.414
Male (%)
13 (61.9)
7 (63.6)
6 (60.0)
0.999
Ischemic etiology (%)
10 (47.7)
6 (54.6)
4 (40.0)
0.670
Diabetes mellitus (%)
9 (42.9)
4 (36.4)
5 (50.0)
0.670
Hypertension (%)
9 (42.9)
4 (36.4)
5 (50.0)
0.670
Dyslipidaemia (%)
11 (52.4)
6 (54.6)
5 (50.0)
0.999
Smokers (%)
17 (81.0)
11 (100)
6 (60.0)
0.035
ICD (%)
9 (42.9)
5 (45.5)
4 (40.0)
0.999
Drug therapy:
ACEI/ARBs (%)
8 (38.1)
4 (36.4)
4 (40.0)
0.999
ARNI (Sac/Val) (%)
13 (61.9)
7 (63.4)
6 (60.0)
0.999
MRA (%)
17 (81.0)
9 (81.2)
8 (80.0)
0.999
Antiplatelet (%)
8 (38.1)
4 (36.4)
4 (40.0)
0.999
Anticoagulants (%)
2 (9.5)
2 (18.2)
0 (0)
0.476
Diuretics (%)
3 (14.3)
1 (9.1)
2 (20.0)
0.586
ACEI, Angiotensin-converting enzyme inhibitors; ARNI (Sac/Val), angiotensin receptor-neprilysin inhibitor
(sacubitril/valsartan); ICD, implantable cardioverter defibrillator; LVEF, left ventricular ejection fraction; MRA,
mineralocorticoid receptor antagonist. Data are presented as median (25th and 75th percentiles) or frequency
(percentage); p values refer to between-group differences; bold values refer to statistical significance (p ≤ 0.050).
3.2. Group Comparisons
Descriptive group data from the IP and VP, as well as between-phase comparisons,
are presented in Table 2. The median peak work rate during the IP was 55.0 W (46.5;
92.5). The absolute and relative VO2 peak values did not differ between exercise phases
(p = 0.557 and p = 0.400, respectively). RER and VE/VCO2 peak values were lower and
higher, respectively, in the VP than in the IP (p = 0.004 and p = 0.003). The results did not
Int. J. Environ. Res. Public Health 2023, 20, 2764
5 of 10
change when exclusively male or female patients were included in the analyses. Figure 1
shows the Bland-Altman plot for the relative VO2 peak. The mean difference between both
exercise phases was −0.07 mL·kg−1·min−1, while the lower and upper limits of agreement
were −1.59 mL·kg−1·min−1 and 1.45 mL·kg−1·min−1, respectively.
Table 2. Cardiopulmonary responses to the two exercise phases and between-phase comparisons
(n = 21).
Variable
IP
VP
Difference (95% CI)
p
Duration, min
8.5 (7.3; 12.5)
2.8 (2.1; 3.5)
−7.25 (−9.30 to −5.20)
<0.001
HR peak, beats·min−1
112.0 (108.0; 127.0)
109.0 (105.5; 128.0)
−1.00 (−5.92 to 4.92)
0.720
RER peak
1.10 (1.04; 1.12)
1.00 (0.95; 1.08)
−0.08 (−0.13 to −0.02)
0.004
VO2 peak, ml·min−1
1.02 (0.89; 1.59)
1.01 (0.89; 1.58)
−0.007 (−0.026 to 0.012)
0.557
VO2 peak, ml·kg−1·min−1
14.7 (13.1; 17.7)
15.3 (12.7; 17.3)
−0.09 (−0.35 to 0.18)
0.400
O2 pulse, ml·beat−1
10.0 (8.0; 13.0)
10.0 (8.0; 13.0)
0.00 (−0.002 to 0.002)
0.999
VE peak, l·min−1
44.9 (36.4; 64.0)
43.6 (35.8; 62.4)
0.40 (−3.64 to 4.44)
0.550
VE/VO2 peak
36.5 (32.9; 40.9)
35.6 (32.5; 39.1)
0.90 (−2.73 to 4.53)
0.338
VE/VCO2 peak
35.5 (32.4; 36.7)
36.4 (33.4; 39.9)
2.90 (1.33 to 4.47)
0.003
BF peak, breaths·min−1
36.0 (29.5; 41.0)
37.0 (30.5; 41.5)
0.50 (−2.51 to 3.51)
0.746
BF peak, peak breath frequency; CI, confidence interval; HR peak, peak heart rate; IP, incremental phase; O2
pulse, Oxygen pulse; RER peak, peak respiratory exchange ratio, VE peak, peak ventilation; VE/VCO2 peak, peak
ventilatory equivalent for carbon dioxide; VE/VO2 peak, peak ventilatory equivalent for oxygen; VO2 peak, peak
oxygen uptake; VP, verification phase. Exercise phase data are presented as median (25th and 75th percentiles);
p values refer to within-subject comparisons (VP vs. IP); bold values refer to statistical significance (p ≤ 0.050).
Descriptive group data from the IP and VP, as well as between-phase comparisons,
are presented in Table 2. The median peak work rate during the IP was 55.0 W (46.5; 92.5).
The absolute and relative VO2 peak values did not differ between exercise phases (p =
0.557 and p = 0.400, respectively). RER and VE/VCO2 peak values were lower and higher,
respectively, in the VP than in the IP (p = 0.004 and p = 0.003). The results did not change
when exclusively male or female patients were included in the analyses. Figure 1 shows
the Bland-Altman plot for the relative VO2 peak. The mean difference between both
exercise phases was −0.07 mL·kg−1·min−1, while the lower and upper limits of agreement
were −1.59 mL·kg−1·min−1 and 1.45 mL·kg−1·min−1, respectively.
Table 2. Cardiopulmonary responses to the two exercise phases and between-phase comparisons
(n = 21).
Variable
IP
VP
Difference (95% CI)
p
Duration, min
8.5 (7.3; 12.5)
2.8 (2.1; 3.5)
−7.25 (−9.30 to −5.20)
<0.001
HR peak, beats·min−1
112.0 (108.0; 127.0)
109.0 (105.5; 128.0)
−1.00 (−5.92 to 4.92)
0.720
RER peak
1.10 (1.04; 1.12)
1.00 (0.95; 1.08)
−0.08 (−0.13 to −0.02)
0.004
VO2 peak, ml·min−1
1.02 (0.89; 1.59)
1.01 (0.89; 1.58)
−0.007 (−0.026 to 0.012)
0.557
VO2 peak, ml·kg−1·min−1
14.7 (13.1; 17.7)
15.3 (12.7; 17.3)
−0.09 (−0.35 to 0.18)
0.400
O2 pulse, ml·beat−1
10.0 (8.0; 13.0)
10.0 (8.0; 13.0)
0.00 (−0.002 to 0.002)
0.999
VE peak, l·min−1
44.9 (36.4; 64.0)
43.6 (35.8; 62.4)
0.40 (−3.64 to 4.44)
0.550
VE/VO2 peak
36.5 (32.9; 40.9)
35.6 (32.5; 39.1)
0.90 (−2.73 to 4.53)
0.338
VE/VCO2 peak
35.5 (32.4; 36.7)
36.4 (33.4; 39.9)
2.90 (1.33 to 4.47)
0.003
BF peak, breaths·min−1
36.0 (29.5; 41.0)
37.0 (30.5; 41.5)
0.50 (−2.51 to 3.51)
0.746
BF peak, peak breath frequency; CI, confidence interval; HR peak, peak heart rate; IP, incremental
phase; O2 pulse, Oxygen pulse; RER peak, peak respiratory exchange ratio, VE peak, peak
ventilation; VE/VCO2 peak, peak ventilatory equivalent for carbon dioxide; VE/VO2 peak, peak
ventilatory equivalent for oxygen; VO2 peak, peak oxygen uptake; VP, verification phase. Exercise
phase data are presented as median (25th and 75th percentiles); p values refer to within-subject
comparisons (VP vs. IP); bold values refer to statistical significance (p ≤ 0.050).
Figure 1. Bland-Altman plot for relative peak oxygen uptake response between the two exercise
phases. Dashed line represents the mean bias, and highlighted zone indices are the limits of agreement
(mean ± 1.96 standard deviation).
3.3. Individual Comparisons
An IP-derived VO2 peak was confirmed (i.e., VO2 max) in 11 (52.4%) patients and not
confirmed (i.e., VO2 peak) in 10 (47.6%). Regarding the patients in whom the VO2 peak
was attained, five showed higher IP-derived VO2 peak values and five showed lower IP-
derived VO2 peak values, compared with the VP-derived VO2 peak value. As to the patient
characteristics, comparisons showed that the proportion of smokers was higher (p = 0.035)
in the confirmed group (100%) than in the not confirmed group (60%). Interestingly, the
percentage of female participants did not differ between groups (p = 0.999). Moreover, there
were no between-group differences in any of the remaining analysed variables (p > 0.050)
(see Table 1).
Int. J. Environ. Res. Public Health 2023, 20, 2764
6 of 10
3.4. Confirmed and Not Confirmed Groups
The patients’ responses to both exercise phases in the confirmed and not confirmed
groups can be found in Table S1. Between-phase comparisons showed the same results
as those found when all patients had been included (see Table 2). On the other hand,
between-group comparisons during each exercise phase are shown in Table S2. Although
no statistically significant differences were found (p > 0.050), the relative VO2 peak value
was higher in the confirmed group compared to the not confirmed group both in the IP
(2.62 mL·kg−1·min−1 [95%CI = −2.38 to 7.62]; p = 0.305) and the VP (2.35 mL·kg−1·min−1
[95%CI = −2.90 to 7.60]; p = 0.380).
3.5. Agreement between the Traditional and Verification Phase Criteria
When the traditional criterion (i.e., RER peak) was used, VO2 max was confirmed in
10 patients and VO2 peak was attained in 11 patients. The median VO2 peak values in
the confirmed and not-confirmed groups during the IP were 16.2 mL·kg−1·min−1 (13.0;
19.9) and 14.1 mL·kg−1·min−1 (13.0; 17.6), while the median values during the VP were
15.7 mL·kg−1·min−1 (13.6; 21.6) and 13.7 mL·kg−1·min−1 (11.5; 16.8), respectively. Regard-
ing the agreement between the two criteria for the determination of VO2 max, there were
10 agreements (47.6%), six false negative cases (28.6%), and five false positive cases (23.8%).
Moreover, the Kappa index showed that there was no significant agreement between both
criteria (Kappa = −0.045; p = 0.583).
4. Discussion
The main objective of this study was to investigate whether the submaximal VP is
a safe and reliable method to validate VO2 max in male and female patients with HFrEF.
To accomplish this, we used both group and individual approaches. Additionally, we
investigated the agreement between the RER and VP criteria to determine VO2 max.
Regarding our results, no adverse events were observed during the exercise tests,
suggesting that VP is a safe method for determining VO2 max in patients with HFrEF. In
agreement with our findings, Bowen et al. [23] also reported no adverse events in patients
with HFrEF. There is also previous evidence showing that the use of the VP was well-
tolerated in patients with other diseases, such as cancer [29], prehypertension [30], and
metabolic syndrome [31], who are normally sedentary and not familiarised with high-
intensity exercise. Nonetheless, it should be noted that, in the current study, patients who
had a high risk of adverse events (those who presented symptoms of ischaemia or ectopic
heartbeats during the IP) were exempt from performing the VP. Moreover, several patients
had difficulty cycling since they were unfamiliar with the cycle ergometer. In this regard,
Manresa-Rocamora et al. [32] reported a greater improvement in the VO2 max after an
exercise-based cardiac rehabilitation programme in studies that conducted the incremental
exercise test on a cycle ergometer compared to studies that used a treadmill in patients with
coronary artery disease. The lack of habituation to the cycle ergometer could explain, in
part, the higher training-induced effect found in these studies, seeing as their baseline VO2
max results were worse than those who used a treadmill. Therefore, a familiarisation period
should be performed before conducting the incremental exercise test to avoid terminating
the test due to peripheral fatigue.
As for the use of individual versus group comparisons for the analysis of VP, con-
tradictory findings were obtained based on the type of approach used to conduct the
analyses. Group comparisons showed that both exercise phases (i.e., IP and VP) led to
similar median VO2 peak values. These results did not change when only male or female
patients were included in the analysis. Therefore, based on this approach, the VO2 peak
values reached during the IP can be considered as maximal (i.e., VO2 max) in all patients.
This finding is in line with those of Murias et al. [33] and Bowen et al. [23] in healthy
males and patients with HFrEF, respectively. In this regard, Murias et al. [33], who did
not conduct individual comparisons, concluded that both the submaximal VP (i.e., 85%
of peak power output) and supramaximal VP (i.e., 105% of peak power output) were not
Int. J. Environ. Res. Public Health 2023, 20, 2764
7 of 10
necessary to confirm the VO2 max values reached during the preceding IP. In the same line,
Astorino and Emma [22] and Costa et al. [34], who respectively conducted a review and
a meta-analysis (54 studies), reported no differences in mean VO2 peak values between
the two exercise phases in a sizable number of studies conducted with healthy adults and
individuals with pathology. Previous studies also failed to find differences between both
exercise phases in endurance-trained athletes [35,36]. Interestingly, in line with our results,
Costa et al. [34] found that the sex of the participants did not influence their results and
reported no differences in the aggregate VO2 peak values in male and female participants.
In contrast to these findings, Moreno-Cabañas et al. [31] and Schaun et al. [37] found higher
mean VO2 peaks during the VP than during the IP in male and female older adults with
obesity and hypertension, respectively. It should be noted that a supramaximal VP (i.e.,
constant load and multistage) preceded by a passive recovery period (i.e., 10–15 min) in the
seated position was conducted, which could explain in part these controversial findings. In
contrast, Costa et al. [34] reported in their meta-analysis no differences in mean VO2 peaks
regardless of the VP intensity (i.e., submaximal vs. supramaximal), type of recovery (i.e.,
active vs. passive), verification timing (i.e., same day vs. different day), and verification
phase duration (e.g., less than 80 s) in apparently healthy adults. Bhammar and Chien [30],
who conducted a supramaximal VP, also found no differences in VO2 peak values in adults
with prehypertension. Therefore, our findings and previous evidence support that the
submaximal VP is not necessary to confirm VO2 max when group comparisons are used,
while the utility of the supramaximal VP, which could lead to controversial findings, in
patients with HFrEF requires future study. Nonetheless, the achievement of a VO2 peak is
an individual phenomenon and group comparisons may cloud individual differences.
In relation to individual comparisons, our results showed that VO2 max was confirmed
by the VP in 52% of the patients with HFrEF, while a VO2 peak was attained (i.e., the
individual between-phase difference in VO2 peak values higher than 3%) in the remaining
patients (48%). Moreover, the percentage of female patients was the same in the confirmed
and not confirmed groups, suggesting that individual comparisons could be used in both
male and female patients. Nonetheless, the low number of female patients included
warrants future studies to confirm our results. Similarly, Bowen et al. [23], who only
recruited male patients, found that the VO2 max was confirmed in 58% of the patients
with HFrEF included in their study. It should be highlighted that, in contrast to our
study, statistical comparisons between both exercise phases were performed to conduct
individual comparisons and validate the VO2 max. In conclusion, regardless of the criteria
used to carry out individual comparisons, an individual approach should be prioritised to
determine the VO2 max in patients with HFrEF, in accordance with previous reports in the
literature [24,37].
On the other hand, we found no difference in median VO2 peak values between the
two exercise phases in the confirmed and not confirmed groups, which also agrees with the
results of Bowen et al. [23]. In the same line, the current and the former study showed no
difference between the two groups in aggregate VO2 peak values reached during the IP.
However, although statistical significance was not reached, both studies showed that the
group VO2 peak values achieved during the IP were higher in the confirmed group (15.9
and 15.1 mL·kg−1·min−1) than in the not confirmed group (13.3 and 13.7 mL·kg−1·min−1).
These findings seem to support a greater difference in VO2 peak values between both
exercise phases (i.e., VO2 peak attained) in patients with lower cardiorespiratory fitness.
Furthermore, Moreno-Cabañas et al. [31], who included older and less physically fit par-
ticipants with obesity, observed higher VP-derived VO2 peak in 40% of the participants,
while Wood et al. [38], who recruited younger and fitter patients with obesity, only found a
difference in VO2 peak values between the two exercise phases in 15% of the participants.
Therefore, our findings and previous evidence seem to support that patients with lower
cardiorespiratory fitness may show a greater difference in VO2 peaks between the two
exercise phases, with the use of the VP being even more important for the validation of the
VO2 max in this group of patients.
Int. J. Environ. Res. Public Health 2023, 20, 2764
8 of 10
Finally, regarding the comparison between traditional criteria (i.e., RER) and the VP for
VO2 max determination, we found no agreement between both methods, which is similar
to previous evidence [23]. Interestingly, based on the RER criterion, the VO2 max was
confirmed in five patients who showed higher VO2 peaks during the VP compared with
the IP (i.e., false positive). There is evidence showing that the RER criterion can be reached
at submaximal intensities (e.g., 80% VO2 max) [11,18], which concurs with our findings.
Moreover, Bowen et al. [23] showed a direct relationship between RER and workload
increase in patients with HFrEF. In the same line, Moreno-Cabañas et al. [31] observed
that the VO2 plateau was not reliable for determining the VO2 max in participants with
obesity. Therefore, the results of the current and previous studies confirm that traditional
criteria (e.g., RER and VO2 plateau) should not be due to their lack of validity to verify
the VO2 max.
5. Limitations
Some limitations should be mentioned. First, we did not perform a familiarisation
period with the equipment (e.g., cycle ergometer) and, consequently, some patients showed
difficulty cycling. Future studies conducted with patients who are sedentary should include
a familiarisation phase before starting the study protocol. Second, there was an uneven sex
distribution among the participants (i.e., 13 males and 8 females). Therefore, to support
our findings, additional research including more female patients with HFrEF should be
conducted. Third, no prior power analysis was conducted to estimate the optimum number
of patients the study should include.
6. Conclusions
The submaximal VP is a safe and suitable method to determine the VO2 max in
patients with HFrEF. When comparing both exercise phases, an individual approach is
preferable, seeing as aggregate comparisons could mask patients who showed differences
in VO2 peaks between both exercise phases.
Supplementary Materials: The following supporting information can be downloaded at: https:
//www.mdpi.com/article/10.3390/ijerph20042764/s1, Table S1: Cardiopulmonary responses to the
two exercise phases in the confirmed and not confirmed groups, and between-phase comparisons;
Table S2: Between-group comparisons during the two exercise phases.
Author Contributions: Conceptualization, A.M.-R., N.S.-R. and J.M.S.; Methodology, A.M.-R., L.F.-K.,
C.B.-P., N.S.-R., J.M.S. and V.C.-P.; Formal analysis, A.M.-R., L.F.-K. and C.B.-P.; Data curation, A.M.-R.,
L.F.-K. and C.B.-P.; Writing—original draft, A.M.-R., C.B.-P. and J.M.S.; Writing—review & editing,
L.F.-K. and V.C.-P. All authors have read and agreed to the published version of the manuscript.
Funding: This research was funded by the Institute of Health Carlos III (ISCIII, grant number
DTS21/00171, European Commission, FEDER funds) and by the Institute for Health and Biomedical
Research of Alicante (ISABIAL, grant number A2022-0018).
Institutional Review Board Statement: Not applicable.
Informed Consent Statement: Not applicable.
Data Availability Statement: The datasets generated from the current study are available from the
corresponding author upon reasonable request.
Conflicts of Interest: The authors declare no conflict of interest.
Int. J. Environ. Res. Public Health 2023, 20, 2764
9 of 10
References
1.
Dickstein, K.; Cohen-Solal, A.; Filippatos, G.; McMurray, J.J.; Ponikowski, P.; Poole-Wilson, P.A.; Strömberg, A.; van Veldhuisen, D.J.;
Atar, D.; Hoes, A.W.; et al. ESC guidelines for the diagnosis and treatment of acute and chronic heart failure 2008: The Task Force for
the diagnosis and treatment of acute and chronic heart failure 2008 of the European Society of Cardiology. Developed in collaboration
with the Heart Failure Association of the ESC (HFA) and endorsed by the European Society of Intensive Care Medicine (ESICM). Eur.
J. Heart Fail 2008, 10, 933–989. [CrossRef] [PubMed]
2.
Members, W.C.; Hunt, S.A.; Abraham, W.T.; Chin, M.H.; Feldman, A.M.; Francis, G.S.; Ganiats, T.G.; Jessup, M.; Konstam, M.A.;
Mancini, D.M.; et al. 2009 focused update incorporated into the ACC/AHA 2005 Guidelines for the Diagnosis and Management
of Heart Failure in Adults: A report of the American College of Cardiology Foundation/American Heart Association Task Force
on Practice Guidelines: Developed in collaboration with the International Society for Heart and Lung Transplantation. Circulation
2009, 119, e391–e479. [CrossRef] [PubMed]
3.
McMurray, J.J.; Adamopoulos, S.; Anker, S.D.; Auricchio, A.; Böhm, M.; Dickstein, K.; Falk, V.; Filippatos, G.; Fonseca, C.;
Gomez-Sanchez, M.A.; et al. ESC Guidelines for the diagnosis and treatment of acute and chronic heart failure 2012: The Task
Force for the Diagnosis and Treatment of Acute and Chronic Heart Failure 2012 of the European Society of Cardiology. Developed
in collaboration with the Heart Failure Association (HFA) of the ESC. Eur. Heart J. 2012, 33, 1787–1847. [CrossRef] [PubMed]
4.
Mancini, D.M.; Eisen, H.; Kussmaul, W.; Mull, R.; Edmunds, L.H.; Jr Wilson, J.R. Value of peak exercise oxygen consumption for
optimal timing of cardiac transplantation in ambulatory patients with heart failure. Circulation 1991, 83, 778–786. [CrossRef]
[PubMed]
5.
Myers, J.; Gullestad, L.; Vagelos, R.; Do, D.; Bellin, D.; Ross, H.; Fowler, M.B. Cardiopulmonary exercise testing and prognosis in
severe heart failure: 14 mL/kg/min revisited. Am. Heart J. 2000, 139 (Pt 1), 78–84. [CrossRef]
6.
Francis, D.; Shamim, W.; Davies, L.C.; Piepoli, M.; Ponikowski, P.; Anker, S.; Coats, A. Cardiopulmonary exercise testing for
prognosis in chronic heart failure: Continuous and independent prognostic value from VE/VCO(2)slope and peak VO(2). Eur.
Heart J. 2000, 21, 154–161. [CrossRef] [PubMed]
7.
Wagner, P.D. Determinants of maximal oxygen transport and utilization. Annu. Rev. Physiol. 1996, 58, 21–50. [CrossRef]
8.
O’Neill, J.O.; Young, J.B.; Pothier, C.E.; Lauer, M.S. Peak oxygen consumption as a predictor of death in patients with heart failure
receiving beta-blockers. Circulation 2005, 111, 2313–2318. [CrossRef]
9.
Ross, R.; Blair, S.N.; Arena, R.; Church, T.S.; Després, J.P.; Franklin, B.A.; Haskell, W.L.; Kaminsky, L.A.; Levine, B.D.;
Lavie, C.J.; et al. Importance of Assessing Cardiorespiratory Fitness in Clinical Practice: A Case for Fitness as a Clinical Vi-
tal Sign: A Scientific Statement From the American Heart Association. Circulation 2016, 134, e653–e699. [CrossRef]
10.
Astorino, T.A. Alterations in VOmax and the VO plateau with manipulation of sampling interval. Clin. Physiol. Funct. Imaging
2009, 29, 60–67. [CrossRef]
11.
Poole, D.C.; Jones, A.M. Measurement of the maximum oxygen uptake
.
Vo(2max):
.
Vo(2peak) is no longer acceptable. J. Appl.
Physiol. 2017, 122, 997–1002. [CrossRef]
12.
Midgley, A.W.; McNaughton, L.R.; Polman, R.; Marchant, D. Criteria for determination of maximal oxygen uptake: A brief
critique and recommendations for future research. Sports Med. 2007, 37, 1019–1028. [CrossRef] [PubMed]
13.
Poole, D.C.; Wilkerson, D.P.; Jones, A.M. Validity of criteria for establishing maximal O2 uptake during ramp exercise tests. Eur. J.
Appl. Physiol. 2008, 102, 403–410. [CrossRef] [PubMed]
14.
Niemelä, K.; Palatsi, I.; Linnaluoto, M.; Takkunen, J. Criteria for maximum oxygen uptake in progressive bicycle tests. Eur. J. Appl.
Physiol. Occup. Physiol. 1980, 44, 51–59. [CrossRef] [PubMed]
15.
Day, J.R.; Rossiter, H.B.; Coats, E.M.; Skasick, A.; Whipp, B.J. The maximally attainable VO2 during exercise in humans: The peak
vs. maximum issue. J. Appl. Physiol. 2003, 95, 1901–1907. [CrossRef]
16.
Howley, E.T.; Bassett, D.R., Jr.; Welch, H.G. Criteria for maximal oxygen uptake: Review and commentary. Med. Sci. Sports Exerc.
1995, 27, 1292–1301. [CrossRef]
17.
Mezzani, A.; Agostoni, P.; Cohen-Solal, A.; Corrà, U.; Jegier, A.; Kouidi, E.; Mazic, S.; Meurin, P.; Piepoli, M.; Simon, A.; et al.
Standards for the use of cardiopulmonary exercise testing for the functional evaluation of cardiac patients: A report from the
Exercise Physiology Section of the European Association for Cardiovascular Prevention and Rehabilitation. Eur. J. Cardiovasc.
Prev. Rehabil. 2009, 16, 249–267. [CrossRef]
18.
Beltz, N.M.; Gibson, A.L.; Janot, J.M.; Kravitz, L.; Mermier, C.M.; Dalleck, L.C. Graded Exercise Testing Protocols for the
Determination of VO(2)max: Historical Perspectives, Progress, and Future Considerations. J. Sports Med. 2016, 2016, 3968393.
[CrossRef]
19.
Midgley, A.W.; Carroll, S. Emergence of the verification phase procedure for confirming ’true’ VO(2max). Scand. J. Med. Sci.
Sports 2009, 19, 313–322. [CrossRef]
20.
Schaun, G.Z. The Maximal Oxygen Uptake Verification Phase: A Light at the End of the Tunnel? Sports Med. Open 2017, 3, 44.
[CrossRef]
21.
Villanueva, I.R.; Campbell, J.C.; Medina, S.M.; Jorgensen, T.M.; Wilson, S.L.; Angadi, S.S.; Gaesser, G.A.; Dickinson, J.M.
Comparison of constant load exercise intensity for verification of maximal oxygen uptake following a graded exercise test in
older adults. Physiol. Rep. 2021, 9, e15037. [CrossRef]
22.
Astorino, T.A.; Emma, D. Utility of Verification Testing to Confirm Attainment of Maximal Oxygen Uptake in Unhealthy
Participants: A Perspective Review. Sports 2021, 9, 9080108. [CrossRef] [PubMed]
Int. J. Environ. Res. Public Health 2023, 20, 2764
10 of 10
23.
Bowen, T.S.; Cannon, D.T.; Begg, G.; Baliga, V.; Witte, K.K.; Rossiter, H.B. A novel cardiopulmonary exercise test protocol and
criterion to determine maximal oxygen uptake in chronic heart failure. J. Appl. Physiol. 2012, 113, 451–458. [CrossRef]
24.
Noakes, T.D. Maximal oxygen uptake as a parametric measure of cardiorespiratory capacity: Comment. Med. Sci. Sports Exerc.
2008, 40, 585–586. [CrossRef] [PubMed]
25.
Dixon, P.M.; Saint-Maurice, P.F.; Kim, Y.; Hibbing, P.; Bai, Y.; Welk, G.J. A Primer on the Use of Equivalence Testing for Evaluating
Measurement Agreement. Med. Sci. Sports Exerc. 2018, 50, 837–845. [CrossRef] [PubMed]
26.
Lamarra, N.; Whipp, B.J.; Ward, S.A.; Wasserman, K. Effect of interbreath fluctuations on characterizing exercise gas exchange
kinetics. J. Appl. Physiol. 1987, 62, 2003–2012. [CrossRef]
27.
Kaiser, J. An exact and a Monte Carlo proposal to the Fisher–Pitman permutation tests for paired replicates and for independent
samples. Stata J. 2007, 7, 402–412. [CrossRef]
28.
Campbell, M.J.; Gardner, M.J. Calculating confidence intervals for some non-parametric analyses. Br. Med. J. (Clin. Res. Ed.) 1988,
296, 1454–1456. [CrossRef]
29.
Schneider, J.; Schlüter, K.; Wiskemann, J.; Rosenberger, F. Do we underestimate maximal oxygen uptake in cancer survivors?
Findings from a supramaximal verification test. Appl. Physiol. Nutr. Metab. 2020, 45, 486–492. [CrossRef]
30.
Bhammar, D.M.; Chien, L.C. Quantification and Verification of Cardiorespiratory Fitness in Adults with Prehypertension. Sports
2021, 9, 9010009. [CrossRef]
31.
Moreno-Cabañas, A.; Ortega, J.F.; Morales-Palomo, F.; Ramirez-Jimenez, M.; Mora-Rodriguez, R. Importance of a verification test
to accurately assess
.
VO(2) max in unfit individuals with obesity. Scand J. Med. Sci. Sports 2020, 30, 583–590. [CrossRef] [PubMed]
32.
Manresa-Rocamora, A.; Sarabia, J.M.; Sánchez-Meca, J.; Oliveira, J.; Vera-Garcia, F.J.; Moya-Ramón, M. Are the Current Cardiac
Rehabilitation Programs Optimized to Improve Cardiorespiratory Fitness in Patients? A Meta-Analysis. J. Aging Phys. Act. 2021,
29, 327–342. [CrossRef] [PubMed]
33.
Murias, J.M.; Pogliaghi, S.; Paterson, D.H. Measurement of a True [Formula: See text]O(2max) during a Ramp Incremental Test Is
Not Confirmed by a Verification Phase. Front. Physiol. 2018, 9, 143. [CrossRef] [PubMed]
34.
Costa, V.A.B.; Midgley, A.W.; Carroll, S.; Astorino, T.A.; de Paula, T.; Farinatti, P.; Cunha, F.A. Is a verification phase useful for
confirming maximal oxygen uptake in apparently healthy adults? A systematic review and meta-analysis. PLoS ONE 2021, 16,
e0247057. [CrossRef] [PubMed]
35.
Midgley, A.W.; Carroll, S.; Marchant, D.; McNaughton, L.R.; Siegler, J. Evaluation of true maximal oxygen uptake based on a
novel set of standardized criteria. Appl. Physiol. Nutr. Metab. 2009, 34, 115–123. [CrossRef]
36.
Foster, C.; Kuffel, E.; Bradley, N.; Battista, R.A.; Wright, G.; Porcari, J.P.; Lucia, A.; Dekoning, J.J. VO2max during successive
maximal efforts. Eur. J. Appl. Physiol. 2007, 102, 67–72. [CrossRef]
37.
Schaun, G.Z.; Alberton, C.L.; Gomes, M.L.B.; Santos, L.P.; Bamman, M.M.; Mendes, G.F.; Häfele, M.S.; Andrade, L.S.; Alves, L.;
DE Ataides, V.A.; et al. Maximal Oxygen Uptake Is Underestimated during Incremental Testing in Hypertensive Older Adults:
Findings from the HAEL Study. Med. Sci. Sports Exerc. 2021, 53, 1452–1459. [CrossRef]
38.
Wood, R.E.; Hills, A.P.; Hunter, G.R.; King, N.A.; Byrne, N.M. Vo2max in overweight and obese adults: Do they meet the threshold
criteria? Med. Sci. Sports Exerc. 2010, 42, 470–477. [CrossRef]
Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual
author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to
people or property resulting from any ideas, methods, instructions or products referred to in the content.
| Is the Verification Phase a Suitable Criterion for the Determination of Maximum Oxygen Uptake in Patients with Heart Failure and Reduced Ejection Fraction? A Validation Study. | 02-04-2023 | Manresa-Rocamora, Agustín,Fuertes-Kenneally, Laura,Blasco-Peris, Carles,Sempere-Ruiz, Noemí,Sarabia, José Manuel,Climent-Paya, Vicente | eng |
PMC6239296 | S2 Appendix. Comparison to oxygen uptake
measurement
While it is assuring to see below that our model can explain and predict record and
individual racing times, a more direct comparison to power output during running is
desirable to probe the logarithmic decline of the maximal power output with exercise
duration, as predicted by Eq. (6). This is of particular importance in the anaerobic
range where different functional forms, e.g., exponential decays, have been proposed [6].
However, running power, as measured by oxygen utilization, can be directly determined
only in the aerobic regime. For (supra-maximal) exercise with substantial contributions
from anaerobic systems where power output exceeds maximal oxygen uptake, Medbo et
al. showed that the oxygen demand can be estimated by extrapolating each runner’s
individual nominal linear relationship between running speed and submaximal oxygen
uptake [1]. The difference between the extrapolated oxygen utilization and the
measured oxygen uptake is the accumulated oxygen deficit. Using this method, Medbo
et al. determined from treadmill exercise at speeds that caused exhaustion within
different predetermined durations the oxygen demand relative to the maximal uptake.
Translated to percent of maximal aerobic power output, this oxygen demand is given by
100 × Pmax(T)/Pm in our model, with Pmax(T) given in Eq. (6).
While a logarithmic dependence for Pmax(T) has been deduced from purely
empirical data analyses for world records for times above tc before [6], to our knowledge
a logarithmic scaling has not been proposed for shorter exercise with large anaerobic
involvement. Hence, it is interesting that there exists experimental estimates of the
maximal oxygen utilization that can be maintained for a given duration. As explained
above, Medbo et al. [1] obtained for 11 runners data that correspond to
100 × Pmax(T)/Pm which is shown as function of T < 5min ∼ tc in Fig. A. We have
fitted the prediction of our model to the data, and the results for the runner with
smallest and largest oxygen demand are shown in the same figure. The agreement
between the data and our model prediction appears to be rather convincing. This
suggests that there exists indeed a logarithmic relation between maximally sustainable
power and duration in the range of supra-maximal intensities, resembling observation
that were made before in the sub-maximal zone.
PLOS
1/2
Fig A. Relative nominal oxygen demand as function of the maximum
duration over which it can be sustained. Original plot and data for 11
runners from Ref. [1]. The two curves are fits of Eq. (6) to the data for the
runners with smallest and largest relative oxygen demand.
References
1. Medbo JI, Mohn AC, Tabata I, Bahr R, Vaage O, Sejersted OM. Anaerobic
capacity determined by maximal accumulated O2 deficit. J Appl Physiol.
1988;64(1):50–60.
PLOS
2/2
| A minimal power model for human running performance. | 11-16-2018 | Mulligan, Matthew,Adam, Guillaume,Emig, Thorsten | eng |
PMC10739691 |
Supplementary Online Material
Anaerobic Threshold Using Sweat Lactate Sensor under Hypoxia
Hiroki Okawara, Yuji Iwasawa, Tomonori Sawada, Kazuhisa Sugai, Kyohei Daigo, Yuta Seki, Genki Ichihara, Daisuke Nakashima, Motoaki Sano, Masaya Nakamura, Kazuki
Sato, Keiichi Fukuda, Yoshinori Katsumata
This supplementary material has been provided by the authors to give readers additional information about their work.
APPENDIX
Supplementary Figure 1. Imaging of sweat lactate levels, local sweat rate, and blood lactate values during incremental exercise under normoxia
Representative graphs of sweat lactate levels (orange), local sweat rate (blue), and blood lactate values (red) during hypoxic exercise with a stepwise incremental protocol (25
W/min) ergometer are shown.
Abbreviations: VT=ventilatory threshold; sLT=sweat lactate threshold.
Supplementary Figure 2. Measured parameters in normoxia.
The graph shows the measured parameters (a; VO2/Body weight, b; Heart rate, c; Sweat lactate, d; sweat rate) at rest, warm up, VT, and peak in hypoxia. Data are shown as
mean (±standard deviation).
Abbreviations: VO2=oxygen uptake: VT=ventilatory threshold: HR=heart rate: sLA=sweat lactate: SR=sweat rate.
a) VO2
c) sLA
d) SR
b) HR
Supplementary Figure 2
Supplementary Figure 3. Reliability testing of the time at sLT determined by the same evaluator in normoxia.
(a) The graph shows the relationship between the repeatedly determined sweat lactate threshold (sLT) by the same evaluator (b) The graph shows the Bland–Altman plots,
which indicate the respective differences between the repeatedly determined sLT by the same evaluator (y-axis) for each individual against the mean of the time at the
repeatedly determined sLT (x-axis) in normoxia. R, correlation coefficient; p, p-value; ventilatory threshold; sLT, sweat lactate threshold.
Supplementary Figure 3
a)
b)
R = 0.70
p < 0.01
Supplementary Figure 4. Validity testing of the time at VT and sLT in normoxia
(a) The graph shows the relationship between the time from the start of the measurement (seconds) at VT and sLT. (b) The graph shows the Bland–Altman plots, which
indicate the respective differences between the time from the start of measurement (s) at the VT and sLT (y-axis) for each individual against the mean of the time at the VT
and sLT (x-axis) in hypoxia. R, correlation coefficient; VT, ventilatory threshold; sLT, sweat lactate threshold.
pp
y
g
a)
b)
R = 0.69
p < 0.01
Supplementary Table 1. Intra-evaluator reliability of sweat lactate threshold determination in normoxia
Hypoxia
N
Evaluator 1
Evaluator 2
Evaluator 3
ICC (95%CI)
sLT [sec]
Mean
20
553.3
486.3
533.6
0.782 (0.607 - 0.898)
SD
84.4
89.8
80.8
bLT [sec]
Mean
20
643.9
605.6
611.3
0.621 (0.363 - 0.813)
SD
77.1
67.4
81.9
VT [sec]
Mean
20
563.1
552.2
552.6
0.711 (0.500 - 0.861)
SD
56.5
45.3
60.0
ICC, intraclass correlation; sLT, sweat lactate threshold; bLT, blood lactate threshold; VT, ventilatory threshold;
SD, standard deviation.
| Anaerobic threshold using sweat lactate sensor under hypoxia. | 12-21-2023 | Okawara, Hiroki,Iwasawa, Yuji,Sawada, Tomonori,Sugai, Kazuhisa,Daigo, Kyohei,Seki, Yuta,Ichihara, Genki,Nakashima, Daisuke,Sano, Motoaki,Nakamura, Masaya,Sato, Kazuki,Fukuda, Keiichi,Katsumata, Yoshinori | eng |
PMC8059023 | Identification of heart rate dynamics
during treadmill exercise: comparison of first‑
and second‑order models
Hanjie Wang* and Kenneth J. Hunt
Background
Characterisation of heart rate (HR) dynamics with respect to changes in exer-
cise intensity provides models that can be used to synthesise control algorithms to
maintain target HR levels [1]. The control of HR is important in the design of train-
ing protocols that aim both to maintain and to improve cardiorespiratory fitness;
this applies to healthy individuals [2] and also in different patient populations [3,
Abstract
Background: Characterisation of heart rate (HR) dynamics and their dependence on
exercise intensity provides a basis for feedback design of automatic HR control systems.
This work aimed to investigate whether the second-order models with separate Phase
I and Phase II components of HR response can achieve better fitting performance com-
pared to the first-order models that do not delineate the two phases.
Methods: Eleven participants each performed two open-loop identification tests
while running at moderate-to-vigorous intensity on a treadmill. Treadmill speed was
changed as a pseudo-random binary sequence (PRBS) to excite both the Phase I and
Phase II components. A counterbalanced cross-validation approach was implemented
for model parameter estimation and validation.
Results: Comparison of validation outcomes for 22 pairs of first- and second-order
models showed that root-mean-square error (RMSE) was significantly lower and
fit (normalised RMSE) significantly higher for the second-order models: RMSE was
2.07 bpm ± 0.36 bpm vs. 2.27 bpm ± 0.36 bpm (bpm = beats per min), second
order vs. first order, with p = 2.8 × 10−10 ; fit was 54.5% ± 5.2 % vs. 50.2% ± 4.8 %,
p = 6.8 × 10−10.
Conclusion: Second-order models give significantly better goodness-of-fit than first-
order models, likely due to the inclusion of both Phase I and Phase II components of
heart rate response. Future work should investigate alternative parameterisations of
the PRBS excitation, and whether feedback controllers calculated using second-order
models give better performance than those based on first-order models.
Keywords: Heart rate dynamics, System identification, Treadmills
Open Access
© The Author(s) 2021. This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing,
adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the
source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this
article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not
included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted
use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/
licen ses/ by/4. 0/. The Creative Commons Public Domain Dedication waiver (http:// creat iveco mmons. org/ publi cdoma in/ zero/1. 0/) applies
to the data made available in this article, unless otherwise stated in a credit line to the data.
RESEARCH
Wang and Hunt BioMed Eng OnLine (2021) 20:37
https://doi.org/10.1186/s12938‑021‑00875‑7
BioMedical Engineering
OnLine
*Correspondence:
[email protected]
Department of Engineering
and Information Technology,
Division of Mechanical
Engineering, Institute
for Rehabilitation
and Performance Technology,
Bern University of Applied
Sciences, 3400 Burgdorf,
Switzerland
Page 2 of 10
Wang and Hunt BioMed Eng OnLine (2021) 20:37
4]. Target heart rate profiles come in various forms such as high-intensity inter-
val training (HIIT) that repeats high-intensity exercise connected by low-intensity
recovery intervals; HIIT has potential to enhance cardiovascular health and fitness
when compared to training at constant work rates (systematic reviews: [5, 6]).
Recent work investigated the effect of exercise intensity and time on HR dynamics
using first-order models [7], but it may be beneficial to include higher order effects:
based on physiological study, the dynamics of both oxygen uptake and HR responses
to changes in exercise intensity are known to have three distinct phases [8]. These
are: (i) a Phase I component lasting ∼ 15 s with a relatively small-magnitude venti-
latory response, but where HR can increase by about 50% of its total response [9];
(ii) a Phase II component between around 15 s and 3 min contributing the further
increase of cardiopulmonary response; and then, (iii) if the applied exercise inten-
sity exceeds the anaerobic threshold, a Phase III component is prolonged and rises
slowly. The three components can each be modelled as single exponentials (first-
order systems) each with their own time delay, gain, and time constant [10]. In addi-
tion to these primary dynamic responses, the phenomenon of heart rate variability
(HRV) can be added to the model to represent the regulatory activities of the auto-
nomic nervous system; in the context of feedback control of HR, HRV represents a
broad-spectrum disturbance term [1].
Because it can be challenging to estimate the separate Phase I and II components
using data which is noisy, those two phases have often been identified as a com-
bined single exponential model with a time constant termed the mean response time
(MRT, [8]), which is effectively the first-order approach taken in the previous studies
that focused on system identification [7] and feedback control [1]. In feedback con-
trol, the slow Phase III component can readily be neglected as it is compensated by
inclusion of an integrator in the controller. The focus of the present work is there-
fore the investigation of whether the separate identification of Phase I and II compo-
nents, i.e., the employment of a second-order model, can give better model fidelity.
Other recent approaches to HR dynamics identification focused mainly on the
modelling of the Phase II and III components of the HR response. Several studies
employed a non-linear state-space model structure comprising two different states
( x1 and x2 ) to separately describe the Phase II and Phase III dynamics [11–15]. Other
work used linear time-varying systems to model the slow Phase III dynamic [16,
17]. While inclusion of Phase III may improve overall model fidelity, it will, as noted
above, have negligible impact on feedback-control performance as it will be elimi-
nated by the integral action. In contrast, it can be anticipated that separate mod-
elling of the Phase I and II components might lead to better control performance
when the model is used as the basis of an analytical, model-based feedback design.
To this end, this work aimed to investigate whether second-order models with sep-
arate Phase I and Phase II components of HR response can achieve better fitting
performance compared to first-order models that do not delineate the two phases.
Here, an input signal of PRBS (pseudo-random binary sequence) form was designed
to excite both the Phase I and Phase II components.
Page 3 of 10
Wang and Hunt BioMed Eng OnLine (2021) 20:37
Results
To illustrate the procedures of data preprocessing and model validation, an exemplary
result from participant P04 is shown (Fig. 1); the raw data for the same participant are
shown above below in the section ‘Method’. For this example, the second-order model P2
gave better performance than the first-order model P1 : fit was 51.9% vs. 50.9% ( P2 vs. P1 )
and RMSE was 2.01 bpm vs. 2.05 bpm.
The overall statistical comparison of validation outcomes for the 22 pairs of first- and
second-order models showed that RMSE was significantly lower and fit significantly
higher for the second-order models: RMSE was 2.07 bpm ± 0.36 bpm vs. 2.27 bpm ±
0.36 bpm, P2 vs. P1 , with p = 2.8 × 10−10 (Table 1; Fig. 2a); fit was 54.5% ± 5.2 % vs.
50.2% ± 4.8 %, p = 6.8 × 10−10 (Table 1; Fig. 2b). The graphical illustration of overall
outcomes (Fig. 2) shows how widely individual samples and their differences are dis-
persed, together with means and their 95% confidence intervals (CIs). These plots allow
visual determination of significant differences, if they exist: whenever there is a signifi-
cant difference, the value 0 will not be contained within the corresponding CI.
The sample size was estimated a priori by a statistical power calculation that used esti-
mates of expected effect sizes and sample standard deviations, with significance level set
to 5% ( α = 0.05) and with a statistical power of 80% ( 1 − β = 0.8).
The observed outcomes show large effect sizes (approximately 9% for both outcomes)
and extremely low p values (on the order of 10−10 ), thus pointing to a well-powered sta-
tistical analysis. In fact, post hoc statistical power analysis based on observed effect sizes
and sample dispersions gives an observed power of 100% for both outcomes.
A graphical illustration of the dispersion of estimated model parameters for the 22
first- and 22 second-order models is provided (Fig. 3). The overall first- and second-
order models were obtained by averaging the individual gains and time constants. For
the first-order models, the overall gain was k1 = 28.57 bpm/(m/s) ± 5.27 bpm/(m/s)
time/s
300
600
900
1200
1500
1800
-20
-10
0
10
20
heart rate/bpm
300
600
900
1200
1500
1800
-0.5
-0.25
0
0.25
0.5
speed/(m/s)
Fig. 1 Data preprocessing and model validation: exemplary data for participant P04 (the raw data for this
test are shown in section ‘Method’). Upper plot: HR measurement from validation data set after detrending
(solid black line), simulated HR response of first-order model ( P1sim , blue dashed line), and simulated HR
response of second-order model ( P2sim , green dashed line). Lower plot: treadmill speed from validation data
set after mean removal
Page 4 of 10
Wang and Hunt BioMed Eng OnLine (2021) 20:37
(mean ± standard deviation) while the time constant was τ1 = 70.56 s ± 16.84 s. For the
second-order models, the overall gain was k2 = 24.70 bpm/(m/s) ± 5.07 bpm/(m/s) and
the overall time constants were τ21 = 18.60 s ± 7.88 s and τ22 = 37.95 s ± 16.01 s. This
gives the average transfer functions for first- and second-order models as follows:
Discussion
This study aimed to investigate whether second-order models with separate Phase
I and Phase II components of heart rate response can achieve better fitting perfor-
mance compared to first-order models that do not delineate the two components.
(1)
u → y: P1(s) =
28.57
70.56s + 1,
(2)
u → y: P2(s) =
24.70
(18.60s + 1)(37.95s + 1).
Table 1 Overall outcomes for first- and second-order models and comparison of outcome
differences (see also Fig. 2)
n = 22
P1 first‑order models, P2 second‑order models, SD standard deviation, MD mean difference, 95% CI confidence interval for
the mean difference , p-value paired one‑sided t tests, RMSE root‑mean‑square error, fit normalised root‑mean‑square error,
bpm beats per min
Mean ± SD
MD (95% CI)
p-value
P1
P2
P2 − P1
RMSE/bpm
2.27 ± 0.36
2.07 ± 0.36
−0.19 ( −∞ , −0.16)
2.8 × 10−10
fit/%
50.2 ± 4.8
54.5 ± 5.2
4.3 (3.6, +∞)
6.8 × 10−10
****
a Root-mean-square error, RMSE.
****
b Normalised root-mean-square error, fit.
Fig. 2 Primary outcomes: data samples and differences for RMSE and fit between 22 first-order models, P1 ,
and 22 second-order models, P2 (see also Table 1). Sample pairs for each participant are connected by green
lines; mean values are shown as red horizontal bars (with numerical values given in Table 1). Sample-pair
differences are shown as D ( P2 − P1 ). The mean difference (MD) is depicted as a red bar and the blue arrow
is the corresponding 95% confidence interval (CI). For both RMSE and fit, the 95% CI does not contain the
value 0, thus showing a significant improvement for P2 vs. P1 ( p < 0.05 , Table 1; the notation **** denotes
p < 0.0001)
Page 5 of 10
Wang and Hunt BioMed Eng OnLine (2021) 20:37
The results clearly demonstrate that second-order models give significantly better
goodness-of-fit, in terms of both RMSE and fit (NRMSE): RMSE was on average
10
15
20
25
30
35
40
45
50
0
20
40
60
80
100
120
first-order models
average model
a First-order models.
10
15
20
25
30
35
40
45
50
0
20
40
60
80
second-order models
average model
10
15
20
25
30
35
40
45
50
0
20
40
60
80
second-order models
average model
b Second-order models.
Fig. 3 Dispersion of estimated model parameters for 22 first- and 22 second-order models. The stars depict
the average models. The 95% confidence intervals for the mean gains and time constants are shown as
rectangular boxes
-25
-10
speed/(m/s)
evaluation period (290 s to 2085 s )
0.5 m/s
time/min
PRBS (schematic)
warm up
rest
formal measurement phase
cool down
0
10
20
30
40
50
-
5
2.0
+
5
2.0
a
0
300
600
900
1200
1500
1800
2100
100
150
200
heart rate/bpm
0
300
600
900
1200
1500
1800
2100
time/s
1
1.5
2
2.5
3
speed/(m/s)
b
Fig. 4 Identification test protocol. a Test phases and treadmill speed. b Original data record from one
participant (P04; upper plot—HR measurement; lower plot—speed of the treadmill); the evaluation period is
depicted by the red horizontal bar
Page 6 of 10
Wang and Hunt BioMed Eng OnLine (2021) 20:37
0.19 bpm lower and fit 4.3% higher for the second-order model structure (p values
were on the order of 10−10 in both cases); that these significance levels were achieved
with a small sample size of only 11 participants underline the difference.
The approach taken here focused on control-orientated model structures, in the
sense that the estimated models would be intended to be used for analytical (model-
based) design of heart rate control systems. For this reason, slow Phase III com-
ponents in the data were eliminated by detrending prior to parameter estimation.
This is consistent with feedback-control scenarios where slowly drifting Phase III
variations in heart rate are automatically compensated using integral action in the
controller.
A further difference between the methodology employed here and heart rate mod-
elling approaches taken in the physiological literature, [8], is that a nominal operat-
ing point was assumed, and small deviations around this point were considered (in
this case, the operating point was set at the transition between exercise levels con-
sidered to be moderate and vigorous). This is consistent with linear feedback design
approaches, which are implicitly based on models that are small-signal linearisations
around an operating point; the purpose of feedback control is indeed to maintain the
controlled variable, viz., heart rate, close to a target level.
For these reasons, it is not possible to compare the overall estimated model param-
eters (gains and time constants, Eqs. (1) and (2) with values given in the physiologi-
cal literature (e.g., [9, 10]), because, there, responses are usually recorded using large
steps from a resting or low-intensity baseline.
A consequence of the control-orientated methodology followed here is that the
design of the PRBS input signal becomes important. For non-linear systems, it is
known that the parameters of linear approximations are input dependent [18], which
motivates further work to explore the effect of PRBS amplitude and frequency con-
tent on model fidelity; in particular, it is important to focus the information content
on frequencies around the intended crossover band of the closed-loop system [19].
Future work should investigate whether the observed improvement in model fidel-
ity translates into better feedback-control performance, i.e., whether controllers
designed on the basis of second-order models perform better, in some sense, than
those designed using first-order models. Because of the fundamental property of
feedback that plant uncertainty (including modelling error) is reduced, the answer
to this question will likely not be as clear cut as in the open-loop identification case.
Conclusions
Second-order models give significantly better goodness-of-fit than first-order mod-
els, likely due to the inclusion of both Phase I and Phase II components of heart rate
response. Future work should investigate alternative parameterisations of the PRBS
excitation, and whether feedback controllers calculated using second-order models
give better performance than those based on first-order models.
Page 7 of 10
Wang and Hunt BioMed Eng OnLine (2021) 20:37
Methods
Participants
Eleven healthy participants were recruited (8 males, 3 females) with age 32.5 years
± 12.3 years (mean ± standard deviation), body mass 75.5 kg ± 14.4 kg, and height
179 cm ± 12 cm. For inclusion, each participant was required to be a regular exerciser
(30-min bouts, 3 times per week) and non-smoker, and to be free of injury and illness.
Test protocols
To generate separate estimation and validation data sets, each participant took part
in two identification tests; there was an interval of at least 48 h between the two tests.
Before each test, participants were asked to meet the following requirements: refrain
from strenuous activity for 24 h, caffeine for 12 h, avoid large meals for 3 h. Each test
session had four phases: a 15 min warm up, a 10 min rest, a 36 min formal measure-
ment, and a 10 min cool down (Fig. 4a).
In the warm up, a feedback-control system was employed to automatically regulate
the speed of the treadmill to maintain a constant target HR. The target HR, denoted
HRref , was computed individually for each participant and corresponded to the HR
at the transition between intensity levels considered to be moderate or vigorous [3],
as follows: HRref = 0.765 × (220 − age) [beats/min, bpm] (except for participant P03,
for whom the factor 0.7 was used, because 0.765 led to HR remaining in the vigor-
ous-intensity regime). The mean speed of the treadmill during the final 2 min of the
warm up phase was subsequently used as the mid-level speed, denoted vm , for the
next phase.
In the formal measurement phase, the speed of the treadmill, denoted v, was
designed as a fifth-order PRBS with mean speed vm and amplitude 0.25 m/s, i.e.,
v = vm ± 0.25 m/s (to illustrate, a single original data record is provided; Fig. 4b).
Model parameter estimation and validation was performed over a full cycle of the
PRBS using an evaluation period from 290 s to 2085 s (Fig. 4); the first 5 min were
excluded to eliminate the initial transient. During the cool down phase, the speed of
the treadmill was kept constant at v = vm − 0.5 m/s.
Equipment
All tests were carried out using a treadmill (model Venus, h/p/cosmos Sports & Med-
ical GmbH, Germany) controlled by a PC running real-time Matlab/Simulink (The
MathWorks, Inc., USA). HR recording was performed with a chest strap (H10, Polar
Electro Oy, Finland) and a wireless receiver (Heart rate Monitor Interface, Spark-
fun Electronics, USA) connected to the Simulink model via a serial port. HR meas-
urements were received at a rate of 1 Hz and then downsampled to a sample rate of
0.2 Hz (sample period 5 s) by averaging consecutive batches of five individual samples.
Data preprocessing, model identification, and outcome measures
As noted above, each participant completed two identification tests, thus gener-
ating individual data sets (I and II) for model parameter estimation and validation.
To prevent over-fitting and to eliminate potential order-of-presentation effects, a
Page 8 of 10
Wang and Hunt BioMed Eng OnLine (2021) 20:37
counterbalanced cross-validation approach was implemented: for each participant,
data set I was used to estimate model parameters and data set II was used as valida-
tion data for the estimated models; then, for the same participant, data set II was used
for model estimation and data set I for validation. Thus, for the 11 participants, a total
of 22 estimation data sets and 22 validation data sets were obtained.
According to the test protocol (Sect. 5.2, Fig. 4a), an evaluation interval from 290 s to
2085 s was used to estimate and validate model parameters. This interval, within one
single PRBS period, was selected, such that the number of samples where the input
was high ( v = vm + 0.25m/s) equalled the number of samples where the input was low
( v = vm − 0.25m/s). Here, on the evaluation period from 290 s to 2085 s and with a sam-
ple period of 5 s, the total number of samples was N = 360, thus giving 180 low samples
and 180 high samples.
To remove any potential drifting Phase III dynamic of the HR response, the mean
value and any trend were removed (Matlab “detrend” function) prior to estimation and
validation; the mean value of the input signal was also removed. An exemplary data set
following this preprocessing procedure is provided (Fig. 1), with raw data are shown
above (Fig. 4b).
For each estimation data set, two linear time-invariant transfer functions were
employed to model the dynamic response from treadmill speed to HR: a first-order
transfer function (Eq. 3) which combined Phases I and II into a single time constant, and
a second-order transfer function (Eq. 4) with separate time constants for Phases I and
II. Hence, for the 11 participants, a total of 22 first-order models and 22 second-order
models were estimated:
Here, k1 and k2 are steady-state gains, and τ1 , τ21 , and τ22 are time constants. Model
parameters were calculated from the estimation data sets using a least-squares optimisa-
tion procedure (“procest” function from the Matlab System Identification Toolbox; The
Mathworks, Inc., USA).
After model estimation, the corresponding validation data sets were used to compute
goodness-of-fit measures for the resulting first- and second-order models. Two outcome
measures were used: the normalised root-mean-square error [denoted fit, Eq. (5)], and
the root-mean-square error [denoted RMSE, Eq. (6)], as follows:
(3)
u → y: P1(s) =
k1
τ1s + 1,
(4)
u → y: P2(s) =
k2
(τ21s + 1)(τ22s + 1).
(5)
fit (NRMSE) [%] =
1 −
N
i=1(HR(i) − HRsim(i))2
N
i=1(HR(i) − HR)2
× 100 %,
(6)
RMSE [bpm] =
1
N
N
i=1
(HRsim(i) − HR(i))2.
Page 9 of 10
Wang and Hunt BioMed Eng OnLine (2021) 20:37
Here, HRsim is the simulated HR response obtained using the estimated models and the
input signal, and HR is the measured HR from the validation data. ¯
HR is the mean value
of HR . i is the discrete time index and N is the number of discrete samples considered
(as described above, N = 360 ). Both of the above outcomes were calculated using the
“compare” function from the Matlab System Identification Toolbox.
Statistics
Statistical analysis was performed to test the hypothesis that the goodness-of-fit out-
comes of second-order models are better (higher fit and lower RMSE) compared to first-
order models. Prior to analysis, normality of differences between the goodness-of-fit
outcomes was formally assessed using the Matlab “lilliefors” function (this implements
a Kolmogorov–Smirnov test with correction according to the Lillifors method). As it
transpired that all differences were not significantly different from a normal distribution,
paired one-sided t tests were employed for hypothesis testing. Hypothesis testing used
a significance threshold of 5% ( α = 0.05 ). The Matlab Statistics and Machine Learning
Toolbox (The Mathworks, Inc., USA) was employed.
Abbreviations
bpm: Beats/min; CI: Confidence interval; fit/NRMSE: Normalised root-mean-square error; HR: Heart rate; HRV: Heart rate
variability; MD: Mean difference; PRBS: Pseudo-random binary sequence; RMSE: Root-mean-square error; SD: Standard
deviation; k: Steady-state gain; τ: Time constant.
Acknowledgements
Lars Brockmann (Institute for Rehabilitation and Performance Technology, Bern University of Applied Sciences) critically
reviewed the manuscript for important intellectual content.
Authors’ contributions
KH and HW designed the study. HW did the data acquisition. HW and KH contributed to the analysis and interpretation
of the data. HW wrote the manuscript; KH revised it critically for important intellectual content. Both authors read and
approved the final manuscript.
Funding
This study was funded by the Swiss National Science Foundation as part of the project “Heart Rate Variability, Dynamics
and Control During Exercise” (Ref. 320030-185351).
Availability of data and materials
The datasets used and/or analysed during the current study are available from the corresponding author on reasonable
request.
Declarations
Ethics approval and consent to participate
This research was performed in accordance with the Declaration of Helsinki. The study was reviewed and approved by
the Ethics Committee of the Swiss Canton of Bern (Ref. 2019-02184). All participants provided written, informed consent.
Consent for publication
Not applicable.
Competing interests
The authors declare that they have no competing interests.
Received: 13 January 2021 Accepted: 28 March 2021
References
1.
Hunt KJ, Fankhauser SE. Heart rate control during treadmill exercise using input-sensitivity shaping for disturbance
rejection of very-low-frequency heart rate variability. Biomed Signal Process Control. 2016;30:31–42.
2.
Garber CE, Blissmer B, Deschenes MR, Franklin BA, Lamonte MJ, Lee I-M, Nieman DC, Swain DP. American College of
Sports Medicine Position Stand. Quantity and quality of exercise for developing and maintaining cardiorespiratory,
Page 10 of 10
Wang and Hunt BioMed Eng OnLine (2021) 20:37
•
fast, convenient online submission
•
thorough peer review by experienced researchers in your field
•
rapid publication on acceptance
•
support for research data, including large and complex data types
•
gold Open Access which fosters wider collaboration and increased citations
maximum visibility for your research: over 100M website views per year
•
At BMC, research is always in progress.
Learn more biomedcentral.com/submissions
Ready to submit your research
Ready to submit your research ? Choose BMC and benefit from:
? Choose BMC and benefit from:
musculoskeletal, and neuromotor fitness in apparently healthy adults: guidance for prescribing exercise. Med Sci
Sports Exerc. 2011;43(7):1334–59.
3.
Riebe D, Ehrman JK, Liguori G, Magal M, editors. ACSM’s guidelines for exercise testing and prescription. 10th ed.
Philadelphia, USA: Wolters Kluwer; 2018.
4.
Mezzani A, Hamm LF, Jones AM, McBride PE, Moholdt T, Stone JA, Urhausen A, Williams MA. Aerobic exercise inten-
sity assessment and prescription in cardiac rehabilitation. Eur J Prev Cardiol. 2013;20(3):442–67.
5.
Weston M, Taylor KL, Batterham AM, Hopkins WG. Effects of low-volume high-intensity interval training (HIT) on
fitness in adults: a meta-analysis of controlled and non-controlled trials. Sports Med. 2014;44:1005–17.
6.
Ramos JS, Dalleck LC, Tjonna AE, Beetham KS, Coombes JS. The impact of high-intensity interval training versus
moderate-intensity continuous training on vascular function: a systematic review and meta-analysis. Sports Med.
2015;45:679–92.
7.
Hunt KJ, Fankhauser SE, Saengsuwan J. Identification of heart rate dynamics during moderate-to-vigorous treadmill
exercise. BioMed Eng Online. 2015;14:117.
8.
Wasserman K, Hansen JE, Sue DY, Stringer WW, Sietsema KE, Sun X-G, Whipp BJ. Principles of exercise testing and
interpretation. 5th ed. Philadelphia, USA: Lippincott, Williams and Wilkins; 2011.
9.
Whipp BJ, Ward SA, Lamarra N, Davis JA, Wasserman K. Parameters of ventilatory and gas exchange dynamics during
exercise. J Appl Physiol. 1982;52(6):1506–13.
10. Bearden SE, Moffat RJ. VO2 and heart rate kinetics in cycling: transitions from an elevated baseline. J Appl Physiol.
2001;90(6):2081–7.
11. Cheng TM, Savkin AV, Celler BG, Su SW, Wang L. Nonlinear modeling and control of human heart rate response dur-
ing exercise with various work load intensities. IEEE Trans Biomed Eng. 2008;55(11):2499–508.
12. Girard C, Ibeas A, Vilanova R, Esmaeili A. Robust discrete-time linear control of heart rate during treadmill exercise. In:
Proc. 24th Iran. Conf. Electr. Eng. (ICEE), 2016; pp. 1113–1118.
13. Sambeda Sarkar AS. Quasi sliding mode control: an application to heart rate regulation. In: Proc. Int. Conf. Control
Power Commun. Comput. Technol. (ICCPCCT), 2018; pp. 299–304.
14. Esmaeili A, Ibeas A, Herrera J, Balaguer P, Herrera J, Esmaeili N. Identification and robust control of heart rate during
treadmill exercise at large speed ranges. J Control Eng Appl Inform. 2019;21:51–60.
15. Du D, Hu Z, Du Y. Model identification and physical exercise control using nonlinear heart rate model and particle
filter. In: Proc. 15th Int. Conf. Autom. Sci. Eng. (CASE), 2019; pp. 405–410.
16. Baig D, Su H, Cheng TM, Savkin AV, Su SW, Celler BG. Modeling of human heart rate response during walking, cycling
and rowing. In: Proc. Ann. Int. Conf. IEEE Eng. Med. Biol. 2010; pp. 2553–2556.
17. Argha A, Ye L, Su SW, Nguyen H, Celler BG. Heart rate regulation during cycle-ergometer exercise using damped
parameter estimation method. In: Proc. 38th Ann. Int. Conf. IEEE Eng. Med. Biol. Soc. (EMBC), 2016; pp. 2676–2679.
18. Ljung L. System identification: theory for the user. 2nd ed. Upper Saddle River, New Jersey, USA: Prentice Hall; 1998.
19. Åström KJ, Wittenmark B. Computer controlled systems: theory and design. 3rd ed. Mineola, New York, USA: Dover
Publications; 2011.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
| Identification of heart rate dynamics during treadmill exercise: comparison of first- and second-order models. | 04-21-2021 | Wang, Hanjie,Hunt, Kenneth J | eng |
PMC7037403 | International Journal of
Environmental Research
and Public Health
Article
Seven Weeks of Jump Training with Superimposed
Whole-Body Electromyostimulation Does Not Affect
the Physiological and Cellular Parameters of
Endurance Performance in Amateur Soccer Players
Nicolas Wirtz 1,*
, André Filipovic 2, Sebastian Gehlert 2,3
, Markus de Marées 4,
Thorsten Schiffer 5, Wilhelm Bloch 2 and Lars Donath 1
1
Institute of Training Science and Sport Informatics, Department of Intervention Research in Exercise
Training, German Sport University Cologne, 50933 Cologne, Germany; [email protected]
2
Institute of Cardiology and Sports Medicine, Department of Molecular and Cellular Sports Medicine,
German Sport University Cologne, 50933 Cologne, Germany; andre.fi[email protected] (A.F.);
[email protected] (S.G.); [email protected] (W.B.)
3
Institute of Sport Science, University of Hildesheim, 50933 Hildesheim, Germany
4
Section of Sports Medicine and Sports Nutrition, Faculty of Sports Science, Ruhr University of Bochum,
44801 Bochum, Germany; [email protected]
5
Outpatient Clinic for Sports Traumatology and Public Health Consultation, German Sport University
Cologne, 50933 Cologne, Germany; t.schiff[email protected]
*
Correspondence: [email protected]; Tel.:+49-221-4982-6044
Received: 6 January 2020; Accepted: 1 February 2020; Published: 10 February 2020
Abstract: Intramuscular density of monocarboxylate-transporter (MCT) could affect the ability to
perform high amounts of fast and explosive actions during a soccer game. MCTs have been proven
to be essential for lactate shuttling and pH regulation during exercise and can undergo notable
adaptational changes depending on training. The aim of this study was to evaluate the occurrence and
direction of potential effects of a 7-weeks training period of jumps with superimposed whole-body
electromyostimulation on soccer relevant performance surrogates and MCT density in soccer players.
For this purpose, 30 amateur soccer players were randomly assigned to three groups. One group
performed dynamic whole-body strength training including 3 x 10 squat jumps with WB-EMS (EG,
n = 10) twice a week in addition to their daily soccer training routine. A jump training group (TG,
n = 10) performed the same training routine without EMS, whereas a control group (CG, n = 8)
merely performed their daily soccer routine. 2 (Time: pre vs. post) x 3 (group: EG, TG, CG) repeated
measures analyses of variance (rANOVA) revealed neither a significant time, group nor interaction
effect for VO2peak, Total Time to Exhaustion and Lamax as well as MCT-1 density. Due to a lack
of task-specificity of the underlying training stimuli, we conclude that seven weeks of WB-EMS
superimposed to jump exercise twice a week does not relevantly influence aerobic performance or
MCT density.
Keywords: electrostimulation; soccer; lactate; VO2peak; monocarboxylate transporter
1. Introduction
The physical demands of soccer players have increased notably within the last 10 to 20 years due
to modern game tactics and their variability. For example, the ability of a team to successfully play
high pressing mainly depends on the physical characteristics of the players. The distances covered
in the higher intensities and the number of quick and explosive actions such as accelerations, turns,
and jumps have increased within the recent years [1–3]. A player’s capacity to perform numerous of
Int. J. Environ. Res. Public Health 2020, 17, 1123; doi:10.3390/ijerph17031123
www.mdpi.com/journal/ijerph
Int. J. Environ. Res. Public Health 2020, 17, 1123
2 of 13
those actions with a highly intense load is considered crucial in modern soccer. This ability relies on
(1) adequate intra- and intermuscular coordination of soccer-specific movements and (2) metabolism
that ensures proper energy delivery [4]. Both has been shown to be affected by electromyostimulation
(EMS) training [5,6].
Jumps with superimposed Whole-Body EMS (WB-EMS) in addition to soccer training sessions can
be effective for improving accelerations, turns, jumps, and kicking velocity [5]. WB-EMS potentially
supports the athlete achieving higher power outputs and faster sport-specific movement velocities
using resistance training [7] by increased firing rates and synchronization of motor units, resulting
in a more pronounced activation of fast-twitch fibers at relatively low force levels [8]. Previous
studies showed that local EMS is beneficially affecting muscle metabolism and can elevate energy
expenditure and carbohydrate oxidation to a higher degree than voluntary contraction only [9–11].
Moreover, WB-EMS seem to stimulate anaerobic glycolysis for energy production with higher lactate
accumulation [12,13]. The beneficial effects of EMS on transportation of lactate have to be taken into
account as lactate shuttling via monocarboxylate transporters (MCTs) has been shown to improve
high-intensity intermittent exercise performance [14,15].
MCTs are considered essential for lactate shuttling and pH regulation during exercise and can
undergo notable adaptational changes depending on physical activity levels [16,17]. Due to a 1:1 ratio of
lactate and H+ being transported by MCTs, an increase in the two isoforms MCT-1 and MCT-4 in skeletal
muscle reduce the intracellular pH perturbations [18]. In line with this, studies revealed that the density
of MCT-1 and MCT-4 proteins in muscle is elevated after a macrocycle of endurance training [19–21].
However, some training studies did not find relevant increases in MCT-4 density [22–24]. It has been
assumed that MCT-1 production is more sensitive to physical stress than MCT-4. Since the biochemical
characteristics of MCT-1 favors lactate uptake, it has been suggested that erythrocytes provide a lactate
storage compartment in situations of physical exercise, thereby reducing the exercise-induced increase
in plasma lactate concentration [25].
Interestingly, Fransson et al. [26] showed remarkable changes in MCT-4 protein expression after
4 weeks of soccer specific training regimes like speed endurance (+30%) and small sided games (+61%)
in well-trained soccer players. An increase in MCT-1 and MCT-4 density in skeletal muscle after
6 weeks of strength training was however merely reported by Juel et al. [27]. No available study
investigated the effects of WB-EMS on relevant endurance capacities like VO2max and MCT-1 and
MCT-4 in soccer players. Against this background, the aim of our 3-armed randomized controlled trial
was to elucidate whether WB-EBS supplemented to a traditional soccer training routine can improve
endurance capacities indices and MCT density of soccer players. Our primary hypothesis was that a
training program of jumps with superimposed EMS may pronouncedly stimulate MCT-1 and MCT-4
density. Our secondary hypothesis was that endurance performance surrogates will not be affected by
the training program, because of the relatively low additional training volume and the subject´s high
overall training status.
2. Material and Methods
2.1. Participants
Only healthy field soccer players were included which means no cardiovascular or metabolic
diseases and no preinjury in the tested muscle groups. Participants needed to compete on a regional
level for the last 3 years and train 2–4 session per week with strength and conditioning training contents
and play one soccer match per week. In a randomized control trial twenty-eight soccer players from
10 different teams were assigned to three different groups. Control group was assigned based on
preferences and availability, whereas both intervention arms have been assigned based on coin toss.
The EMS group (EG, n = 10) performed jumps with superimposed WB-EMS twice a week accompanied
by 3 × 10 squat jumps in addition to the daily soccer routine over a period of 7 weeks that is a sufficient
intervention period with WB-EMS to improve strength abilities [5,28,29]. To differentiate between the
Int. J. Environ. Res. Public Health 2020, 17, 1123
3 of 13
effects caused by EMS and by the squat jumps and soccer training respectively, two control groups
were included. A jump training group (TG, n = 10) performed the same number of squat jumps
without EMS on the same days as the EG and a control group (CG, n = 8) that only performed the daily
soccer routine. All subjects were non-smokers. Basal anthropometric parameters of the subjects were
presented in Table 1.
This study was carried out in accordance with the recommendations of the “Ethics Committee of
the German Sports University Cologne”. All subjects gave written informed consent in accordance
with the Declaration of Helsinki. The protocol was approved by the “Ethics Committee of the German
Sports University Cologne” (06–02–2014).
Table 1. Anthropometric data (mean ± SD) and Total Training Load (arbitrary units) during the 7-weeks
intervention period calculated by Polar Team-2 Software according to training time spent in defined
heart rates zones.
Group
Age [Year]
Height
[m]
Weight
[kg]
Bodyfat
[%]
relVO2peak
[ml/kg*min-1]
Sessions/
Week
Total Training Load
[a.u.]
EG (n = 10)
24.4 ± 4.2
1.82 ± 0.03
81.4 ± 5.3
12.9 ± 2.1
52.1 ± 3.4
3.4 ± 1.2
3431 ± 911
TG (n = 10)
21.1 ± 1.9
1.83 ± 0.06
79.7 ± 5.5
10.8 ± 2.8
56.3 ± 5.7
3.4 ± 1.3
3479 ± 1723
CG (n = 8)
23.6 ± 3.9
1.82 ± 0.05
79.7 ± 7.5
14.1 ± 3.6
54.3 ± 7.2
2.6 ± 0.7
2644 ± 1437
2.2. Daily Soccer Routine
The participants performed 3.2 ± 1.0 soccer training sessions per week and competed once a week
in the championships. The standard training sessions lasted approximately 90 min including technical
skill activities, offensive and defensive tactics, athletic components with various intensities, small-sided
game plays and continuous play. In a normal training week during season with a match on Sunday
training was scheduled on Tuesdays, Wednesdays (optional), Thursdays and Fridays. Number of
training sessions and the training days varied according to the game schedule playing Sunday-Sunday
or Sunday-Saturday. The number of training sessions and the total training minutes were documented.
The training load was measured according to the training time spent in defined heart rate zones during
soccer training or match via Polar Team-2 Software (Polar Electro, Büttelborn, Germany) (see Table 1).
The training load [arbitrary units] provided by the Polar-Software aims to determine internal training
load based on background variables (sex, training history, metabolic thresholds, and maximal oxygen
consumption [VO2max]) and parameters measured during training sessions (exercise mode, and
energy expenditure) (c.f. [30]). The heart rate zones (100–90%, 89–80%, 79–70%, 69–60%, 59–50%)
were defined according to the individual maximum heart rate measured in the maximal ramp test
(see endurance test).
The players were asked to maintain their usual food intake und hydration according to the
recommendations for soccer players [31] and no nutrition supplementation was used. Additional
strength training was not allowed during the study.
All players had a constant training volume during the first half of the season (July till December)
and were in a well-trained condition with a relative VO2peak of 54.2 ± 5.9 mL/kg·min−1. All players
regularly conducted strength training during first half of the season and had overall experience in
strength training of 5.4 ± 3.9 years. The intervention period started after the three weeks mid-season
break from end of December till mid of January. During these three weeks the training load was
relatively low (moderate endurance training twice per week) in order to maintain fitness level and not
negatively affect Baseline testing.
2.3. WB-EMS Application and Protocol
In order to obtain a rest interval of 48 h between the two sessions and the championship game
on Sunday WB-EMS training was conducted on Tuesdays and Friday. All subjects abstained from
alcohol consumption for 24 h prior to and during the training intervention. The EMS Training was
conducted with a WB-EMS-system by Miha Bodytec (Augsburg, Germany). WB-EMS was applied
Int. J. Environ. Res. Public Health 2020, 17, 1123
4 of 13
with an electrode vest to the upper body with integrated bilaterally two paired surface electrodes for
the chest (15 × 5 cm), upper and lower back (14 × 11 cm), latissimus (14 × 9 cm), and the abdominals
(23 × 10 cm) and with a belt system to the lower body including the muscles of the glutes (13 × 10
cm), thighs (44 × 4 cm) and calves (27 × 4 cm). Biphasic rectangular wave pulsed currents (80 Hz)
were used with an impulse width of 350 µs [5]. The stimulation intensity (mA) was determined and
set separately for each muscle group (0–120 mA) by using a Borg Rating of Perceived Exertion [32].
The training intensity was defined for each player in a familiarization session two weeks before and
set at a sub-maximal level that still assures a clean dynamic jump movement (RPE 16–19 “hard to
very hard”) and was saved on a personalized chip card. The EG performed 3 × 10 maximal squat
jumps with a set pause of 60 s (no currents) per session. Every impulse for a single jump lasted for
4 s (range of motion: 2 s eccentric from standing position to an knee angle of 90◦–1 s isometric–0.1 s
explosive concentric–1 s landing and stabilisation) followed by a rest period of 10 s (duty cycle approx.
28%). This results in an overall time of 8.5 min per session an effective stimulation time of 2 min
per session. The players started with a 2–3 min standardized warm-up with movement preparations
including squats, skipping and jumps in different variations (squat jumps, jumps out of skipping or
double jumps) at a light to moderate stimulation intensity. The players were told to slowly increase
the intensity every few impulses. The training started when the players reached the defined training
intensity that was saved on the chip card from the last session according to the RPE 16–19 (“hard to
very hard”). The stimulation intensity was constantly increased individually every week (Tuesdays)
controlled by the coaches in order to maintain a high stimulation intensity. The intensity was increased
after the warm-up during the first and the second set of 10 squat jumps starting from calves up to the
chest electrodes. The TG conducted the same standardized warm-up and performed the same amount
of jumps with identical interval and conduction twice per week without EMS. The CG only performed
the 2–4 soccer training session plus one match per week.
2.4. Experimental Protocol
2.4.1. Endurance Test and Assessment of Anthropometrics
For determination of the endurance parameters spirometry was performed on a WOODWAY
treadmill (Woodway GmbH, Weil am Rhein, Germany) one week before (Baseline) and after the
7-weeks intervention period (Post-test) (Figure 1). Furthermore, bodymass and body composition
were determined via bioelectrical impendence analysis (TANITA corp., Tokyo, Japan). Endurance
tests were conducted three days after the soccer match to assure adequate recovery and not negatively
influence performance. Respiratory gases were analyzed via the ZAN600-System and ZAN-Software
GPI 3.xx (ZAN Austria e.U., Steyr-Dietach, Austria), using standard algorithms with dynamic account
for the time delay between the gas consumption and volume signal. To calibrate the device according
to the manufacturer´s guidelines, a gas mixture consisting of 5% CO2, 16% O2, and rest nitrogen was
used (Praxair Deutschland GmbH, Düsseldorf, Germany). To measure the maximum oxygen uptake
(VO2peak), the subjects performed an incremental ramp test [33]. Thereby, the players performed a
warmup at moderate speed (3 m·s−1) with 1% incline for 3 min. In the last 30 s the incline was increased
to 2.5%. Subsequently, running speed was then increased every 30 sec by 0.3 m/s until subjective
exhaustion was reported. Heart rate was documented in the last 10s of a ramp stage. The VO2peak
was determined as average maximum oxygen uptake of the first 20 s after ending the test. Additionally,
maximum heart rate, time to exertion (TTE) and maximum lactate concentration (Lamax) was recorded.
27 players completed the two endurance diagnostics. One player of the TG was removed from the
study due to an ankle joint injury prior to post testing.
Int. J. Environ. Res. Public Health 2020, 17, 1123
5 of 13
Figure 1. Timeline of endurance testing and muscle biopsy withdrawal during the study in the 2nd
half of the season.
2.4.2. Muscle Biopsies and Tissue Treatment.
Muscle biopsies via Bergström method [34] were taken from each player two weeks before
(Baseline) and in the week after the last training intervention (Post-test). All biopsies were obtained
under local anaesthetic from the middle portion of the vastus lateralis between the lateral part of the
patella and spina iliaca anterior superior 2.5 cm below the fascia. The muscle samples were freed from
blood and non-muscular material and embedded in tissue freezing medium (TISSUE TEK, Sakura,
Zoeterwoude, The Netherlands). Samples were frozen in liquid nitrogen-cooled isopentane and stored
at −80 ◦C for further analysis. The distance between the Baseline and Post-test incision was approx.
2.5 cm.
2.5. Immunohistochemistry
Muscle samples from 26 subjects were used for histology. 7 µm cross-sectional slices were obtained
from the frozen muscle tissue using a cryo-microtome Leica CM 3050 C (Leica Microsystems, Nußbach,
Germany) and placed on Polysine™ microscope slides (VWR International, Leuven, Belgium) [35].
Sections were fixed for 5 min in −20 ◦C pre-cooled acetone and air dried for 10 min at room temperature
(RT), before blocking for one hour at RT with TBS (tris buffered saline, 150 mM NaCl, 10 mM Tris-HCl,
pH 7.6) containing 5% BSA (bovine serum albumin). After blocking, sections were incubated overnight
(4 ◦C) with primary antibody for MCT-1 (ab3538P; 1:500; Merck Millipore, Burlington, MA, USA) and
MCT-4 (sc-376140; 1:400; Santa Cruz Biotechnology, Dallas, TX, USA), diluted in 0.8% BSA. To confirm
antibody specificity, control sections were incubated in TBS containing 0.8% BSA but without primary
antibody. After incubation, sections were washed 5 times short and twice for 10 min with TBS and
incubated for one hour with biotinylated goat anti-rabbit secondary antibody for MCT-1 (VECTOR
Laboratories, Burlingame, CA, USA), diluted 1:500 in TBS and goat anti-mouse for MCT-4 (VECTOR
Laboratories), diluted 1:400 in TBS, at RT.
After that, sections were washed again 5 times for 30 s and twice for 10 min before incubation
with fluorescent Alexa 488 secondary antibodies (Life Technologies, Carlsbad, CA, USA); diluted 1:500
in TBS for an hour. Afterwards sections were blocked with 5% BSA (TBS-Tween) for 30 min. Slides
were then incubated overnight at 4 ◦C with A4.951primary antibodies (A4951; type-I myosin heavy
chain; Developmental Studies Hybridoma Bank, Iowa City, IA, USA) diluted 1:200 in 0.8% BSA.
On the third day, sections were washed 5 times short and twice for 10 min with TBS before
incubated again with secondary antibodies and fluorescent Alexa 555 (red) diluted 1:500 in TBS for an
hour at RT. After washing again the samples, fixed on microscope glass slides, were embedded with
aqualpolymount and stored at RT.
Int. J. Environ. Res. Public Health 2020, 17, 1123
6 of 13
2.6. Data Analysis
The analysis of immunofluorescence stained myofibers were conducted with a confocal laser
scanning microscope (LSM 510, Zeiss, Jena, Germany) at 63X fold magnification. For the analysis of
MCT density in sarcoplasm and myofiber membranes, two laser channels 543 nm (for Alexa 555) and
488 nm (for Alexa 488) were used.
Two separate line-scans were conducted per measurement of membrane and sarcoplasmic areas
of single myofibers to determine the staining intensity for MCT 1 and MCT 4 and the means was used
for analysis (Figure 2). 1000 pixels were standardized analyzed per line scan along the membrane
and the sarcoplasm. MCT density was then calculated as the mean staining intensity of all pixels
along each line scan. For the analysis of type I fibers, only the green channel (Alexa 488) was used for
analysis and the red channel (Alex 555) was used for fiber type determination. Laser intensity was
standardized for each subject without changing throughout the analysis.
Figure 2. Representative pictures of immunofluorescence stained myofiber cross-sections showing
specific MCT-1 and MCT-4 staining (green) and type 1 myofiber staining (red) within membrane and
sarcoplasmic areas of myofibers (10× fold magnification). (A) MCT-4 Posttest, (B) MCT-1 Posttest.
2.7. Statistical Analysis
To determine the effect of the training interventions on endurance parameters, MCT-1 and MCT-4,
separate 2 (time: pre vs. post) × 3 (group: EG, TG, CG) mixed ANOVA with repeated measures
were conducted. ANOVA assumption of homogenous variances was tested using Maulchy-test of
Sphericity. A Greenhouse-Geisser correction was used when a violation of Mauchly´s test was observed.
To estimate overall time and interaction effect sizes, partial eta squared (η2p) was computed with
η2p ≥ 0.01 indicating small, ≥0.059 medium and ≥0.138 large effects [36]. If 2 × 3 mixed ANOVA
revealed a time*group interaction effect on any variable, this effect was further investigated using
Bonferroni post hoc tests for pairwise comparison. For all inferential statistical analyses, significance
was defined as a p-value less than 0.05. All descriptive and inferential statistical analyses were
conducted using SPSS 25® (IBM®, Armonk, NY, USA). Results were presented as means and standard
deviations (SDs). Figures were created with Prism 6 (GraphPad Software Inc., La Jolla, CA, USA).
Int. J. Environ. Res. Public Health 2020, 17, 1123
7 of 13
3. Results
3.1. Training Load
No significant differences were observed between the groups in the total number of training
sessions (EG 23.9 ± 7.8; TG 25.9 ± 6.6; CG 18.1 ± 5.6 sessions), training minutes (EG 2103 ± 630; TG 1812
± 919; CG 1437 ± 381Min), and the total recorded training load via Polar Team-2 software (Table 1).
All subjects of TG and EG had a compliance of 100% (14 training sessions) for jump training and
WB-EMS sessions, respectively.
3.2. Endurance Parameters
2 × 3 (time × group) ANOVA of repeated measures revealed no significant time, group or
interaction effect for relative VO2peak, TTE and Lamax. No group differences were observed at Baseline
or Posttest in none of the analyzed parameters (Figure 3).
Figure 3. (A) Relative maximum oxygen uptake (relVO2peak), (B) maximal lactate concentration,
(C) maximal running time till exertion (TTE), and (D) maximal heart rate) determined at the endurance
ramp-test on the treadmill in EMS-Group (EG), Training-Group (TG) and Control-Group (CG) measured
before (Baseline) and after the 7 weeks intervention period (Posttest). Values are presented in means
± SD.
3.3. MCT-4
3.3.1. Type-I Fibers
The 2 × 3 (time × group) repeated measures ANOVA revealed no significant time (p = 0.119,
η2p = 0.102), group or intervention effect (p = 0.165, η2p = 0.145) for the MCT-4 density in the membrane
of type-I muscle fibers. Regarding cytoplasm density of the MCT-4, a large significant effect over time
(p = 0.009, η2p = 0.26) was shown. No group*time effect (p = 0.318, η2p = 0.095) was however observed.
Subsequent post-hoc analysis showed a significant decrease in MCT-4 density after 7 weeks for TG
only (p = 0.005). No group differences were detected at Baseline and Posttest for MCT-4 density in
membrane and cytoplasm in type-I fibers (Figure 4).
Int. J. Environ. Res. Public Health 2020, 17, 1123
8 of 13
3.3.2. Type-II Fibers
With respect to the membrane density of the MCT-4, no time effect (p = 0.172, η2p = 0.079)
or group*time interaction (p = 0.315, η2p = 0.096) of type-II fibers was shown. For the cytoplasm
density of the MCT-4 a large significant main effect for the factor time (p = 0.001, η2p = 0.382) was
observed in type-II fibers. No group*time interaction effect (p = 0.333, η2p = 0.091) was observed.
Subsequent post-hoc analysis showed a significant decrease in MCT-4 distribution only for TG
(p = 0.004). No differences were shown between the groups at Baseline or Posttest for MCT-4 density
in the membrane and cytoplasm of the type-II fibers (Figure 4).
Figure 4.
MCT-4 density in type-I fiber (A) membrane and (B) cytoplasm, and in type-II fiber
(C) membrane and (D) cytoplasm for EMS-Group (EG), Training-Group (TG) and Control-Group (CG)
measured before (Baseline) and after the 7 weeks intervention period (Posttest). Values are presented
in means ± SD.
3.4. MCT-1
3.4.1. Type-I Fibers
The 2 × 3 (time × group) repeated measure ANOVA showed no significant effect over time
(p = 0.230, η2p = 0.065) as well as no significant group*time interaction effect (p = 0.045, η2p = 0.246) for
MCT-1 density in the membrane. Post-hoc analysis showed a significant decrease in density for the TG
(p = 0.032). For the cytoplasm density of the MCT-1 no main effects over time (p = 0.114, η2p = 0.110) or
group*time (p = 0.416, η2p = 0.077) were found. No group differences were detected at Baseline and
Posttest for MCT-1 density in the membrane and cytoplasm of the type-I fibers (Figure 5).
3.4.2. Type-II Fibers
The 2 × 3 ANOVA revealed a large significant time effect for cytoplasm MCT-1 (p = 0.009,
η2p = 0.269) but no group × time interaction effect (p = 0.933, η2p = 0.006). However, no significant
alternations in the cytoplasm were found in the three different groups over time. For membrane density
of the MCT-1 neither a time (p = 0.104, η2p = 0.115) nor an interaction effect (p = 0.480, η2p = 0.065)
was observed in type-II fibers. Group comparison revealed no differences between the three groups
at baseline and post-testing for MCT-1 density in the membrane and cytoplasm of the type-II fibers
(Figure 5).
Int. J. Environ. Res. Public Health 2020, 17, 1123
9 of 13
Figure 5. MCT-1 density in type-I fiber (A) membrane and (B) cytoplasm, and in type-II fiber (C)
membrane and (D) cytoplasm for EMS-Group (EG), Training-Group (TG) and Control-Group (CG)
measured before (Baseline) and after the 7 weeks intervention period (Posttest). Values are presented
in means ± SD.
4. Discussion
The main finding of this intervention is that 7 weeks of a dynamic WB-EMS program (2 sessions per
week) in addition to the regular soccer training does not relevantly influence endurance performance
indices as well as MCT-1 or MCT-4 density in the muscle. We surprisingly observed that MCT-4 density
in the cytoplasm and MCT-1 density in the membrane of type I muscle fibers notably decreased in TG,
the group that completed jumps without EMS. Participants of all three groups (EG, TG, CG) did their
weekly soccer sessions since years and training volume and training intensity was not changed during
the intervention period. Thus, the results could be explained by the high overall training status of the
subjects. Due to the documented effects of strength training on runner´s performance [37] and effects
of EMS application on runner´s VO2max [6], we analysed effects on some endurance parameter for EG.
Indeed, WB-EMS intervention with dynamic exercises and training status of the subjects (VO2max:
53 mL/min/kg) were similar to the study of Amaro-Gahete et al. [6]. However, the differential results
might be attributed to differences in current frequency (12–90 vs. 85 Hz), higher time under tension
(6 vs. 2 min) and higher intensities of exercises with superimposed EMS (strength and interval exercise
vs. jumps) in the cited study. With regard to the training status of the subjects and the general high
metabolic demand in soccer games and -training, the EMS stimulus could have been too low for further
adaptations in endurance capacities. Consequently, no changes were observed in any parameter
obtained during incremental treadmill running test. With respect to the results of Amaro-Gahete and
coworkers and in order to improve endurance parameter, it might be promising to adjust exercise to
higher training intensities, e.g., by shortening rest intervals of jumps or include other exercises within
high intensity intervals. The authors provided recommendations for an undulating modulation of
current adjustments [38], but without physiological explanation or reasoning. There are no studies
available that support these results and we found only one study that applied EMS during endurance
training. In this regard, Mathes and coworkers showed that, although metabolic stimuli and markers
of muscle damage were higher in cycling with superimposed EMS compared to cycling without EMS,
improvements of endurance performance and capacity were not significantly different between both
training methods [39].
Int. J. Environ. Res. Public Health 2020, 17, 1123
10 of 13
The disposed EMS-protocol concurrently to soccer training enhanced strength and myofiber
adaptations [40]. Furthermore it revealed to be effective for accelerations, direction changes, vertical
jumping ability and kicking velocity in elite soccer players [5]. Training design was identical within the
present study. Improving such surrogate parameters of aerobic or anaerobic endurance capacities seems
also promising to improve indices of soccer performance. The ability to perform sprints with high
intensity bouts is influenced by anaerobic capacity. The ability to do that repeatedly critically depends
on the aerobic metabolism. Both metabolic pathways are inter-linked with each other. However,
for the recommendation of WB-EMS, it would be also important that no degradation occurs since
soccer players need concurrent abilities of strength, speed, and endurance in the sense of repeated
high-intensity actions.
MCT-1 and MCT-4 content in the muscle was not influenced by the intervention of EG. Interventions
that showed increases of MCT-1 and MCT-4 conducted higher intensities and metabolic demanding
exercises. It is known that high-intensity endurance training increases MCT-1 in trained subjects [41,42]
and strength training increases MCT-1 and MCT-4 in untrained subjects [27]. The training program of
3 × 10 maximum jumps and 10 s between each jump was seemingly less intense. Indeed, jumps are
metabolically demanding, but 10 s of rest enable adequate delivery of oxygen. Unfortunately, lactate
accumulation was not measured during training in the present study. However, hundreds of repeated
jumps with 8 s rest between the jumps can result in moderate steady state lactate concentrations of
3–4 mmol·L−1 [43]. It might be a question of the stimulus´ intensity or accumulation that need to be
analyzed in exercise constellations that increase metabolic stimulus like high intensity interval training.
The superimposed WB-EMS on/off-time ratio should be increased accordingly.
Our results show a significant decrease in MCT-4 and MCT-1 content after jump training without
EMS (TG), which can be hardly explained, as EG and CG did not show significant differences in MCT-4
and MCT-1 content. Although the EG and TG showed equal training load (see Table 1) generally,
high-intensity anaerobic effort in soccer greatly varies according to the playtime and different playing
position requirements within a squad [44,45]. Furthermore, there can be differences of intensity in
daily soccer training routine that could lead to fluctuations of MCT´s. This speculation is indicated
by large standard deviations in total training load of TG (Table 1). Although subjects were assigned
to play and train as usual, it was not possible to adjust for this influence in the study. Replication
studies with accelerometer-based monitoring of the total loads are required to verify this issue.
Additionally, findings warrant further studies about strength training effects to MCT. In this regard,
authors have demonstrated the importance of detailed characterization of the training stimulus and
the subjects [46,47]. A specification of muscular time under tension and movement dynamics like
reactivity are missing in recent studies that dealt with EMS or MCT´s. The reduction effect of MCT
could also be attributed to a shift of MCTs to the membrane. A previous study has shown that MCT-1
localisation after training in diabetic patients increased in the sarcolemma of muscle fibers while the
sarcoplasmic content was reduced [48].
Some more limitations of the present study have to be mentioned for further research on the
effects of WB-EMS to endurance capacities. Due to the small sample size the study has pilot character.
Since we did not measure lactate production during training, we are not able to characterize the
metabolic stimulus of the training. Moreover, the effects of running performance mainly include
improved running economy, time trial performance and sprint performance all of which were not tested
in the current work. A more sports-specific testing set such as the Yo-Yo Intermittent Recovery test in
combination to sprint tests would have been useful in terms of ecological validity. A Further aspect of
limitation is that including players from different teams can result in differences in training sessions
for the CG (Table 1). A more detailed documentation of players match and training loads would be
helpful to avoid bias. Future research may consider changing study design to evoke higher metabolic
stimuli by increasing time under tension, reducing rest intervals or increasing intervention duration.
However, the present training stimulus was designed to improve soccer specific high-intensity actions
and could be integrated into daily training on a professional level [5].
Int. J. Environ. Res. Public Health 2020, 17, 1123
11 of 13
5. Conclusions
We conclude despite findings that the disposed WB-EMS protocol can enhance strength [5]
and myofiber adaptations [40] it is not a potent stimulation to improve VO2max and lactate
transport proteins.
Author Contributions: Conceptualization: A.F. and N.W.; Methodology: A.F., T.S. and M.d.M.; Software: A.F.;
Validation: S.G., N.W. and A.F.; Formal Analysis: A.F.; Investigation: A.F., T.S. and M.d.M.; Resources: W.B.; Data
Curation: A.F.; Writing—Original Draft Preparation: N.W.; Writing—Review and Editing: L.D., S.G. and N.W.;
Visualization: A.F.; Supervision: L.D.; Project Administration: A.F.; Funding Acquisition: A.F. All authors have
read and agreed to the published version of the manuscript.
Funding: This research was funded by the German Federal Institute of Sport Science, grant number AZ
070101/16-17.
Acknowledgments: The authors would like to thank Bianca Collins, Benedikt Seeger, Anke Schmitz, and Anika
Voß for excellent technical assistance.
Conflicts of Interest: The authors declare no conflict of interest.
References
1.
Barnes, C.; Archer, D.T.; Hogg, B.; Bush, M.; Bradley, P.S. The evolution of physical and technical performance
parameters in the English Premier League. Int. J. Sports Med. 2014, 35, 1095–1100. [CrossRef]
2.
Bush, M.; Barnes, C.; Archer, D.T.; Hogg, B.; Bradley, P.S. Evolution of match performance parameters
for various playing positions in the English Premier League. Hum. Mov. Sci. 2015, 39, 1–11. [CrossRef]
[PubMed]
3.
Wallace, J.L.; Norton, K.I. Evolution of World Cup soccer final games 1966-2010: Game structure, speed and
play patterns. J. Sci. Med. Sport 2014, 17, 223–228. [CrossRef] [PubMed]
4.
Glaister, M. Multiple sprint work: Physiological responses, mechanisms of fatigue and the influence of
aerobic fitness. Sports Med. 2005, 35, 757–777. [CrossRef] [PubMed]
5.
Filipovic, A.; Grau, M.; Kleinoder, H.; Zimmer, P.; Hollmann, W.; Bloch, W. Effects of a Whole-Body
Electrostimulation Program on Strength, Sprinting, Jumping, and Kicking Capacity in Elite Soccer Players.
J. Sports Sci Med. 2016, 15, 639–648.
6.
Amaro-Gahete, F.J.; De-la, O.A.; Sanchez-Delgado, G.; Robles-Gonzalez, L.; Jurado-Fasoli, L.; Ruiz, J.R.;
Gutierrez, A. Whole-Body Electromyostimulation Improves Performance-Related Parameters in Runners.
Front. Physiol. 2018, 9, 1576. [CrossRef]
7.
Young, W.B. Transfer of strength and power training to sports performance. IntJ. Sports Physiol Perform. 2006,
1, 74–83. [CrossRef]
8.
Gregory, C.M.; Bickel, C.S. Recruitment patterns in human skeletal muscle during electrical stimulation.
Phys. Ther. 2005, 85, 358–364. [CrossRef]
9.
Paillard, T.; Noe, F.; Passelergue, P.; Dupui, P. Electrical stimulation superimposed onto voluntary muscular
contraction. Sports Med. 2005, 35, 951–966. [CrossRef]
10.
Hamada, T.; Hayashi, T.; Kimura, T.; Nakao, K.; Moritani, T. Electrical stimulation of human lower extremities
enhances energy consumption, carbohydrate oxidation, and whole body glucose uptake. J. Appl. Physiol.
2004, 96, 911–916. [CrossRef]
11.
Wahl, P.; Schaerk, J.; Achtzehn, S.; Kleinoder, H.; Bloch, W.; Mester, J. Physiological responses and perceived
exertion during cycling with superimposed electromyostimulation. J. Strength Cond. Res. 2012, 26, 2383–2388.
[CrossRef] [PubMed]
12.
Spriet, L.L.; Soderlund, K.; Bergstrom, M.; Hultman, E. Skeletal muscle glycogenolysis, glycolysis, and pH
during electrical stimulation in men. J. Appl. Physiol. 1987, 62, 616–621. [CrossRef] [PubMed]
13.
Vanderthommen, M.; Duteil, S.; Wary, C.; Raynaud, J.S.; Leroy-Willig, A.; Crielaard, J.M.; Carlier, P.G.
A comparison of voluntary and electrically induced contractions by interleaved 1H- and 31P-NMRS in
humans. J. Appl. Physiol. 2003, 94, 1012–1024. [CrossRef]
14.
Gibala, M.J.; Little, J.P.; Macdonald, M.J.; Hawley, J.A. Physiological adaptations to low-volume, high-intensity
interval training in health and disease. J. Physiol. 2012, 590, 1077–1084. [CrossRef] [PubMed]
Int. J. Environ. Res. Public Health 2020, 17, 1123
12 of 13
15.
Iaia, F.M.; Bangsbo, J. Speed endurance training is a powerful stimulus for physiological adaptations and
performance improvements of athletes. Scand. J. Med. Sci. Sports 2010, 20 (Suppl. 2), 11–23. [CrossRef]
16.
Juel, C. Regulation of pH in human skeletal muscle: Adaptations to physical activity. Acta Physiol. 2008, 193,
17–24. [CrossRef]
17.
Thomas, C.; Bishop, D.J.; Lambert, K.; Mercier, J.; Brooks, G.A. Effects of acute and chronic exercise on
sarcolemmal MCT1 and MCT4 contents in human skeletal muscles: Current status. Am. J. Physiol. Regul.
Integr. Comp. Physiol. 2012, 302, R1–R14. [CrossRef]
18.
Juel, C.; Halestrap, A.P. Lactate transport in skeletal muscle—role and regulation of the monocarboxylate
transporter. J. Physiol. 1999, 517, 633–642. [CrossRef]
19.
Bonen, A.; McCullagh, K.J.; Putman, C.T.; Hultman, E.; Jones, N.L.; Heigenhauser, G.J. Short-term training
increases human muscle MCT1 and femoral venous lactate in relation to muscle lactate. Am. J. Physiol. 1998,
274, E102–E107. [CrossRef]
20.
Pilegaard, H.; Domino, K.; Noland, T.; Juel, C.; Hellsten, Y.; Halestrap, A.P.; Bangsbo, J. Effect of high-intensity
exercise training on lactate/H+ transport capacity in human skeletal muscle. Am. J. Physiol. 1999, 276,
E255–E261. [CrossRef]
21.
Dubouchaud, H.; Butterfield, G.E.; Wolfel, E.E.; Bergman, B.C.; Brooks, G.A. Endurance training, expression,
and physiology of LDH, MCT1, and MCT4 in human skeletal muscle. Am. J. Physiol. Endocrinol. Metab. 2000,
278, E571–E579. [CrossRef] [PubMed]
22.
Evertsen, F.; Medbo, J.I.; Bonen, A. Effect of training intensity on muscle lactate transporters and lactate
threshold of cross-country skiers. Acta Physiol. Scand. 2001, 173, 195–205. [CrossRef] [PubMed]
23.
Bickham, D.C.; Bentley, D.J.; Le Rossignol, P.F.; Cameron-Smith, D. The effects of short-term sprint training
on MCT expression in moderately endurance-trained runners. Eur. J. Appl. Physiol. 2006, 96, 636–643.
[CrossRef] [PubMed]
24.
Perry, C.G.; Heigenhauser, G.J.; Bonen, A.; Spriet, L.L. High-intensity aerobic interval training increases
fat and carbohydrate metabolic capacities in human skeletal muscle. Appl. Physiol. Nutr. Metab. 2008, 33,
1112–1123. [CrossRef]
25.
Sara, F.; Hardy-Dessources, M.D.; Marlin, L.; Connes, P.; Hue, O. Lactate distribution in the blood
compartments of sickle cell trait carriers during incremental exercise and recovery. Int. J. Sports Med.
2006, 27, 436–443. [CrossRef]
26.
Fransson, D.; Nielsen, T.S.; Olsson, K.; Christensson, T.; Bradley, P.S.; Fatouros, I.G.; Krustrup, P.;
Nordsborg, N.B.; Mohr, M. Skeletal muscle and performance adaptations to high-intensity training in
elite male soccer players: Speed endurance runs versus small-sided game training. Eur. J. Appl. Physiol.
2018, 118, 111–121. [CrossRef]
27.
Juel, C.; Holten, M.K.; Dela, F. Effects of strength training on muscle lactate release and MCT1 and MCT4
content in healthy and type 2 diabetic humans. J. Physiol. 2004, 556, 297–304. [CrossRef]
28.
Wirtz, N.; Zinner, C.; Doermann, U.; Kleinoeder, H.; Mester, J. Effects of Loaded Squat Exercise with and
without Application of Superimposed EMS on Physical Performance. J. Sports Sci. Med. 2016, 15, 26–33.
29.
Dörmann, U.; Wirtz, N.; Micke, F.; Morat, M.; Kleinöder, H.; Donath, L. The Effects of Superimposed
Whole-Body Electromyostimulation During Short-Term Strength Training on Physical Fitness in Physically
Active Females: A Randomized Controlled Trial. Front. Physiol. 2019, 10, 72. [CrossRef]
30.
Schumann, M.; Botella, J.; Karavirta, L.; Hakkinen, K. Training-Load-Guided vs Standardized Endurance
Training in Recreational Runners. IntJ. Sports Physiol Perform. 2017, 12, 295–303. [CrossRef]
31.
Garcia-Roves, P.M.; Garcia-Zapico, P.; Patterson, A.M.; Iglesias-Gutierrez, E. Nutrient intake and food habits
of soccer players: Analyzing the correlates of eating practice. Nutrients 2014, 6, 2697–2717. [CrossRef]
[PubMed]
32.
Tiggemann, C.L.; Korzenowski, A.L.; Brentano, M.A.; Tartaruga, M.P.; Alberton, C.L.; Kruel, L.F.M. Perceived
Exertion in Different Strength Exercise Loads in Sedentary, Active, and Trained Adults. J. Strength Cond. Res.
2010, 24, 2032–2041. [CrossRef] [PubMed]
33.
Sperlich, P.F.; Holmberg, H.C.; Reed, J.L.; Zinner, C.; Mester, J.; Sperlich, B. Individual versus Standardized
Running Protocols in the Determination of VO2max. J. Sports Sci. Med. 2015, 14, 386–393. [PubMed]
34.
Bergstrom, J. Percutaneous needle biopsy of skeletal muscle in physiological and clinical research. Scand. J.
Clin. Lab. Investig. 1975, 35, 609–616. [CrossRef]
Int. J. Environ. Res. Public Health 2020, 17, 1123
13 of 13
35.
Jacko, D.; Bersiner, K.; Hebchen, J.; de Marees, M.; Bloch, W.; Gehlert, S. Phosphorylation of alphaB-crystallin
and its cytoskeleton association differs in skeletal myofiber types depending on resistance exercise intensity
and volume. J. Appl. Physiol. 2019, 126, 1607–1618. [CrossRef] [PubMed]
36.
Cohen, J. Statistical Power Analysis for the Behavioral Sciences; Lawrence Erlbaum Associates: Hillsdale NJ,
USA, 1988.
37.
Berryman, N.; Mujika, I.; Arvisais, D.; Roubeix, M.; Binet, C.; Bosquet, L. Strength Training for Middle- and
Long-Distance Performance: A Meta-Analysis. Int. J. Sports Physiol. Perform. 2018, 13, 57–63. [CrossRef]
[PubMed]
38.
Amaro-Gahete, F.J.; De-la, O.A.; Sanchez-Delgado, G.; Robles-Gonzalez, L.; Jurado-Fasoli, L.; Ruiz, J.R.;
Gutierrez, A. Functional Exercise Training and Undulating Periodization Enhances the Effect of Whole-Body
Electromyostimulation Training on Running Performance. Front. Physiol. 2018, 9, 720. [CrossRef]
39.
Mathes, S.; Lehnen, N.; Link, T.; Bloch, W.; Mester, J.; Wahl, P. Chronic effects of superimposed
electromyostimulation during cycling on aerobic and anaerobic capacity. Eur. J. Appl. Physiol. 2017,
117, 881–892. [CrossRef]
40.
Filipovic, A.; DeMarees, M.; Grau, M.; Hollinger, A.; Seeger, B.; Schiffer, T.; Bloch, W.; Gehlert, S. Superimposed
Whole-Body Electrostimulation Augments Strength Adaptations and Type II Myofiber Growth in Soccer
Players During a Competitive Season. Front. Physiol. 2019, 10, 1187. [CrossRef]
41.
Burgomaster, K.A.; Cermak, N.M.; Phillips, S.M.; Benton, C.R.; Bonen, A.; Gibala, M.J. Divergent response of
metabolite transport proteins in human skeletal muscle after sprint interval training and detraining. Am. J.
Physiol Regul Integr Comp. Physiol 2007, 292, R1970–R1976. [CrossRef]
42.
McGinley, C.; Bishop, D.J. Influence of training intensity on adaptations in acid/base transport proteins,
muscle buffer capacity, and repeated-sprint ability in active men. J. Appl. Physiol. 2016, 121, 1290–1305.
[CrossRef] [PubMed]
43.
Wahl, P.; Sanno, M.; Ellenberg, K.; Frick, H.; Bohm, E.; Haiduck, B.; Goldmann, J.P.; Achtzehn, S.;
Bruggemann, G.P.; Mester, J.; et al. Aqua Cycling Does Not Affect Recovery of Performance, Damage
Markers, and Sensation of Pain. J. Strength Cond. Res. 2017, 31, 162–170. [CrossRef] [PubMed]
44.
Suarez-Arrones, L.; Torreno, N.; Requena, B.; Saez De Villarreal, E.; Casamichana, D.; Barbero-Alvarez, J.C.;
Munguia-Izquierdo, D. Match-play activity profile in professional soccer players during official games and
the relationship between external and internal load. J. Sports Med. Phys. Fit. 2015, 55, 1417–1422.
45.
Di Salvo, V.; Gregson, W.; Atkinson, G.; Tordoff, P.; Drust, B. Analysis of high intensity activity in Premier
League soccer. Int. J. Sports Med. 2009, 30, 205–212. [CrossRef]
46.
Toigo, M.; Boutellier, U. New fundamental resistance exercise determinants of molecular and cellular muscle
adaptations. Eur. J. Appl. Physiol. 2006, 97, 643–663. [CrossRef]
47.
Mujika, I. Quantification of Training and Competition Loads in Endurance Sports: Methods and Applications.
IntJ. Sports Physiol Perform. 2017, 12, S2–S9. [CrossRef] [PubMed]
48.
Opitz, D.; Kreutz, T.; Lenzen, E.; Dillkofer, B.; Wahl, P.; Montiel-Garcia, G.; Graf, C.; Bloch, W.; Brixius, K.
Strength training alters MCT1-protein expression and exercise-induced translocation in erythrocytes of men
with non-insulin-dependent type-2 diabetes. Can. J. Physiol. Pharm. 2014, 92, 259–262. [CrossRef] [PubMed]
© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access
article distributed under the terms and conditions of the Creative Commons Attribution
(CC BY) license (http://creativecommons.org/licenses/by/4.0/).
| Seven Weeks of Jump Training with Superimposed Whole-Body Electromyostimulation Does Not Affect the Physiological and Cellular Parameters of Endurance Performance in Amateur Soccer Players. | 02-10-2020 | Wirtz, Nicolas,Filipovic, André,Gehlert, Sebastian,Marées, Markus de,Schiffer, Thorsten,Bloch, Wilhelm,Donath, Lars | eng |
PMC4473093 | Electronic Supplementary Material Appendix S1
Specification of the search strategy used in the Pubmed database:
(488 HITS, 23th of june, 2014)
(((("Running"[Mesh]) AND (((("Athletic Injuries"[Mesh]) OR running injur*) OR running-related
injur*) OR "1000 hours")) NOT ((((((((((((((((((((((((("Addresses"[Publication Type]) OR
"Bibliography"[Publication Type]) OR "Biography"[Publication Type]) OR "Case
Reports"[Publication Type]) OR "Clinical Conference"[Publication Type]) OR
"Comment"[Publication Type]) OR "Congresses"[Publication Type]) OR "Dictionary"[Publication
Type]) OR "Directory"[Publication Type]) OR "Editorial"[Publication Type]) OR
"Festschrift"[Publication Type]) OR "Government Publications"[Publication Type]) OR
"Interview"[Publication Type]) OR "Lectures"[Publication Type]) OR "Legal Cases"[Publication
Type]) OR "Legislation"[Publication Type]) OR "Letter"[Publication Type]) OR
"News"[Publication Type]) OR "Newspaper Article"[Publication Type]) OR "Retracted
Publication"[Publication Type]) OR "Retraction of Publication"[Publication Type]) OR
"Review"[Publication Type]) OR "Scientific Integrity Review"[Publication Type]) OR "Technical
Report"[Publication Type]) OR "Validation Studies"[Publication Type])) NOT "Soccer"[Mesh])
NOT "Football"[Mesh] Filters: Danish; English
| Incidence of Running-Related Injuries Per 1000 h of running in Different Types of Runners: A Systematic Review and Meta-Analysis. | [] | Videbæk, Solvej,Bueno, Andreas Moeballe,Nielsen, Rasmus Oestergaard,Rasmussen, Sten | eng |
PMC8838374 |
Citation: Martínez-Rodríguez, A.;
Miralles-Amorós, L.;
Vicente-Martínez, M.; Asencio-Mas,
N.; Yáñez-Sepúlveda, R.;
Martínez-Olcina, M. Ramadan
Nutritional Strategy: Professional
Soccer Player Case Study. Nutrients
2022, 14, 465. https://doi.org/
10.3390/nu14030465
Academic Editor: Lauri Byerley
Received: 17 December 2021
Accepted: 18 January 2022
Published: 21 January 2022
Publisher’s Note: MDPI stays neutral
with regard to jurisdictional claims in
published maps and institutional affil-
iations.
Copyright:
© 2022 by the authors.
Licensee MDPI, Basel, Switzerland.
This article is an open access article
distributed
under
the
terms
and
conditions of the Creative Commons
Attribution (CC BY) license (https://
creativecommons.org/licenses/by/
4.0/).
nutrients
Article
Ramadan Nutritional Strategy: Professional Soccer Player
Case Study
Alejandro Martínez-Rodríguez 1,2,*
, Laura Miralles-Amorós 1
, Manuel Vicente-Martínez 3,
Nuria Asencio-Mas 1, Rodrigo Yáñez-Sepúlveda 4 and María Martínez-Olcina 1
1
Department of Analytical Chemistry, Nutrition and Food Science, Faculty of Sciences, University of Alicante,
03690 Alicante, Spain; [email protected] (L.M.-A.); [email protected] (N.A.-M.);
[email protected] (M.M.-O.)
2
Alicante Institute for Health and Biomedical Research (ISABIAL Foundation), 03010 Alicante, Spain
3
Faculty of Health Science, Miguel de Cervantes European University, 47012 Valladolid, Spain;
[email protected]
4
Escuela de Educación, Pedagogía en Educación Física, Universidad Viña del Mar, Viña del Mar 7055, Chile;
[email protected]
*
Correspondence: [email protected]
Abstract: The period of Ramadan induces changes in the usual eating patterns of individuals. During
this period, Muslims must abstain from drinking and eating from dawn to dusk. Therefore, some
research conducted on professional soccer players has observed that during and/or after Ramadan,
performance, running speed, agility, dribbling speed, and endurance and/or skill performance in
athletic events may be negatively affected by Ramadan intermittent fasting (RIF). The objective of this
study was to analyze the influence of a dietary plan during RIF on performance and body composition
in a professional soccer player. A 20-year-old elite player (86.0 kg, 188.5 cm) followed a dietary-
nutritional plan with an isocaloric diet and was supplemented with glycerol. The athlete’s strength
and power in the lower limbs was assessed by performing a countermovement jump (CMJ) and
Abalakov vertical jump (ABK) before and after Ramadan. After nutritional planning, the patient’s
body composition improved in terms of fat loss (6.61 to 5.70%) and muscle mass gain (50.26 to
51.50%). In addition, this translated into improvements in performance tests, both in the CMJ (36.72
to 40.00 cm) and ABK (39.16 to 49.34 cm). In conclusion, during a period of fasting, personalised
nutritional planning and an appropriate supplementation and rest protocol can improve the body
composition and performance of soccer players.
Keywords: nutrition; sport; athletic performance; sports supplements; body composition; dietary tools
1. Introduction
The period of Ramadan induces changes in the usual dietary patterns of individu-
als; during this period, Muslins must abstain from drinking and eating from sunrise to
sunset [1]. According to the Holy Quran, it takes place in the ninth month of the Islamic
calendar, and, as a lunar month, it lasts between 29 and 30 days [2].
According to the Federation International Football Association (FIFA), soccer has pro-
moted fair play, heterogeneity and fairness and is an internationally recognized sport [1].
The month of Ramadan often coincides with the calendars of many soccer leagues [1].
Researchers have proceeded to conduct studies on adult and youth Muslim soccer players
because of their commitment to their faith and the game. It has been shown that, dur-
ing and/or after Ramadan, performance, running speed, agility, dribbling speed, and
endurance and/or skill performance in athletic tests can be negatively affected by Ramadan
intermittent fasting (RIF) [3,4].
In the same way, sleep duration is compromised, as dinners and breakfasts are usually
early during Ramadan [5,6]. Indeed, by altering the sleep-wake cycle of circadian rhythms,
Nutrients 2022, 14, 465. https://doi.org/10.3390/nu14030465
https://www.mdpi.com/journal/nutrients
Nutrients 2022, 14, 465
2 of 8
a modification in anticipation and adaptation to environmental changes occurring during
the day may occur in the metabolism and cardiovascular system [2]. Therefore, biological
rhythms and how they affect human performance must be taken into account, as some
factors, such as hepatic and muscular glucogen stores, fluid stores, decreased blood glucose
levels, increased uric acid and the risk of dehydration during prolonged physical activity,
could negatively impair athletic performance [7].
Currently, there are no studies evaluating the effects of Ramadan that intervene at
the dietary-nutritional level, so there are still no adapted plans for practitioners practicing
RIF [1,8]. However, the variation in total energy intake before and after Ramadan has
been evaluated [6,9,10]. In these investigations, it has been observed that the intake was
insufficient to meet the needs of the athletes and, in addition, the estimated total daily
energy intake and the relative proportions of carbohydrate, fat and protein in the diet did
not change during Ramadan in the participants [6,9,10].
Physical performance tests such as counter movement jump (CMJ) and Abalakov
jump are often used in these investigations to assess recovery from baseline levels and
serve to measure power and fatigue [11,12]. Results for CMJ have been inconclusive in
the literature [8]. Rebaï et al. [13] observed significant increases in CMJ during Ramadan;
the increases were significant during versus before Ramadan for the group that reduced
their training load. However, some of the studies found do not evaluate the effect of RIF on
Abalakov’s jumping tests, with free arm jumps being predominant in this sport [9,10,13].
Another aspect that influences sports performance and can be affected by RIF is body
composition. In soccer, it is important to control body fat, as optimal fat levels allow players
to move more efficiently during training and matches [11,14]. Further relevant is the lean
mass compartment, specifically muscle mass, because insufficient or excessive training
loads can impact the physique, affecting performance, as well as speed, strength, power
and injury risk [15].
Although inconsistent findings on body composition for RIF have been observed in
the scientific literature [1,8], some studies use data obtained from bioimpedance or the
summation of 4 folds [6,9]. In addition, other research proposes, but does not carry out, the
complete profile of 5 components: skin, body fat, muscle mass, bone tissue and residual
mass. This model allows to get a photographic shape of the athlete’s profile [16].
Another aspect that lacks research so far is the medio somatotype and the Heath and
Carter anthropometric method classification [17] for Muslim RIF performers.
In order to respond to all these gaps in the scientific literature, the main objective of this
research was to describe the effects of RIF with personalized dietary-nutritional planning
on the athletic performance and body composition of an elite soccer player. The initial
hypothesis was that performance would be negatively affected by the month of Ramadan.
2. Materials and Methods
2.1. Study Design
A case report was used to analyse the influence of RIF on performance and body
composition in a professional soccer player. In this design, an initial measurement prior to
the intervention (start) and a measurement after the intervention (final) was conducted.
The participant was informed about the possible risks and discomforts that could arise
and was asked to complete a health history questionnaire and sign a consent form. The
research strictly followed the Declaration of Helsinki (Edinburgh review in 2000), and the
procedures were also in accordance with the recommendations of the EEC Good Clinical
Practice (document 111/3976/88 of July 1990). The subject signed an informed consent to
participate in the study after receiving all the information. All procedures were previously
approved by the Ethics Committee of the University of Alicante (UA-2021-03-11).
2.2. Participant
The participant was a 20-year old African soccer player since he was 5 years old,
having now 15 years of experience. He has been playing soccer at a professional level, for
Nutrients 2022, 14, 465
3 of 8
the team in La Liga for the last two seasons. He usually trains 12 h per week and his only
dedication is soccer. He is healthy, does not take any medication, has never undergone
surgery and has an optimal blood biochemistry.
2.3. Data Collection
2.3.1. Body Composition
Body weight was measured early in the morning, fasting and with minimal clothing,
using a Tanita BC-545n, (Tanita Corporation, Arlington Heights, IL, USA) to the near-
est 0.1 kg. Standing height without shoes was measured using a Seca 213 stadiometer
(Seca, Hamburg, Germany) to the nearest 0.1 cm. To minimize the potential source of
Bioimpedance systems (BIA) related to total body weight and height, body composition
assessment was performed by a level three anthropometrist following the International
Society for the Advancement of Kinanthropometry (ISAK) recommendations [18]. All mea-
surements were performed at the same location, at room temperature and under baseline
conditions. The technical error of measurement for perimeters, circumferences, lengths and
heights was less than 1% and less than 5% for skinfolds.
Anthropometric measurements were performed following the complete profile of ISAK
II methodology [18]. Skinfolds, girths, lengths and breadths were measured with a caliper,
flexible metallic tape, segmometer and pachymeter, respectively (Holtain, Crymych, UK).
Bone and muscle mass were obtained through Rocha’s equation [19] and Lee’s for-
mula [20], respectively. Fat mass was estimated using Carter’s formula [21], Faulkner
formula [22] and Withers’ equation [23]. Residual mass was calculated from the difference
between the total body weight minus the sum of the bone, muscle and fat masses. Accord-
ing to the Spanish Committee of Kinanthropometry, these methods are the most suitable
for high performance players [24].
Somatotype components were calculated from the assessment of different body com-
partments, including fat mass for endomorphy, muscle mass for mesomorphy, and leanness
and relative bone linearity for ectomorphy. The differences between each individual so-
matopoint with respect to the mean value was calculated using the somatotype attitudinal
mean (SAM) [25,26].
Proportionality analyses were performed using the Phantom stratum [19]. Each
variable was fitted to the Phantom size using the z-score. The z-values have a mean of 0, so
a z-value of 0.0 indicates that the given variable is proportionally equal to the Phantom; a z-
value greater than 0.0 means that the subject is proportionally greater than the Phantom for
that variable; and, conversely, a z-value less than 0.0 shows that the subject is proportionally
less than the Phantom for the variable [19].
2.3.2. Performance Measures
As indicators of global strength, the countermarch jump (CMJ) and the Abalakov
jump test [27] were performed. In the CMJ test, the participant performed a maximal
vertical jump starting from a standing position, without allowing the arms to swing, and
bending the knees 90◦. The participant performed several familiarisation trials prior to
the test [28]. A contact platform (Optojump Next Microgate, Bolzano, Italy) was used to
measure the CMJ. Three measurements were performed with 30 s recovery in between. The
flight time was used to calculate the jump height. The best jump was used for subsequent
analysis [29,30].
The Abalakov jump test followed the same protocol. The participant also performed
3 countermovement jumps with 30 s rest between jumps, on a platform with an optical
(infrared) data collection system (Optojump Next Microgate, Bolzano, Italy) to calculate
the Abalakov jump height [31]. The player had to stand up and perform a 90◦ knee flexion
followed by the fastest extension to reach the highest possible jump height. Of the three
results, the best one was used for statistical analysis.
Nutrients 2022, 14, 465
4 of 8
2.4. Nutritional Intervention Protocol
2.4.1. Nutrient Intake
For the dietary-nutritional intervention, quantitative estimates were made of total
energy expenditure based on basal metabolism, using the Harris–Benedict formula [32]
and corrected body weight. Physical activity expenditure was estimated from standardized
factors [33]. The proposed diet was based on an isocaloric intake following the recommen-
dations for elite soccer players [34].
Table 1 shows the values of macronutrients and vitamin B12 provided by the dietary-
nutritional guideline established according to the recommendations for elite soccer players.
Fiber intake was around 25 g per day and total cholesterol was less than 200 mg per day. The
software used in the elaboration of the diet was Dietopro (Dietopro, Valencia, Spain) [35].
Table 1. Dietary nutrient intake.
Calories (Kcal)
3432.1
Carbohydrates (g)
398.1
Proteins (g)
219.4
Fat (g)
112.7
Vitamin B12 (mg)
17.3
Carbohydrates (g/kg/day)
4.6
Protein (g/kg/day)
2.5
PFA/MFA
1.2
Kcal = kilocalories; g = grams; kg = kilograms; PFA = polyunsaturated fatty acids; MFA = monounsaturated
fatty acids.
Because the time of ingestion is different than usual during the RIF, a schedule was
established with SK for each of the different meals (Table 2).
Table 2. Meal times.
Intake
Time of Day
Breakfast
5:30–6:00
Snack
20:00
Dinner
22:30
Snack
0:30
2.4.2. Supplementation
Glycerol (1,2,3-propanetriol) is produced from glucose, protein, pyruvate, triacylglyc-
erols and other glycerolipid metabolic pathways and is a binding metabolite in numerous
pathways [27].
During RIF, glycerol is the only source of gluconeogenesis, as glycogen stores are
depleted within two days of fasting [28]. Considering its role as an energy substrate,
glycerol could effectively improve sports performance [28]. The combined ingestion of
glycerol and fluid has been used to increase body water volume, thus maintaining hydration
by reducing the rate of water elimination from the kidneys [27]. The intake protocol is
shown in Table 3.
Table 3. Glycerol intake protocol.
Time
Glycerol (g)
Liquid (L)
1–3 days
25
0.5
4–6 days
50
1
7–10 days
75
1.5
11–final
100
2
g = grams; L = liters.
Nutrients 2022, 14, 465
5 of 8
3. Results
3.1. Body Composition
The soccer player remained weight-stable throughout the study period (the initial
weight was 86.1 kg, while at the end of the intervention the weight reached 85.9 kg).
The athlete claimed to have followed the entire nutritional plan, based on 3400 kcal,
4.6 g/kg/day of carbohydrates and 2.5 g/kg/day of protein.
Table 4 shows the results of the body composition assessment between the beginning
and the end of the RIF. It can be observed that both weight and fat mass percentage
decreased after the intervention. However, muscle mass increased (from 50.26 to 51.50%).
The somatotype presented is ecto-mesomorphic, where mesomorphy predominates and
ectomorphy is higher than endomorphy.
Table 4. Body composition results.
Body Composition
Start
Final
Weight (kg)
86.1
85.9
Height (cm)
188.5
188.5
BFM Carter (%)
3.80
3.30
BFM Faulkner (%)
9.30
8.90
BFM Withers + Siri (%)
6.61
5.70
Muscle mass (kg) Lee 2000
43.27
44.20
Muscle mass (%) Lee 2000
50.26
51.50
Bone mass (kg) Rocha
13.96
14.01
Bone mass (%) Rocha
16.21
16.31
Residual mass (kg)
23.18
22.81
Residual mass (%)
26.92
26.55
Endomorphy
1.38
1.16
Mesomorphy
5.46
5.98
Ectomorphy
2.67
2.69
BFM: Body Fat Mass; kg = kilograms; cm = centimeters; % = percentage.
With regard to skinfolds, the triceps (4.5 to 3.5 mm), subscapular (8.2 to 7.6 mm),
biceps (3.4 to 2.5 mm), supraspinal (4.5 to 4.2 mm), abdominal (5.8 to 4.8 mm), thigh (6.3
to 6.1 mm) and leg (4.3 to 2.8 mm) skinfolds decreased, while the triceps (4.3 to 2.8 mm)
skinfold decreased, (5 to 4.2 mm), abdominal (5.8 to 4.8 mm), thigh (6.3 to 6.1 mm) and leg
(4.3 to 2.8 mm) decreased, while the iliac crest crease increased (6.1 to 6.4 mm).
Differences were observed in the different variables for anthropometric dimensions
and proportionality profile. The results for Z-perimeter calf (0.06 to 0.20), Z-perimeter thigh
(−0.08 to −0.25), Z-perimeter hip (−0.95 to −1.72), Z-perimeter waist (−1.17 to −1.20),
Z-perimeter forearm (1.96 to 2.28), Z-perimeter flexed arm (2.49 to 3.15) and Z-perimeter
relaxed arm (1.98 to 2.06) were higher after the intervention. However, the Z-perimeter
mesosternale decreased from −0.28 to −0.20.
3.2. Sport Performance
Regarding sport performance results, lower limb power measured with the CMJ and
Abalakov jumps tests shows a considerable improvement in performance after the RIF and
the dietary-nutritional intervention performed. The variation of the Abalakov jump was
from 39.16 to 49.34 cm, while the countermovement jump improved from 36.72 to 40 cm.
4. Discussion
The main objective of this research was to evaluate body composition, physical con-
dition (through different physical tests) and a personalized dietary-nutritional plan in a
professional soccer player at the beginning and at the end of the Ramadan period.
Differences in daily energy intake among different studies could be due to dietary
habits, cultural customs, biopsychosocial environment, and geographical differences of the
study participants. The present findings demonstrate that the estimated total daily energy
Nutrients 2022, 14, 465
6 of 8
intake was not affected by Ramadan fasting, since the player followed the dietary pattern,
where the nutritional plan was adapted to the fasting schedule [29–31].
In terms of body composition, the player presented lower initial characteristics of
muscle mass and higher fat mass compared to the end of the study. This can also be seen in
the measurement of skinfolds, where all decreased, except for the iliac crest. So far, no study
has analyzed these parameters in such depth. Only one study by Durnin and Womersley
et al. [36] analyzed the sum of 4 skinfolds (triceps, biceps, subscapular and suprailiac); it
was observed that there was no significant difference after the RIF period [37]. Another
investigation by Meckel et al. [9] found a slight increase in skinfolds. These studies did not
include the implementation of a personalized dietary plan, so the results may be due to an
inadequate diet.
It should be noted that this is the first study to investigate the effect of a dietary-
nutritional planning adapted to Ramadan on lower limb physical performance variables
and 5-component body composition. This study consists of the CMJ and Abalakov jump
tests, being the first study to evaluate Abalakov in a Muslim professional athlete. In
reference to the scores, all were higher after the intervention. Therefore, the player had
better lower body power at the end of the study than at the beginning. These results at
first indicate an increase in physical performance tests, however, they are different from
those obtained in previous studies, as some studies showed that Ramadan fasting decreases
physical performance [9,37], while others revealed no effect [3,4,6].
Furthermore, since Abalakov tests can be used as a performance indicator and as part
of the selection process of youth soccer players for national teams [38], future research
should include this test in its variables.
On the other hand, some researchers [27] underline the importance of the qualities of
glycerol for success in aerobic and anaerobic activities, emphasizing the effect on perfor-
mance, since it can be used as an energy substrate. The findings of such supplementation
evidence its relevance and usefulness in the athlete population, since it has been observed
that hypohydration decreases maximal aerobic power [27]. Therefore, the effects of such
supplementation could have had a positive impact on the performance tests of the subject
included in the present study, following the same line as other investigations [39].
Existing scientific evidence shows that adequate hydration and nutrition are essential
to improve performance in soccer during Ramadan periods [1,40]. It should be borne in
mind that soccer is a high-intensity intermittent sport played in hot environments, so there
is an increased risk of dehydration, with sweating rates in elite players of 1–2.5 L/h [39].
Therefore, both physiological function and cognitive and athletic performance are com-
promised. To maximise the effects of training during Ramadan and better adapt to the
conditions presented, it is essential to make appropriate dietary choices, ensuring optimal
energy intake during periods of high-intensity, long-duration training [1].
The results of this case study should be interpreted with caution, as they are based on
the performance of a single athlete using an experimental protocol (n = 1) in which neither
the participant nor the researchers were unaware of the project during the intervention
phase. In fact, only one set of performance tests was conducted at the beginning and end of
the intervention. Future research will consider the previously mentioned limitation. It is
recommended that researchers in the field provide more detailed information on fitness
assessment, body composition and dietary habits in future studies on professional soccer
players and the RIF, as research is still scarce.
5. Conclusions
During a fasting period such as Ramadan, an optimal and personalized nutritional
planning according to the demands and needs of each athlete, accompanied by a supple-
mentation of glycerol and rest protocol, allows for improvements on body composition and
lower limb performance indicators in soccer players.
Author Contributions: Conceptualization, A.M.-R. and M.M.-O.; methodology, A.M.-R. and R.Y.-S.;
software, M.V.-M.; validation, R.Y.-S. and A.M.-R.; formal analysis, M.M.-O.; investigation, M.M.-O.,
Nutrients 2022, 14, 465
7 of 8
L.M.-A., N.A.-M. and M.V.-M.; resources, M.V.-M.; data curation, L.M.-A. and N.A.-M.; writing—
original draft preparation, M.M.-O. and L.M.-A.; writing—review and editing, A.M.-R. and R.Y.-S.;
visualization, M.V.-M.; supervision, A.M.-R.; project administration, A.M.-R. All authors have read
and agreed to the published version of the manuscript.
Funding: This research received no external funding.
Institutional Review Board Statement: The study was conducted according to the guidelines of the
Declaration of Helsinki and also approved by the Ethics Committee of the University of Alicante
(UA-2021-03-11, 26 June 2021).
Informed Consent Statement: Written informed consent has been obtained from the patient(s) to
publish this paper.
Data Availability Statement: The data presented in this study is available on request from the
corresponding author. The data are not publicly available due to is personal health information.
Acknowledgments: To the European Institute of Exercise and Health (EIEH) of Alicante University
for their selfless collaboration in this research.
Conflicts of Interest: The authors declare no conflict of interest.
References
1.
DeLang, M.D.; Salamh, P.A.; Chtourou, H.; Ben Saad, H.; Chamari, K. The Effects of Ramadan Intermittent Fasting on Football
Players and Implications for Domestic Football Leagues Over the Next Decade: A Systematic Review. Sports Med. 2021, 1–16.
[CrossRef]
2.
Kalsbeek, A.; Scheer, F.A.; Perreau-Lenz, S.; La Fleur, S.E.; Yi, C.X.; Fliers, E.; Buijs, R.M. Circadian disruption and SCN control of
energy metabolism. FEBS Lett. 2011, 585, 1412–1426. [CrossRef]
3.
Chtourou, H.; Hammouda, O.; Souissi, H.; Chamari, K.; Chaouachi, A.; Souissi, N. The effect of Ramadan fasting on physical
performances, mood state and perceived exertion in young footballers. Asian J. Sports Med. 2011, 2, 177–185. [CrossRef]
4.
Zerguini, Y.; Kirkendall, D.; Junge, A.; Dvorak, J. Impact of Ramadan on physical performance in professional soccer players. Br.
J. Sports Med. 2007, 41, 398–400. [CrossRef]
5.
Herrera, C.P. Total sleep time in Muslim football players is reduced during Ramadan: A pilot study on the standardized
assessment of subjective sleep-wake patterns in athletes. J. Sports Sci. 2012, 30. [CrossRef]
6.
Bouzid, M.A.; Abaïdia, A.E.; Bouchiba, M.; Ghattassi, K.; Daab, W.; Engel, F.A.; Chtourou, H. Effects of Ramadan Fasting on
Recovery Following a Simulated Soccer Match in Professional Soccer Players: A Pilot Study. Front. Physiol. 2019, 10, 1480.
[CrossRef] [PubMed]
7.
Drust, B.; Ahmed, Q.; Roky, R. Circadian variation and soccer performance: Implications for training and match-play during
Ramadan. J. Sports Sci. 2012, 30, S43–S52. [CrossRef] [PubMed]
8.
Chtourou, H.; Trabelsi, K.; Boukhris, O.; Ammar, A.; Shephard, R.J.; Bragazzi, N.L. Effects of ramadan fasting on physical
performances in soccer players: A systematic review. Tunis. Med. 2019, 97, 1114–1131.
9.
Meckel, Y.; Ismaeel, A.; Eliakim, A. The effect of the Ramadan fast on physical performance and dietary habits in adolescent
soccer players. Eur. J. Appl. Physiol. 2008, 102, 651–657. [CrossRef] [PubMed]
10.
Aloui, A.; Chaouachi, A.; Chtourou, H.; Wong, D.P.; Haddad, M.; Chamari, K.; Souissi, N. Effects of Ramadan on the diurnal
variations of repeated-sprint performance. Int. J. Sports Physiol. Perform. 2013, 8, 254–262. [CrossRef]
11.
Hammami, M.; Hermassi, S.; Gaamouri, N.; Aloui, G.; Comfort, P.; Shephard, R.J.; Chelly, M.S. Field Tests of Performance
and Their Relationship to Age and Anthropometric Parameters in Adolescent Handball Players. Front. Physiol. 2019, 10, 1124.
[CrossRef]
12.
Saavedra, J.M.; Kristjánsdóttir, H.; Einarsson, I.; Guðmundsdóttir, M.L.; Þorgeirsson, S.; Stefansson, A. Anthropometric char-
acteristics, physical fitness, and throwing velocity in elite women’s handball teams. J. Strength Cond. Res. 2018, 32, 2294–2301.
[CrossRef] [PubMed]
13.
Rebaï, H.; Chtourou, H.; Zarrouk, N.; Harzallah, A.; Kanoun, I.; Dogui, M.; Souissi, N.; Tabka, Z. Reducing resistance training
volume during ramadan improves muscle strength and power in football players. Int. J. Sports Med. 2014, 35, 432–437. [CrossRef]
14.
La Monica, M.B.; Fukuda, D.H.; Miramonti, A.A.; Beyer, K.S.; Hoffman, M.W.; Boone, C.H.; Tanigawa, S.; Wang, R.; Church, D.D.;
Stout, J.R.; et al. Physical differences between forwards and backs in American collegiate rugby players. J. Strength Cond. Res.
2016, 30, 2382–2391. [CrossRef]
15.
Bernal-Orozco, M.F.; Posada-Falomir, M.; Quiñónez-Gastélum, C.M.; Plascencia-Aguilera, L.P.; Arana-Nuño, J.R.; Badillo-
Camacho, N.; Márquez-Sandoval, F.; Holway, F.E.; Vizmanos-Lamotte, B. Anthropometric and Body Composition Profile of
Young Professional Soccer Players. J. Strength Cond. Res. 2020, 34, 1911–1923. [CrossRef] [PubMed]
Nutrients 2022, 14, 465
8 of 8
16.
Ross, W.D.; Kerr, D.A. Fraccionamiento de la Masa Corporal: Un Nuevo Método para Utilizar en Nutrición, Clínica y Medicina
Deportiva—G-SE/Editorial Board/Dpto. Contenido. PubliCE. 1993. Available online: https://g-se.com/fraccionamiento-de-la-
masa-corporal-un-nuevo-metodo-para-utilizar-en-nutricion-clinica-y-medicina-deportiva-261-sa-Q57cfb27120415 (accessed on
8 December 2021).
17.
Heath, B.H.; Carter, J.E.L. A modified somatotype method. Am. J. Phys. Anthropol. 1967, 27, 57–74. [CrossRef] [PubMed]
18.
Esparza-Ros, F.; Vaquero-Cristóbal, R.; Marfell-Jones, M. Protocolo internacional para la valoración antropométrica. Perf. Complet.
Murcia Int. Soc. Adv. Kinanthropometry-ISAK 2019.
19.
Rocha, M.S.L. Peso ósseo do brasileiro de ambos os sexos de 17 a 25 años. Arq. Anatomía Antropol. 1975, 1, 445–451.
20.
Lee, R.C.; Wang, Z.; Heo, M.; Ross, R.; Janssen, I.; Heymsfield, S.B. Total-body skeletal muscle mass: Development and
cross-validation of anthropometric prediction models. Am. J. Clin. Nutr. 2000, 72, 796–803. [CrossRef]
21.
Carter, J.E.L. Body composition of montreal olympic athletes. In Physical Structure of Olympic Athletes; Karger Publishers: San
Diego, CA, USA, 1982; Volume 16, pp. 107–116.
22.
Faulkner, J.A. Physiology of swimming. Res. Quarterly. Am. Assoc. Health Phys. Educ. Recreat. 1966, 37, 41–54. [CrossRef]
23.
Withers, R.T.; Craig, N.P.; Bourdon, P.C.; Norton, K.I. Relative body fat and anthropometric prediction of body density of male
athletes. Eur. J. Appl. Physiol. Occup. Physiol. 1987, 56, 191–200. [CrossRef]
24.
Ramón, J.; Cruz, A.; Dolores, M.; Porta, J. Protocolo de valoración de la composición corporal para el reconocimiento médico-
deportivo. Documento de consenso del grupo español de cineantropometría (grec) de la federación española de medicina del
deporte (femede). Versión 2010. Arch. Med. Deporte 2009, XXVI, 166–179.
25.
Carter, J.E.L. The Heath-Carter Anthropometric Somatotype-Instruction Manual-Somatotype Instruction Manual 2 Part 1: The
Heath-Carter Anthropometric Somatotype-Instruction Manual. Ph.D. Thesis, San Diego State University, San Diego, CA, USA,
2002.
26.
Malina, R.M. Anthropometric Assessment of Nutritional Status; Wiley-Liss: New York, NY, USA, 1991; pp. 51–171.
27.
Patlar, S.; Yalçin, H.; Boyali, E. The effect of glycerol supplements on aerobic and anaerobic performance of athletes and sedentary
subjects. J. Hum. Kinet. 2012, 34, 69–79. [CrossRef]
28.
Baba, H.; Zhang, X.J.; Wolfe, R.R. Glycerol gluconeogenesis in fasting humans. Nutrition 1995, 11, 149–153. [PubMed]
29.
Alsubheen, S.A.; Ismail, M.; Baker, A.; Blair, J.; Adebayo, A.; Kelly, L.; Chandurkar, V.; Cheema, S.; Joanisse, D.R.; Basset, F.A.
The effects of diurnal Ramadan fasting on energy expenditure and substrate oxidation in healthy men. Br. J. Nutr. 2017, 118,
1023–1030. [CrossRef]
30.
Beltaifa, L.; Bouguerra, R.; Ben Slama, C.; Jabrane, H.; El Khadhi, A.; Rayana, B.; Doghri, T. Food Intake and Anthropometrical
and Biological Parameters in Adult Tunisians during Fasting at Ramadan. Available online: https://apps.who.int/iris/handle/
10665/119205 (accessed on 8 December 2021).
31.
Bouhlel, E.; Salhi, Z.; Bouhlel, H.; Mdella, S.; Amamou, A.; Zaouali, M.; Mercier, J.; Bigard, X.; Tabka, Z.; Zbidi, A.; et al. Effect of
Ramadan fasting on fuel oxidation during exercise in trained male rugby players. Diabetes Metab. 2006, 32, 617–624. [CrossRef]
32.
Seagle, H.M.; Strain, G.W.; Makris, A.; Reeves, R.S. Position of the American Dietetic Association: Weight management. J. Am.
Diet. Assoc. 2009, 109, 330–346. [CrossRef]
33.
Ainsworth, B.E.; Haskell, W.L.; Whitt, M.C.; Irwin, M.L.; Swartz, A.M.; Strath, S.J.; O’Brien, W.L.; Bassett, D.R.J.; Schmitz, K.H.;
Emplaincourt, P.O.; et al. Compendium of physical activities: An update of activity codes and MET intensities. Med. Sci. Sports
Exerc. 2000, 32, S498–S504. [CrossRef]
34.
Collins, J.; Maughan, R.J.; Gleeson, M.; Bilsborough, J.; Jeukendrup, A.; Morton, J.P.; Phillips, S.M.; Armstrong, L.; Burke,
L.M.; Close, G.L.; et al. UEFA expert group statement on nutrition in elite football. Current evidence to inform practical
recommendations and guide future research. Br. J. Sports Med. 2021, 55, 416. [CrossRef]
35.
García, C.G.; Sebastià, N.; Blasco, E.; Soriano, J.M. Dietopro.com: A new tool for dietotherapeutical management based on cloud
computing technology. Nutr. Hosp. 2014, 30, 678–685. [CrossRef]
36.
Durnin, J.V.G.A.; Womersley, J. Body fat assessed from total body density and its estimation from skinfold thickness: Measure-
ments on 481 men and women aged from 16 to 72 Years. Br. J. Nutr. 1974, 32, 77–97. [CrossRef] [PubMed]
37.
Chennaoui, M.; Desgorces, F.; Drogou, C.; Boudjemaa, B.; Tomaszewski, A.; Depiesse, F.; Burnat, P.; Chalabi, H.; Gomez-Merino, D.
Effects of Ramadan fasting on physical performance and metabolic, hormonal, and inflammatory parameters in middle-distance
runners. Appl. Physiol. Nutr. Metab. 2009, 34, 587–594. [CrossRef]
38.
Ryman Augustsson, S.; Arvidsson, J.; Haglund, E. Jump height as performance indicator for the selection of youth football players
to national teams. J. Sports Med. Phys. Fitness 2019, 59, 1669–1675. [CrossRef] [PubMed]
39.
Siegler, J.C.; Mermier, C.M.; Amorim, F.T.; Lovell, R.J.; McNaughton, L.R.; Robergs, R.A. Hydration, Thermoregulation, and
Performance Effects of Two Sport Drinks during Soccer Training Sessions. J. Strength Cond. Res. 2008, 22, 1394–1401. [CrossRef]
[PubMed]
40.
Aziz, A.R.; Muhamad, A.M.C.; Roslan, S.R.; Mohamed, N.G.; Singh, R.; Chia, M.Y.H. Poorer intermittent sprints performance in
ramadan-fasted muslim footballers despite controlling for pre-exercise dietary intake, sleep and training load. Sports 2017, 5, 4.
[CrossRef] [PubMed]
| Ramadan Nutritional Strategy: Professional Soccer Player Case Study. | 01-21-2022 | Martínez-Rodríguez, Alejandro,Miralles-Amorós, Laura,Vicente-Martínez, Manuel,Asencio-Mas, Nuria,Yáñez-Sepúlveda, Rodrigo,Martínez-Olcina, María | eng |
PMC4687124 | RESEARCH ARTICLE
The Dynamics of Speed Selection and
Psycho-Physiological Load during a Mountain
Ultramarathon
Hugo A. Kerhervé1,2*, Guillaume Y. Millet3,4, Colin Solomon1
1 School of Health and Sport Sciences, University of the Sunshine Coast, Sippy Downs, Australia,
2 Laboratoire de Physiologie de l’Exercice, EA-4338, Université Savoie Mont Blanc, Le Bourget-du-Lac,
France, 3 Human Performance Laboratory, University of Calgary, Calgary, Canada, 4 Laboratoire de
Physiologie de l’Exercice, Université de Lyon, F–42023, Saint–Etienne, France
* [email protected]
Abstract
Background
Exercise intensity during ultramarathons (UM) is expected to be regulated as a result of the
development of psycho-physiological strain and in anticipation of perceived difficulties
(duration, topography). The aim of this study was to investigate the dynamics of speed,
heart rate and perceived exertion during a long trail UM in a mountainous setting.
Methods
Fifteen participants were recruited from competitors in a 106 km trail mountain UM with a
total elevation gain and loss of 5870 m. Speed and gradient, heart rate (HR) and ratings of
perceived exertion (dissociated between the general [RPEGEN] and knee extensor fatigue
[RPEKE] and collected using a voice recorder) were measured during the UM. Self-selected
speed at three gradients (level, negative, positive), HR, RPEGEN and RPEKE were deter-
mined for each 10% section of total event duration (TED).
Results
The participants completed the event in 18.3 ± 3.0 h, for a total calculated distance of 105.6 ±
1.8 km. Speed at all gradients decreased, and HR at all gradients significantly decreased
from 10% to 70%, 80% and 90%, but not 100% of TED. RPEGEN and RPEKE increased
throughout the event. Speed increased from 90% to 100% of TED at all gradients. Average
speed was significantly correlated with total time stopped (r = -.772; p = .001; 95% confidence
interval [CI] = -1.15, -0.39) and the magnitude of speed loss (r = .540; p = .038; 95% CI =
-1.04, -0.03), but not with the variability of speed (r = -.475; p = .073; 95% CI = -1.00, 0.05).
Conclusions
Participants in a mountain UM event combined positive pacing strategies (speed decreased
until 70–90% of TED), an increased speed in the last 10% of the event, a decrease in HR at
PLOS ONE | DOI:10.1371/journal.pone.0145482
December 21, 2015
1 / 13
OPEN ACCESS
Citation: Kerhervé HA, Millet GY, Solomon C (2015)
The Dynamics of Speed Selection and Psycho-
Physiological Load during a Mountain Ultramarathon.
PLoS ONE 10(12): e0145482. doi:10.1371/journal.
pone.0145482
Editor: Pedro Tauler, University of the Balearic
Islands, SPAIN
Received: October 12, 2015
Accepted: December 6, 2015
Published: December 21, 2015
Copyright: © 2015 Kerhervé et al. This is an open
access article distributed under the terms of the
Creative Commons Attribution License, which permits
unrestricted use, distribution, and reproduction in any
medium, provided the original author and source are
credited.
Data Availability Statement: All relevant data are
within the paper.
Funding: No specific funding was used for this study.
The first author was supported by university research
funds and scholarship (USC and Queensland
Education and Training).
Competing Interests: The authors have declared
that no competing interests exist.
70–90% of TED, and an increase in RPEGEN and RPEKE in the last 30% of the event. A
greater speed loss and less total time stopped were the factors associated with increased
total performance. These results could be explained by theoretical perspectives of a com-
plex regulatory system modulating motor drive in anticipation of perceived difficulties such
as elevation changes.
Introduction
Exercise intensity during self-paced events is regulated by a complex protective system inte-
grating instantaneous somatosensory feed-back and anticipatory mechanisms in order to
maintain homeostasis and prevent from catastrophic failure [1–3]. As in shorter duration, self-
paced exercises [2, 4–8], the dynamics of exercise intensity during ultramarathons (UM) are
expected to be regulated as a result of the development of psycho-physiological strain and in
anticipation of perceived difficulties (duration, topography) [1]. Despite being sensitive to
environmental factors such as gradient and wind [9], speed is commonly used as an indicator
of exercise intensity in the study of athletic performance. Speed is predicted to decrease
throughout UM events [10], and systematic descriptions of pacing during field UM events
have been used to indicate that speed decreases overall and becomes more variable as a func-
tion of increased finishing times [11–14].
UM events performed on trails and in mountainous settings include changes in elevation,
surface, obstacles, altitude, remoteness, and adverse atmospheric conditions, which could alter
the dynamics of pacing, compared to flat events. For instance, significant peripheral fatigue
(low-frequency fatigue, indicative of the failure of excitation-contraction coupling typical of
exercises involving intense eccentric and stretch-shortening contractions [15, 16]) was
observed following a ~37 h mountain UM event [17], but not following a 24 h level, treadmill
run [14]. Therefore, it is possible that the dynamics of speed during a mountain UM would dif-
fer compared to a level UM.
Some measures of psycho-physiological load, such as the ratings of perceived exertion
(RPE) and heart rate (HR), are strong predictors of pacing during self-paced exercise [1, 18],
and could assist in characterising the psycho-physiological load pertaining to a runner at vari-
ous stages of an UM. HR is an indicator of the cumulative systemic physiological response to
variations in physiological and psychological load, and is routinely used to monitor exercise
intensity [19]. The RPE are scores of specifically-developed subjective scales [20] widely used
as indicators of psycho-physiological strain [2], and have been proposed to be a primary vari-
able involved in the selection of work rate [1, 2]. During a 68km UM lasting 9.8 ± 0.4 hr, HR
decreased in the second half, and RPE increased throughout the event without reaching maxi-
mal values (15.4 ± 0.4) [21]. RPE was also found to increase and to have slightly higher values
(14.1 ± 2.0, with a maximum under ~18) throughout a 73 km mountain UM lasting 11.5 ± 0.5
hr [22]. More recently, near-maximal RPE values (19.5 ± 1.5) were reported at the end of a 54
km mountain UM despite a moderate exercise intensity as evidenced by the relatively low aver-
age speed (3.83 km h−1), and HR (111.7 ± 5.9 bpm) [23]. However, no description of psycho-
physiological load and pacing has been performed in long, mountain UM events.
Therefore, the aim of this study was to investigate the dynamics of speed, HR and perceived
exertion during a long trail UM in a mountainous setting. We hypothesised that a progressive
decrease in speed (positive pacing) and increased variability would be observed in participants.
Based on shorter duration events [11, 24], we also hypothesised that changes in speed would be
Psycho-Physiological Load during a Mountain Ultramarathon
PLOS ONE | DOI:10.1371/journal.pone.0145482
December 21, 2015
2 / 13
correlated to changes in gradient, and would exhibit different dynamics at level gradients com-
pared to positive and negative gradients. In association with these changes, we hypothesised
that HR would decrease (reverse HR drift), and that RPE would increase throughout the event.
Together, these findings would indicate that pacing is regulated not only as a consequence of
the development of fatigue (simultaneous decrease in speed and HR, [25]), but also in an antic-
ipatory manner to prevent from reaching high levels of exertion prematurely.
Methods
Ethics statement and participants
Ethical approval for the study was granted by the research ethics committee of the University
of the Sunshine Coast (project code S/12/432). The participants were recruited from experi-
enced runners registered to compete in a ~167 km UM running event with ~10,000 m of eleva-
tion gain and loss, and held in Chamonix, France (The North Face1 Ultra-Trail du Mont-
Blanc, UTMB). Nineteen male participants provided written informed consent and were ini-
tially included in the study, fifteen of which completed the entire event and constitute the study
group (age: 43 ± 10 yr, height: 1.78 ± 0.60 m, weight: 74 ± 8 kg).
Study procedures
The participants were individually familiarised with the following study procedures in the 3
days preceding the event. Due to adverse meteorological conditions, the course was shortened
to ~106 km and the elevation gain and loss reduced to ~6,000 m on race day.
Measures of distance, speed and gradient.
The participants were equipped with a Non-
differential Global Positioning System (GPSND) device (BT-Q1000, Qstarz International, Tai-
wan) secured on top of their clothing or gear to record the distance, speed and elevation for the
entire event. The GPSND data were retrieved using the proprietary software, and exported in
columnar format for data analysis. The variables contained in the files were date and time (uni-
versal time constant, UTC), position (latitude, longitude, elevation) and speed (via Doppler
shift).
From the successive geographical coordinates recorded, we calculated the distance between
each data point using the Vincenty great-circle formulae [26], which are spherical trigonome-
try functions calculating the shortest distance between spatial coordinates at the surface of an
ellipsoid (earth dimensions used were the WGS-84 GPS model of reference with equatorial
radius 6,378.137 km, polar radius 6,356.752 314 245 km and flattening f 1=298:257223563).
An automated calculation of the Vincenty formulae can be obtained from an internet-based
utility (GPS Visualizer; www.gpsvisualizer.com), which we compared to our preliminary mea-
sures for 10 data sets and found to be in exact agreement (r = 1.00, p < .001). Therefore, we
used the automated formulae as a simple and generalisable procedure to obtain point-to-point
distances. Point-to-point speed was subsequently calculated using the ratio of the point-to-
point distances and of the GPS epoch time (one data point every 5 s in the current study).
To reduce the effect of signal errors in the analysis, a two-step treatment procedure was
then applied to the data. Preliminary calculations revealed that GPSND devices did not discrim-
inate speeds slower than 1 km h−1 (0.28 m s−1 or 1.39 m in 5 s) based on the typical error in
speed measured in a static position (drift, when a device will record speed values due to the
non geo-synchronous nature of the constellation of satellites). For the high end of the speed
range, it was considered that speeds higher than 20 km h−1 (5.56 m s−1 or 27.8 m in 5 s) were
not expected during a long UM and were likely due to signal jamming (which occurs mainly
when the signal from a satellite becomes too weak and forces the ground based receiver to pair
to another satellite). These erroneous distance and speed data were first assigned a value of
Psycho-Physiological Load during a Mountain Ultramarathon
PLOS ONE | DOI:10.1371/journal.pone.0145482
December 21, 2015
3 / 13
zero, and all speed values were then smoothed in order to further increase the signal-to-noise
ratio. For smoothing, a 3, 9, and 15-pt weighted averages were graphically compared. The 9-pt
weighted average was considered satisfactory as it provided a balanced sensitivity to individual
observations of slow and high speeds. This two-step procedure facilitated the reduction of the
effect of signal drift and jamming, which both artificially increase the distance and speed mea-
sured using GPS devices, while remaining sensitive to periods of null speed values (Fig 1).
GPS-based elevation is considered to be inaccurate [27] due to differences between the
model of reference of the earth used for calculations and the actual shape of the earth, and
therefore an independent source was sought in order to increase the quality of the elevation
data. Due to the size of a typical file containing UM data at the relatively high recording rates
of GPS devices (12 h of data recording at 5 s equals 8640 observations), a digital elevation
model (DEM) was used in order to automate the treatment procedure. Elevation values were
reconstructed from the geographical positions using a DEM (in this study, the NASA SRTM3)
available from the same online utility (GPS Visualizer; gpsvisualizer.com). Data was smoothed
using a 9-pt weighted average. The gradient between two consecutive data points was then
Fig 1. Total time stopped. Total time stopped for each participant, including the relative position of official event checkpoints. “CP” are official event
checkpoints. CP 1 and 6: distance ~7 km (outbound) and ~69 km (inbound), altitude ~1015 m. CP 2: distance ~ 19 km, altitude ~815 m. CP 3 and 5: distance
~29 km (outbound) and ~52 km (inbound), elevation ~1160 m. CP 4: distance ~36.5 km, elevation ~1699 m. CP 7: distance ~93 km, elevation ~1263 m.
doi:10.1371/journal.pone.0145482.g001
Psycho-Physiological Load during a Mountain Ultramarathon
PLOS ONE | DOI:10.1371/journal.pone.0145482
December 21, 2015
4 / 13
calculated as the change in elevation divided by the horizontal distance between two points
(the amount of vertical gain as a function of horizontal distance).
Measures of psycho-physiological load.
HR was measured continuously using a chest
strap and watch (RS800, RS800cx, RS400, or S810, Polar Electro, Kempele, Finland). Each
watch was set to record one data point every 15 s in order to optimise battery life and memory.
RPE scores were recorded using a portable voice recorder (ICD PX312, Sony, Tokyo,
Japan). We instructed the participants to record the time of observation, a general (RPEGEN)
and a local (muscular) RPE focused on the sensation of fatigue or pain of the knee extensor
muscles and excluding any psychological/psychic contribution to exertion (RPEKE) using
Borg’s 10 point category-ratio scale (CR-10) that the subjects carried over the entire race.
Variables and statistical analyses
We reported all variables as a function of total event distance (Figs 1 and 2) or duration (Figs 3
and 4) in order to represent all participants on a comparable scale (where 100% represents the
distance or duration at event completion for every participant). In order to ensure sufficient
data was used at each stage, data for each dependent variable was computed for every 10%
Fig 2. UTMB course outline and elevation profile. (A) Mean (±SD) group elevation data from the Digital Elevation Model values associated with measures
of geographical positions, and (B) group speed data associated with measures of geographical positions (CP, refer to legend of Fig 1 for description).
doi:10.1371/journal.pone.0145482.g002
Psycho-Physiological Load during a Mountain Ultramarathon
PLOS ONE | DOI:10.1371/journal.pone.0145482
December 21, 2015
5 / 13
Fig 3. Dynamics of speed. Mean (±SD) group speed as a function of event duration in (A) level, (B) negative
(C) and positive gradients, respectively. Symbols denote significant differences to (*) 10%, ($) 20%, (#) 30%,
(θ) 40%, (&) 50%, (€) 60%, (ϕ) 70%, (Ω) 90% and (£) 100% of total event duration, at p < .05.
doi:10.1371/journal.pone.0145482.g003
Psycho-Physiological Load during a Mountain Ultramarathon
PLOS ONE | DOI:10.1371/journal.pone.0145482
December 21, 2015
6 / 13
section of the total duration of the event. All statistical analyses were performed using SPSS
(version 21, IBM Corporation, Armonk NY, USA). Data are reported as mean ± SD, and the
level of significance was set at p < .05.
Dynamics of exercise intensity.
We determined the relationship between the variations of
speed and changes in elevation for the entire event using a quadratic regression of individual
speed and gradient. After confirming the assumption of the equality of variances were met, the
effect of exercise duration on speed at each gradient (level, negative and positive inclines) HR,
RPEGEN and RPEKE were determined using a multivariate ANOVA (MANOVA). Post-hoc
one-way, repeated measures ANOVAs with a Fisher’s LSD post-hoc test were used in order to
locate the differences in means for speed at all gradients. Due to incomplete data sets, the
dynamics of HR, RPEGEN and RPEKE were assessed using a one-way ANOVA on ranks (Krus-
kal-Wallis test) with a Student-Newman-Keuls post-hoc test.
As HR does not adjust to exercise intensity instantaneously, it is not possible to treat HR
data in the same way as speed. Instead, we investigated the dynamics of exercise intensity using
Fig 4. Dynamics of psycho-physiological load. (A) Mean (±SD) group heart rate (HR) as a function of total event duration, and (B) general and muscular
(knee extensors) ratings of perceived exertion (RPE) as a function of total event duration. Symbols denote significant differences to (*) 10% and ($) 20%, at p
< .05. Bpm: beats per minute. CR-10: 10-point category-ratio Borg scale.
doi:10.1371/journal.pone.0145482.g004
Psycho-Physiological Load during a Mountain Ultramarathon
PLOS ONE | DOI:10.1371/journal.pone.0145482
December 21, 2015
7 / 13
sections of sustained uphill running, in order to maximise the contribution of metabolic work
compared to passive energy recovery (since the ability to perform eccentric contractions
decreases with the development of peripheral fatigue, refer to [15, 16]) in total work rate [28].
We identified 6 sections of sustained uphill combining at least 300 m of vertical gain at a 10%
average gradient (refer to Fig 2; the section from CP2 to CP3 was only 4.9% gradient and was
not included in the analysis). The effect of hill order (1 to 6) on HR was assessed using a
repeated-measures, one-way ANOVA and a Bonferroni post-hoc test.
However, during uphill running, overground speed is a less relevant metrics than the
amount of vertical gain (in m h−1) to characterise exercise intensity, which is also dependent
on the gradient of the slope [28]. Therefore, to determine whether any drift in HR existed inde-
pendent of exercise intensity, we used a 1-factor principal component analysis to reduce the
dimension of these three variables (termed SVG for speed, vertical gain, gradient). After con-
firming the assumptions of normality using a Shapiro-Wilk test, the relationship between aver-
age HR and SVG in the 6 main ascents was assessed using Pearson’s product-moment
correlation. We further tested the effect of hill order (1–6) on HR using a repeated-measures
one-way ANCOVA, and a Bonferroni post-hoc test, using SVG as a covariate of HR.
Factors of performance. The relationship between final performance (using the individual
average speed, as it allows comparison of various UM distances) and 1) the variability of speed
(using the coefficient of variation of point-to-point speed values), 2) the magnitude of speed
loss (using the slope of the linear regression of speed over the entire event) and 3) the total time
stopped (assumed to correspond to resting, eating, clothing and gear change, toilet, other) were
tested using Pearson’s product-moment correlation after confirming assumptions of normality
were met (Shapiro-Wilk test). The 95% confidence intervals (CI) of the correlations were calcu-
lated using the unstandardised beta-weights of the linear regression of the Z-scores of each
variable.
Results
The following data sets were retrieved: 15 complete GPS traces, 9 HR data sets with at least
80% of event data (due to equipment issues and loss of signal), and 6 RPE data sets with at least
80% of data. The average distance for the event, calculated using filtered point-to-point ortho-
drome, was 105.6 ± 1.8 km (range: 103.0–107.5 km), and the total elevation gain and loss was
5871 ± 239 m. The 15 participants completed the event in 18.3 ± 3.0 h (range: 13.8–23.9 h) at
an average speed of 5.88 ± 0.9 km h−1 (range: 4.58–7.58 km h−1) and an average HR of
132 ± 10 bpm (range: 112–146 bpm). The mean group elevation and speed profiles are repre-
sented in Fig 2. There was a significant quadratic correlation between point-to-point speed and
elevation changes (linear factors model: r = .49, R2 = .24, F-linear = 316.76, p < .001; quadratic
factors model: r = .52, R2 = .27, F-change = 40.99, p < .001; Total factors: F-total = 185.22, p <
.001).
Dynamics of exercise intensity
The changes in speed as a function of total event duration are presented in Fig 3 (panels A, B
and C for level, negative and positive gradients, respectively). Positive pacing was observed on
level (speed loss: -2.91 ± 2.15%), negative (-2.61 ± 0.92%) and positive gradients (-1.31 ±
0.84%). The MANOVA indicated a difference in speed between the speed at level gradient, and
at the negative and positive gradients (p < .001). Speed was not significantly different between
the negative and positive gradients (p = .10). Post-hoc ANOVAs indicated that speed decreased
from 10% to all sections up to 70% of total duration at level inclines (except 60%, where it sig-
nificantly increased compared to 40% and 50%), and that speed increased at 90% compared to
Psycho-Physiological Load during a Mountain Ultramarathon
PLOS ONE | DOI:10.1371/journal.pone.0145482
December 21, 2015
8 / 13
40%, 50%, 70%, 80%, and increased at 100% compared to all sections between 30% and 90%
(Fig 3A). For negative gradients, speed decreased from 10% to all other sections, and from 20%
and 30% to 70% and 90%. Speed then increased between 90% and 100% (Fig 3B). For positive
gradients, speed decreased at all sections until 80% of event duration except 60%. Speed
increased at 100% compared to all observations between 40% and 90% (Fig 3C).
Mean group HR averaged 132.6 ± 13.6 bpm, and decreased -34.2 ± 17.2 bpm over the UM.
HR decreased from 10% to 70%, 80% and 90%, but not 100% (Fig 4A). Mean group RPEGEN
and RPEKE averaged 4.3 ± 1.1 and 3.8 ± 1.3, respectively, and increased 6.9 ± 1.4 and 6.9 ± 2.2,
respectively during the event (Fig 4B). Mean group RPEGEN increased significantly from 10%
to 70%, 80%, 90% and 100%, and from 20% to 70%, 80% and 90% (Fig 4B). Mean group
RPEKE increased significantly from 10% to 70%, 80%, 90% and 100% (Fig 4B). Mean group
RPEGEN and RPEKE were significantly positively correlated (r = .980, p < .001).
There was no significant change in HR as a function of time in the 6 main climbs, as evi-
denced in the ANOVA (using HR alone) as well as in the ANCOVA (using HR and SVG as
covariates). HR was significantly positively correlated with SVG on all uphill sections (r = .663,
p < .001).
Factors of performance
Performance (average speed) was negatively correlated with total time stopped (r = -.772, p =
.001; 95% CI = -1.15, -0.39) and positively correlated with the magnitude of speed loss (r =
.540, p = .038; 95% CI = -1.04, -0.03) but not with the variability of speed (r = -.475, p = .073;
95% CI = -1.00, 0.05).
Discussion
There were three main findings in this study: (i) speed decreased overall at all inclines during
the event (positive pacing), but increased significantly in the last section at all inclines; (ii)
faster participants stopped less and decreased their speed more than slower participants
throughout the event; and (iii) the measures of psycho-physiological load indicated that despite
evidence of a reverse HR drift and increased RPE throughout the event, HR in sustained climbs
did not change, and maximal RPE values were relatively low, suggesting that participants
actively regulated (paced) their physiological and psychological load to complete the event and
avoid premature exhaustion.
During self-paced running exercise, the optimal locomotor speed is adjusted as a function
of environmental factors [10]. One of the main factors influencing speed selection is gradient,
where additional energy is required to run at high negative (to generate braking forces limiting
downward acceleration) and positive (to elevate a runner’s mass against gravity) gradients
compared to level or slightly negative gradients [28, 29]. The curvilinear relationship between
locomotor speed and gradient measured in this study (indicating that speed varies directly as a
function of gradient; Fig 2) had so far been assumed to exist [11] but not measured in UM
events due to limitations in the ability to measure speed and gradient of individual participants.
This relationship was measured despite the overall low speeds (especially at positive gradients)
and positive pacing strategies characteristic of mountain UM events, which could both have
affected the relationship.
The overall decrease in speed during the event indicates that the study participants used
positive pacing strategies (progressively slowing down), in agreement with previous research
findings specific to UM running [10–13, 25, 30]. Speed losses on level gradients were more pro-
nounced (the slope of the linear regression was greater, and the MANOVA indicated that sig-
nificant differences existed with both negative and positive gradients) and occurred at an
Psycho-Physiological Load during a Mountain Ultramarathon
PLOS ONE | DOI:10.1371/journal.pone.0145482
December 21, 2015
9 / 13
earlier point in the event (reaching a minimum at 70% of event duration) compared to both
negative and positive gradients (minimum at 90% of event duration). The increase in speed at
60% of total event duration on level gradients could not be explained using the data we col-
lected. This section corresponded to a section of the course in the main valley of Chamonix,
and therefore we hypothesise that the terrain and surface were conducive for running (in con-
trast to walking), and that the course was accessible by spectators and crew members (which
could have provided support and motivation). The existence of an increase in speed at the end
of the event at all inclines in the current study is unique compared to other studies of long UM
on level ground [13]. Two combined factors could, in part, explain the presence of an increase
in speed in the last section: first, the mainly descending profile at the end of the event (after hill
6, refer to Fig 2), and second, a phenomenon termed cardio-pulmonary [7] or speed reserve [1]
predicting the increase of exercise intensity at the end of self-paced exercises. In this mountain
UM event, the marked elevation gain and loss may have favoured the use of conservative pac-
ing strategies decreasing the risk of premature exhaustion in anticipation of difficulties, when
compared to level or hilly UM events [1]. Together, these findings are indicative of mixed pac-
ing strategies, which is a subset of the three main types of pacing associating positive pacing for
the main part of the event and an increase in speed for the final section of the event.
The significant relationship between performance and speed loss in this project contrasts
with findings in other UM studies [11, 13], where participants with a higher performance level
had greater speed losses. Future studies are required to further investigate this unexpected
result, which could potentially originate from faster participants pacing their race less conser-
vatively from the start aiming to decrease overall time, compared to slower participants, for
whom finishing the event could have been the main goal. Future studies should investigate a
priori pacing strategies, and performance goals and attitudes toward risk taking as a function
of performance level. The inverse correlation between performance and time stopped was
novel in UM running, and extends findings in an ultra-endurance cycling event [31] where
faster athletes spent less time napping. While this result is expected in shorter duration events,
it is commonly believed among participants of ultra-long duration events that a bout of passive
rest can be beneficial to final performance. This finding indicates the marked differences in the
physiological demands of an UM event as a function of performance level due to the differ-
ences in time spent on course, where passive rest may be a relatively important feature of pac-
ing strategies for slower participants. Future studies are also required to determine whether a
threshold exists as a function of performance level in longer (> 300 km) UM events.
We reported that HR decreased from 10% to 70%, 80% and 90%, but not 100% of total
event duration. Although we could not determine the relationship between the dynamics of
speed and HR at each gradient over the entire event (due to the relatively low accuracy of
GPSND devices), it is likely that the variations of exercise intensity (speed) explained most of
the observed changes in HR (reverse HR drift), as this well-described physiological response is
typical of ultra-long duration exercise [1, 25]. Still, we reported that HR in sustained uphill sec-
tions distributed throughout the event (hills 1–6) did not change, including with the use of HR
scaled for exercise intensity (using the factorial component SVG), which indicates that bouts of
sustained uphill running may be regulated differentially given the high risk of exhaustion they
present.
The dissociated RPE (RPEGEN and RPEKE) had similar dynamics (the two measures of RPE
increased from 10% to all sections between 70% and 100% of event duration), and were highly
correlated, suggesting either (i) that the dynamics and magnitude of change of RPEGEN and
RPEKE are similar throughout the course of an UM (which would indicate that the measure of
one variable is sufficient), or (ii) that the two scales measured the same underlying construct (and
therefore, that RPE may be a general, but not location-specific indicator of psycho-physiological
Psycho-Physiological Load during a Mountain Ultramarathon
PLOS ONE | DOI:10.1371/journal.pone.0145482
December 21, 2015
10 / 13
load). Future research is required to investigate the relative contribution of the perceived exertion
specific to a muscle group (knee extensors, plantar flexors) or physiological system (respiratory,
gastric) to the general RPE. The role of pain will also need to be investigated as it may alter the
perception of exertion and fatigue [1], and will be heightened following sustained downhill loco-
motion [32]. Further, although the highest values in RPE (both RPEGEN and RPEKE) were
recorded in the last section of the event, the maximal values were relatively low (6.3 and 6.7
group mean for RPEGEN and RPEKE, respectively). The relatively low maximum values distin-
guishes it from other studies in shorter UM [23], and could contribute to identify the protective
nature of fatigue in preventing the participants from attaining maximal values at the end of the
event [1, 3]. As such, the combined results of the dynamics of pacing and psycho-physiological
load may indicate that participants relied on relatively conservative pacing strategies and used a
functional reserve [1, 7] permitting an increase in speed observed in the last section of the event.
These findings are consistent with theoretical perspectives of a complex protective system regu-
lating work rate based on the interaction of the instantaneous and anticipated psycho-physiologi-
cal state of a participant, and of the environmental conditions in which the exercise is performed
[1–3]. Still, some questions remain regarding the regulation of speed at different gradients, since
some of our results (the pacing on level gradients differed to both downhill and uphill gradients)
were not expected. Previous research hypothesised that the changes in stride patterns were
altered differentially following a level [33] and a mountain UM [32] due to the increased reliance
on eccentric contractions typical of downhill running. Recently, Vernillo and colleagues [34]
measured an increase in the energy cost of running specifically at mild downhill (-5%) gradients
but not on level or uphill (5%) gradients as a function of the development of fatigue in an UM
event. Therefore, further studies are required to investigate the simultaneous variations of pacing
and psycho-physiological load as a function of gradient during an UM event with a greater
resolution.
Limitations
In this study, the main limitation was related to the resolution of measurements made using
non-differential GPS devices. We optimised the calculations to be able to define broad catego-
ries of total distance and duration (10% sections) and gradients (negative [-100 to -2.5%], level
[-2.5 to 2.5%], positive [2.5 to 100%]). Still, the temporal resolution of observations (10% of
event duration) was comparable to other UM studies [13], and was selected as a robust
approach ensuring a sufficient number of observations was available for each variable. Unfore-
seen issues with recording equipment reduced the numbers of data sets in HR, and therefore
limited the findings as we were unable to establish the relationship between HR and pacing at
each stage. In future studies, the use of GPS devices with higher temporal and spatial resolution
could also lead to the development of various indices of running performances in conditions of
trail running, such as the rate of ascent as a function of gradient which would be useful for ath-
letes and coaches, and in scientific research for the analysis of performance.
Conclusion
During a mountain UM, speed decreased over the first 90% of the event and at all gradients,
and speed increased in the last 10% section of the event. A greater speed loss and less total time
stopped were the factors associated with increased total performance. HR decreased overall,
but remained constant in the main ascents of the race, indicating the potential effect of conser-
vative pacing strategies to avoid premature exertion. Perceived exertion increased throughout
the event, but without reaching maximal values. These observations are supported by
Psycho-Physiological Load during a Mountain Ultramarathon
PLOS ONE | DOI:10.1371/journal.pone.0145482
December 21, 2015
11 / 13
theoretical perspectives of a complex protective system regulating motor drive in anticipation
of remaining exercise duration and changes in elevation.
Acknowledgments
The authors would like to thank all the participants for their effort and valuable time. We
would also like to thank the organisers of the UTMB and other researchers who have permitted
this research effort. In particular, the authors would like to thank Dr Roger Ouillon and Dr
Pascal Edouard for conducting medical screenings, Mr Sylvain Battault for providing the voice
recording equipment, and Dr Léonard Féasson for organising and managing the testing facility.
We also thank Ms Benjie Bartos and Mr Dylan Astley for proof-reading and editing the
manuscript.
Author Contributions
Conceived and designed the experiments: HK CS. Performed the experiments: HK GM. Ana-
lyzed the data: HK CS. Wrote the paper: HK CS GM.
References
1.
Millet GY. Can neuromuscular fatigue explain running strategies and performance in ultra-marathons?:
The flush model. Sports Medicine. 2011; 41(6):489–506. doi: 10.2165/11588760-000000000-00000
PMID: 21615190
2.
Tucker R, Noakes TD. The anticipatory regulation of performance: The physiological basis for pacing
strategies and the development of a perception-based model for exercise performance. British Journal
of Sport Medicine. 2009; 43(6):392–400. doi: 10.1136/bjsm.2008.050799
3.
Noakes TD. Fatigue is a brain-derived emotion that regulates the exercise behavior to ensure the pro-
tection of whole body homeostasis. Frontiers in Physiology. 2012; 3:1–13. doi: 10.3389/fphys.2012.
00082
4.
Ansley A, Schabort E, Gibson A, Lambert MI, Noakes TD. Regulation of pacing strategies during suc-
cessive 4-km time trials. Medicine and Science in Sports and Exercise. 2004; 36(10):1819–25. PMID:
15595306
5.
Ansley L, Robson PJ, St Clair-Gibson A, Noakes TD. Anticipatory pacing strategies during supramaxi-
mal exercise lasting longer than 30 s. Medicine and Science in Sports and Exercise. 2004; 36(2):309–
14. PMID: 14767256
6.
Noakes TD, Lambert MI, Hauman R. Which lap is the slowest? An analysis of 32 world mile record per-
formances. British Journal of Sport Medicine. 2009; 43(10):760–4. doi: 10.1136/bjsm.2008.046763
7.
Swart J, Lamberts RP, Lambert MI, St Clair-Gibson A, Lambert EV, Skowno J, et al. Exercising with
reserve: evidence that the central nervous system regulates prolonged exercise performance. British
Journal of Sport Medicine. 2009; 43(10):782–8. doi: 10.1136/bjsm.2008.055889
8.
Tucker R, Noakes TD. The physiological regulation of pacing strategy during exercise: A critical review.
British Journal of Sport Medicine. 2009; 43(6):e1. doi: 10.1136/bjsm.2009.057562
9.
Cavagna GA, Thys H, Zamboni A. The sources of external work in level walking and running. Journal of
Physiology (London). 1976; 262(3):639–57. doi: 10.1113/jphysiol.1976.sp011613
10.
Abbiss CR, Laursen PB. Describing and understanding pacing strategies during athletic competition.
Sports Medicine. 2008; 38(3):239–52. doi: 10.2165/00007256-200838030-00004 PMID: 18278984
11.
Angus SD, Waterhouse BJ. Pacing strategy from high-frequency field data: more evidence for neural
regulation? Medicine and Science in Sports and Exercise. 2011; 43(12):2405–12. PMID: 21606868
12.
Hofmann MD. Pacing by winners of a 161-km mountain ultramarathon. International Journal of Sports
Physiology and Performance. 2014; 9(6):1054–6.
13.
Lambert MI, Dugas JP, Kirkman MC, Mokone GG, Waldeck MR. Changes in running speeds in a 100
km ultra-marathon race. Journal of Sports Science and Medicine. 2004; 3(3):167–73. PMID: 24482594
14.
Martin V, Kerhervé H, Messonnier LA, Banfi J-C, Geyssant A, Bonnefoy R, et al. Central and peripheral
contributions to neuromuscular fatigue induced by a 24-h treadmill run. Journal of Applied Physiology.
2010; 108:000–. doi: 10.1152/japplphysiol.01202.2009
Psycho-Physiological Load during a Mountain Ultramarathon
PLOS ONE | DOI:10.1371/journal.pone.0145482
December 21, 2015
12 / 13
15.
Martin V, Millet GY, Lattier G, Perrod L. Why does knee extensor muscles torque decrease after eccen-
tric-type exercise? Journal of Sports Medicine and Physical Fitness. 2005; 45(2):143–51. PMID:
16355074
16.
Waldron M, Worsfold P, Twist C, Lamb K. Concurrent validity and test–retest reliability of a global posi-
tioning system (GPS) and timing gates to assess sprint performance variables. Journal of Sports Sci-
ences. 2011; 29(15):1613–9. doi: 10.1080/02640414.2011.608703 PMID: 22004326
17.
Millet GY, Tomazin K, Verges S, Vincent C, Bonnefoy R, Boisson R-C, et al. Neuromuscular conse-
quences of an extreme mountain ultra-marathon. PLOS One. 2011; 6(2):e17059. doi: 10.1371/journal.
pone.0017059 PMID: 21364944
18.
de Koning JJ, Foster C, Bakkum A, Kloppenburg S, Thiel C, Joseph T, et al. Regulation of pacing strat-
egy during athletic competition. PLOS One. 2011; 6(1):e15863. PMID: 21283744
19.
Achten J, Jeukendrup A. Heart rate monitoring. Applications and limitations. Sports Medicine. 2003; 33
(7):517–38. doi: 10.2165/00007256-200333070-00004 PMID: 12762827
20.
Borg G. Psychophysical bases of perceived exertion. Medicine and Science in Sports and Exercise.
1982; 14(5):377–81. PMID: 7154893
21.
Utter AC, Kang J, Nieman DC, Vinci DM, McAnulty SR, Dumke CL, et al. Ratings of perceived exertion
throughout an ultramarathon during carbohydrate ingestion. Perceptual and Motor Skills. 2003; 97
(1):175–84. PMID: 14604037
22.
Micklewright D, Papadopoulou E, Parry D, Hew-Butler T, Tam N, Noakes TD. Perceived exertion influ-
ences pacing among ultramarathon runners but post-race mood change is associated with perfor-
mance expectancy. South African Journal of Sports Medicine. 2009; 21(4).
23.
Clemente-Suárez VJ. Psychophysiological response and energy balance during a 14-h ultraendurance
mountain running event. Applied Physiology, Nutrition, and Metabolism. 2014; 40(3):269–73. doi: 10.
1139/apnm-2014-0263 PMID: 25693897
24.
Townshend AD, Worringham CJ, Stewart IB. Spontaneous pacing during overground hill running. Med-
icine and Science in Sports and Exercise. 2010; 42(1):160–9. PMID: 20010117
25.
Gimenez P, Kerhervé H, Messonnier LA, Féasson L, Millet GY. Changes in the energy cost of running
during a 24-h treadmill exercise. Medicine and Science in Sports and Exercise. 2013; 45(9):1807–13.
PMID: 23524515
26.
Vincenty T. Direct and inverse solutions of geodesics on the ellipsoid with application of nested equa-
tions. Survey Review XXIII. 1975; 23(176):88–93. doi: 10.1179/sre.1975.23.176.88
27.
Townshend AD, Worringham CJ, Stewart IB. Assessment of speed and position during human locomo-
tion using nondifferential GPS. Medicine and Science in Sports and Exercise. 2008; 40(1):124–32.
PMID: 18091013
28.
Minetti AE, Moia C, Roi GS, Susta D, Ferretti G. Energy cost of walking and running at extreme uphill
and downhill slopes. Journal of Applied Physiology. 2002; 93(3):1039–46. doi: 10.1152/japplphysiol.
01177.2001 PMID: 12183501
29.
Minetti AE, Ardigo LP, Saibene F. Mechanical determinants of the minimum energy cost of gradient run-
ning in humans. Journal of Experimental Biology. 1994; 195:211–25. PMID: 7964412
30.
Davies CTM, Thompson MW. Physiological responses to prolonged exercise in ultramarathon athletes.
Journal of Applied Physiology. 1986; 61(2):611–7. PMID: 3745051
31.
Knechtle B, Wirth A, Knechtle P, Rüst CA, Rosemann T, Lepers R. No improvement in race perfor-
mance by naps in male ultra-endurance cyclists in a 600-km ultra-cycling race. Chinese Journal of
Physiology. 2012; 55(2):125–33. PMID: 22559737
32.
Morin J-B, Tomazin K, Edouard P, Millet GY. Changes in running mechanics and spring–mass behavior
induced by a mountain ultra-marathon race. Journal of Biomechanics. 2011; 44(6):1104–7. doi: 10.
1016/j.jbiomech.2011.01.028 PMID: 21342691
33.
Morin J-B, Samozino P, Millet GY. Changes in Running Kinematics, Kinetics, and Spring-Mass Behav-
ior over a 24-h Run. Medicine and Science in Sports and Exercise. 2011; 43(5):829–36. PMID:
20962690
34.
Vernillo G, Savoldelli A, Zignoli A, Skafidas S, Fornasiero A, Torre AL, et al. Energy cost and kinematics
of level, uphill and downhill running: fatigue-induced changes after a mountain ultramarathon. Journal
of Sports Sciences. 2015; 33(19):1998–2005. doi: 10.1080/02640414.2015.1022870 PMID: 25751128
Psycho-Physiological Load during a Mountain Ultramarathon
PLOS ONE | DOI:10.1371/journal.pone.0145482
December 21, 2015
13 / 13
| The Dynamics of Speed Selection and Psycho-Physiological Load during a Mountain Ultramarathon. | 12-21-2015 | Kerhervé, Hugo A,Millet, Guillaume Y,Solomon, Colin | eng |
PMC4237511 | Speed Trends in Male Distance Running
Timothy N. Kruse1, Rickey E. Carter2, Jordan K. Rosedahl2, Michael J. Joyner3*
1 The University of Washington School of Medicine, 1959 N. E. Pacific Street, Seattle, WA, 98195, United States of America, 2 Department of Health Sciences Research,
Mayo Clinic, 200 First Street SW, Rochester, MN, 55905, United States of America, 3 Department of Anesthesiology, Mayo Clinic, 200 First Street SW, Rochester, MN, 55905,
United States of America
Abstract
The major cycling ‘‘Grand Tours’’ have shown an attenuation of performance over the last decade. This has been interpreted
as circumstantial evidence that newer anti-doping strategies have reduced the use of performance-enhancing drugs. To
examine this idea under more controlled conditions, speed trends for world class 5000 m, 10000 m, and marathon
performances by men from 1980 to 2013 were analyzed. We obtained comprehensive records from the International
Association of Athletics Federations, Association of Road Racing Statisticians, and the Track and Field All-time Performances
database webpages. The top 40 performances for each event and year were selected for regression analysis. For the three
distances, we noted cumulative performance improvements in the 1990s thru the mid-2000s. After the peak speed years of
the mid 2000 s, there has been limited improvement in the 5000 m and 10,000 m and world records set during that time
remain in place today, marking the longest period of time between new records since the early 1940s. By contrast marathon
speed continues to increase and the world record has been lowered four times since 2007, including in 2013. While the
speed trends for 5000 m and 10000 m track results parallel those seen in elite cycling, the marathon trends do not. We
discuss a number of explanations other than improved anti-doping strategies that might account for these divergent
findings.
Citation: Kruse TN, Carter RE, Rosedahl JK, Joyner MJ (2014) Speed Trends in Male Distance Running. PLoS ONE 9(11): e112978. doi:10.1371/journal.pone.0112978
Editor: Maria F. Piacentini, University of Rome, Italy
Received June 18, 2014; Accepted October 17, 2014; Published November 19, 2014
Copyright: 2014 Kruse et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits
unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Data Availability: The authors confirm that, for approved reasons, some access restrictions apply to the data underlying the findings. Ethical restrictions prevent
public deposition of data. Requests for data may be sent to Michael Joyner, [email protected].
Funding: This work was supported by the National Institutes of Health grants R25 GM075148 and UL1 TR000135. The funders had no role in study design, data
collection and analysis, decision to publish, or preparation of the manuscript.
Competing Interests: The authors have declared that no competing interests exist.
* Email: [email protected]
Introduction
The use of performance-enhancing drugs (doping) can be dated
back to the ancient Olympics [1], [2]. Since then athletes have
used a wide range of substances including red wine, caffeine,
nitroglycerin, cocaine, opiates, amphetamines, growth hormone,
blood transfusions, anabolic steroids, and erythropoietin (EPO) in
an
effort
to
gain
a
physiological
advantage
[3].
Because
performance-enhancing drugs compromises the idealized princi-
ples of pure competition, the World Anti-Doping Agency (WADA)
was created in 1999 [4]. Since the formation of WADA, it has
developed widely applied policies and drug testing protocols
(including regular out of competition testing) in an attempt to stop
the apparently wide spread doping in elite sports competition [5].
If recent improvements in athletic performance have been
driven by doping, then improved doping control might be
reflected by a leveling off or declining performances in sports
where doping is thought to be ubiquitous. In recent analyses of
major cycling races including the Tour de France, Giro d’Italia,
Vuelta A Espan˜a, the average speed has been leveling off or
declining [6], [7] since the introduction of improved techniques to
detect use of exogenous EPO in 2005 [4]. However, the analysis of
cycling is confounded by varying race distances, yearly changes in
course, and weather. Endurance running eliminates many of these
confounding factors. The tracks and courses are identical from
year to year. For the two shorter distances, there are numerous
competitive opportunities per year and at least some would likely
have nearly ideal environmental conditions.
With this information as a background, we tested the hypothesis
that elite distance running times would show a pattern of leveling
in the middle 2000 s similar to that seen in cycling. Such a finding
could be explained by improved doping control. We also discuss
alternate explanations including that humans are reaching the
biological limits of performance and the potential role technical
innovations in training and equipment [8], [9]. Finally, any
potential explanation might also be confounded by changes in the
economic landscape associated with world class distance running.
Materials and Methods
We obtained the top male performances from 1980–2013, by
year, in major endurance running races (5000 m, 10000 m on the
track; marathon on road courses) from the International Associ-
ation of Athletics Federations (IAAF: http://www.iaaf.org/home),
Association of Road Racing Statisticians (ARRS: http://www.
arrs.net), and the Track and Field all-time Performances data-
base websites (http://www.alltime-athletics.com/index.html) [10],
[11], [12].
The 1980–2013 epoch was selected because besides new
performance-enhancing drugs (PEDs), a case can be made that
potentially transformative changes in training or equipment has
not occurred. For example, by 1980 high volume and high
intensity training had been widely adopted by top competitors for
PLOS ONE | www.plosone.org
1
November 2014 | Volume 9 | Issue 11 | e112978
several
decades
and
athletes
from
East
Africa
had
been
participating at the international level since the early 1960s. High
quality synthetic tracks were also widely available by 1980 and
carbohydrate loading was widely practiced in the marathon and it
is unclear if technical changes in shoes have had a measureable
impact on performance. While ideas about training have been
refined the extent to which these have been uniformly adopted by
elite athletes, especially the East Africans, is not known [13], [14],
[15].
Beyond
these
training
and
globalization
related
factors,
professionalism also emerged during the 1980s. We also restricted
our analysis to men because women were not routinely permitted
to participate in long distance racing until the 1970s and
performances
dropped
dramatically
in
the
early
years
of
widespread competition by women. While the gap in world
records has been steady since that time women still lag in
competitive depth in many events [16], [17].
Finally, the first synthetic EPO (Epogen) was approved by the
FDA in 1989 and within a few years it was clear that EPO can
have profound effects on maximal oxygen transport (VO2 max)
[18] in humans and was being used to enhance athletic
performance by the early 1990s. Additionally, because EPO and
related analogues are injectable the logistical challenges of
traditional blood doping (autologous red cell transfusions) are
eliminated [19].
The abstracted data consisted of the total number of perfor-
mances
below
2012’s
Olympic
A
standard:
5000 m-13:20,
10,000 m-27:45, and also performances under 2:10:00 for the
marathon (the Olympic A standard was 2:15 in 2012, a time that
equals an estimated 29.22 10,000 m [20]). Performances below
these standards were considered ‘elite’. To study the speed tends
more formally, the 40 fastest athlete performances (the fastest
performance of the 40 fastest athletes) were recorded per year and
event for regression modeling (described below). Age and country
of residence at time of race were also abstracted.
Initial data summaries included the frequency tabulation elite
performances by event and distance. To analyze the changes in
speed trends over the study period, we used regression techniques
consisting of quadratic splines (cubic B-splines with 3 equally
spaced interior knots) against our years of interest using the top 40
athlete performances per year. These generalized regression
models allowed for flexibility of estimating the change in
performance over time while providing traditional measures of
model fit (e.g., R-square value). It was hypothesized that different
regression profiles would be observed between the top 40 finishing
times and the fastest yearly performances, so the percentage
difference in speed of the fastest performance relative to the 40th
fastest time per year was also modeled using regression splines and
locally weighted smoothers (LOESS). When reporting measures of
model fit for the cubic B-splines regression models (i.e., change in
speed as a function of year), the omnibus F test for the regression
model and R-square were reported. The LOESS curves, which
were used for illustrative purposes of changes in pacing based on
the relative placing, were not summarized using traditional
regression summaries such as R-square on account of their
intended visual utilization.
Cubic B-spli ne regression analyses were conducted using The
SAS System (v9.3, Cary, NC) using PROC ORTHOREG.
LOESS smoothing was conducted using PROC SGPLOT using
default parameters.
Results
The number of performances below the 2012 Olympic A
qualifying standard plus sub 2:10:00 for each distance increased
over the 1980–2013 (Figure 1a–c). The world record for the
5000 m was set in 2004 while the 10000 m world record was set in
2005; these records stand today, which is the longest gap between
world records since the 1940s. The number of performances below
the 2012 Olympic A qualifying standard for the 5000 m and
10,000 m also appears to have leveled off since the middle 2000 s.
Similarly the number of athletes breaking 2:10:00 for the
marathon has also leveled off. 2:10:00 was chosen as a comparable
standard for the marathon because this time is considered
generally similar to or slightly slower than the A standards for
shorter distances based on various empirical point tables, scoring
systems and time conversion programs [20].
All three regression splines presented in Figures 2a–c were
statistically significant (p,0.0001 for each). Furthermore, year
alone explained a large percentage of the variation in the speed
trends (R-square: 53%, 37% and 69% for 5000 m, 10000 m and
marathon, respectively). Consistent with these overall model
estimates, the 5000 m and 10000 m had significant increases in
speed during the 1990s whereas the marathon showed an increase
over the entire three plus decades (Figures 2a–c). In particular, the
5000 m the speed trend levels off starting around 2000. The
marathon and 10000 m times do not show this as a pronounced
tendency.
Figures 3a–c explores the speed trends using an alternative
classification approach to provide additional insights into the
temporal effects. The fastest performance of the year is plotted
alone and summarized using a LOESS smoother. Speed trends of
the mean top 10, mean top 20 and mean top 40 athlete
performances are superimposed in these same figures. With the
5000 m and 10000 m, there is a pronounced ‘outlying’ effect of
the top performance from 1995 to late 2010. The marathon,
however, displays no attenuation of the increased speed over the
epoch sampled and the relative speed of the fastest annual time
does not appear to be an outlier (i.e., the figure lines appear as
roughly parallel).
To better quantify the observations made from Figure 3, the
percentage changes in speeds over time were examined and found
to be consistent with the differential
findings of the top
performance vs. the 40th fastest athlete performance. The degree
to which the fastest times were relatively fast (compared to other
years) was observed during the 2000s in the 5000 m and 10000 m
distances. The regression spline analyses supported these findings
that the fastest relative times for the 5000 m and 10000 m varied
over the epoch (5000 m: p = 0.048, R-square = 32%; 10000 m:
p = 0.0007, R-square = 51%). As illu strated in Figure 3c, trends
for the marathon distance were not clearly identified in the data
(p = 0.29, R-square = 19%).
Discussion
The speed and performance trends for top 5000 m and
10000 m distance running performers on the track show a period
of increased speed among the fastest runners to the mid-2000 s
with an attenuation of speed in either all (5000 m) or the fastest
performances (10000 m) after this period of time. For the
marathon, all indices of speed show a nearly linear increase in
speed with an increased number of elite performances over the
three plus decades we sampled. We believe there are a number of
possible explanations for these findings.
First, the findings for the 5000 m can be interpreted as
consistent with the hypothesis that improved drug testing has
Doping and Running Times
PLOS ONE | www.plosone.org
2
November 2014 | Volume 9 | Issue 11 | e112978
limited the ability of elite athletes to manipulate their oxygen
transport systems with EPO (or other techniques to improve
oxygen transport during exercise) since the middle 2000 s. These
observations are also broadly consistent with recent speed trends in
elite cycling races [6], [7]. This interpretation can also be applied
to the 10000 m results, but only when considering the fastest
times. By contrast, the data for the marathon shows continued
improvements in running speed during the same time period along
with more total elite performances and world records. These
observations challenge the idea that the speed leveling seen in the
5000 m on the track and in the so-called ‘‘Grand Tours’’ of
cycling is due primarily to better drug testing and the reduced use
of performance enhancing drugs.
A second possible interpretation is that world class performanc-
es are leveling off and reaching a physiological upper limit as has
been postulated for equine and canine athletes [8], [9,], [21]. In
the case of the marathon a number of empirical estimates and
physiological modeling suggest the record is relatively slow in
comparison to the 5,000 m and 10,000 m times and is merely
catching up by comparison [20], [22], [23]. In this context, it is
Figure 1. Total number of elite performances by year. Times under 13:20 for the 5000 m, 27:45 for the 10000 m, and 2:10 for the marathon.
doi:10.1371/journal.pone.0112978.g001
Doping and Running Times
PLOS ONE | www.plosone.org
3
November 2014 | Volume 9 | Issue 11 | e112978
interesting to note that top speeds have not fallen for the shorter
races but only leveled off.
The third element of any interpretation focuses on the changing
financial incentives in professional distance running. Prize money
for top marathon performances has increased. Specifically, in 1980
the highest total payout for any marathon was $50,000; just over
two decades later the first million dollar race was run [24]. These
incentives could be attracting a stronger pool of competitors to
‘‘move up’’ and focus on the marathon and forgo record setting
attempts at shorter distances. This could lead to more competitive
races among top runners at the major marathons. Second the
Figure 2. Top 40 athlete performances of the 5000 m, 10000 m,
and marathon (circles). The solid line is a quadratic spline fit with
three equally spaced knots (points of inflection).
doi:10.1371/journal.pone.0112978.g002
Figure 3. LOESS smoothers fit through the fastest yearly
performance by year for the 5000 m (A), 10000 m (B) and
marathon (C). In addition, the mean speed for the top 10, top 20 and
top 40 athlete performances are also plotted for reference.
doi:10.1371/journal.pone.0112978.g003
Doping and Running Times
PLOS ONE | www.plosone.org
4
November 2014 | Volume 9 | Issue 11 | e112978
highest profile marathon races are now being staged in a way
designed for world record attempts that include the use of pacers.
Along these lines, the use of pacers has been wide spread for races
on the track for many years, and many top athletes have bonus
plans and other financial incentives from sponsors that reward fast
times at the shorter distances. There are a number strengths and
limitations to this study. A major strength of our data set and
analysis is that it includes standardized distances and courses with
numerous competitive opportunities at the shorter two distances
when environmental conditions are likely to be optimal. By
contrast, a limitation to our analysis is that we have no idea if
improved approaches to training or equipment (shoes and tracks)
might have contributed to the trends we report. However, we
favor the interpretation that the entire epoch we have analyzed has
been relatively stable from a technical perspective. This includes
widespread use of high volume and high intensity training,
widespread availability of synthetic tracks, and adequate footwear.
Additionally, while ideas about training have been refined it is not
known if how uniformly these have been adopted by elite athletes,
especially the East Africans [13], [14], [15]. This perspective
contrasts to the major improvements in equipment for cycling that
includes use of advanced materials and improved aerodynamic
designs to construct faster bikes.
A final concern whenever the topic of doping is raised is
discussed relates to what might be called the continuous ‘‘cat and
mouse’’ game between those trying to enforce the rules with
improved testing and those trying to circumvent them. This has
engendered speculation that micro-doses of EPO can be titrated
by athletes in a way to achieve high levels of performance and yet
avoid a positive drug test [19], [25], [26]. There is also widespread
speculation about the use less or undetectable compounds and so-
called designer performance enhancing drugs. Advocates of these
points of view have argued that while doping is considered
widespread the number of positive tests in major competitions is
quite low [5]. The counter argument is that the low number of
positive results demonstrates that the testing is working and
deterring doping. The lack of hard data on the true incidence of
doping and how it has or has not been influenced by improved
testing is unknown and a major limitation to any discussion on this
topic. However, it is clear that anonymous questionnaire based
surveys suggest the true incidence of doping it is much higher (14–
39%) than ,2% rate of positive tests suggests [27]. This is clearly
an area of sports sociology that requires increased attention.
It should also be noted that the sociology surrounding the
doping phenomenon along with the ongoing incentives to dope
are complex. In this context, strategies beyond testing alone will be
required to improve the efficacy of doping control. A compre-
hensive discussion of this complex topic is beyond the scope of our
analysis, but there has been much thoughtful discussion of related
topics [28], [29], [30], [31], [32], [33], [34].
Conclusion
In summary, our analysis demonstrates that speed trends for
elite distance running are divergent depending on distance and
have event specific patterns. Thus, any generalizations about
performances in world class competition providing evidence that
drug testing is or is or is not ‘‘working’’ need to be viewed with
caution. Further caution is required given the many caveats and
potential factors that could explain our findings.
Acknowledgments
We would like to thank IAAF, Peter Larsson, and Ken Young for
compiling the data. This work was supported by the National Institutes of
Health grants R25 GM075148 and UL1 TR000135.
Author Contributions
Conceived and designed the experiments: TNK MJJ REC. Performed the
experiments: TNK MJJ. Analyzed the data: TNK MJJ REC JKR.
Contributed reagents/materials/analysis tools: TNK MJJ REC JKR.
Contributed to the writing of the manuscript: TNK MJJ REC.
References
1. Conti AA (2010) Doping in sports in ancient and recent times. Med Secoli 22:
181–190.
2. Papagelopoulos PJ, Mavrogenis AF, Soucacos P (2004) Doping in ancient and
modern Olympic Games. Orthopedics 27: 1226–1231.
3. Gaudard A, Varlet-Marie E, Bressolle F, Audran M (2003) Drugs for increasing
oxygen transport and their potential use in doping - A review. Sports Med 33:
187–212, doi:10.2165/00007256-200333030-00003.
4. World Anti-Doping Agency (2010) A Brief History of Anti-Doping.
5. International Olympic Committee (2014) The Fight Against Doping and
Promotion of Athletes’ Health.
6. El Helou N, Berthelot G, Thibault V, Tafflet M, Nassif H, et al. (2010) Tour de
France, Giro, Vuelta, and classic European races show a unique progression of
road cycling speed in the last 20 years. J Sports Sci 28: 789–796, doi:10.1080/
02640411003739654.
7. Perneger TV (2010) Speed trends of major cycling races: does slower mean
cleaner? Int J Sports Med 31, 261–264, doi:10.1055/s-0030-1247593.
8. Berthelot G, Tafflet M, El Helou N, Len S, Escolano S, et al. (2010) Athlete
atypicity on the edge of human achievement: performances stagnate after the last
peak, in 198. PLoS One 5: e8800, doi:10.137/journal.pone.0008800.
9. Berthelot G, Thibault V, Tafflet M, Escolano S, El Helou N, et al. (2008) The
citius end: world records progression announces the completion of a brief ultra-
physiological quest. PLoS One 3: e1552, doi:10.1371/journal.pone.0001552.
10. Home of World Athletics (2014) International Association of Athletics
Federation. Available: http://www.iaaf.org/records/toplists. Accessed 15 May
2014.
11. Young K (13 June 2014) Yearly Rankings- Marathon. Association of Road
Racing Statisticians. Available: http://www.arrs.net. Accessed 15 May 2014.
12. Larsson P (16 Jun 2014) Men’s standard events. Track & Field All-time
Performances Homepage. Available: http://www.alltime-athletics. com/index.
html. Accessed 15 May 2014.
13. Billat V, Lepretre PM, Heugas AM, Laurence MH, Salim D, et al. (2003)
Training and bioenergetic characteristics in elite male and female Kenyan
runners. Med Sci Sports Exerc 35: 297–304.
14. Wilber RL, Pitsiladis YP (2012) Kenyan and Ethiopian distance runners: what
makes them so good? Int J Sports Physiol Perform 7: 92–102.
15. Bourne ND (2008) Fast science: A history of training theory and methods for
elite runners through 1975, Doctor of Philosophy Thesis, University of Texas.
16. Thibault V, Guillaume M, Berthelot G, Helou NE, Schaal K, et al. (2010)
Women and men in sport performance: the gender gap has not evolved since
1983. J Sports Sci Med 9: 214–223.
17. Hunter SK, Stevens AA, Magennis K, Skelton KW, Fauth M (2011) Is there a
sex difference in the age of elite marathon runners? Med Sci Sports Exerc 43:
656–664, doi:10.1249/MSS.0b013e3181fb4e00.
18. Joyner MJ (2003) VO2MAX, blood doping, and erythropoietin. Br J Sports
Med 37: 190–191.
19. Lundby C, Achman-Andersen NJ, Thomsen JJ, Norgaard AM, Robach P (1985)
Testing for recombinant human erythropoietin in urine: problems associated
with current anti-doping testing. J Appl Physiol 105: 417–419, doi:10.1152/
japplphysiol.90529.2008.
20. Hill R (1 Jun 2014) Race Conversion. Hill Runner. Available: http://www.
hillrunner.com/calculators/raceconversion.php. Accessed 17 May 2014.
21. Denny MW (2008) Limits to running speed in dogs, horses and humans. J Exp
Biol 211: 3836–3849, doi:10.1242/jeb.024968.
22. Joyner MJ, Ruiz JR, Lucia A (2011) The two-hour marathon: who and when?
J Appl Physiol (1985) 110: 275–277. doi:10.1152/japplphysiol. 00563.2010.
23. Joyner MJ (1991) Modeling: optimal marathon performance on the basis of
physiological factors. J Appl Physiol (1985) 70: 683–687.
24. Miller T, Lawrence G, McKay J, Rowe D (2001) Globalization and Sports.
(SAGE).
25. Jelkmann W, Lundby C (2011) Blood doping and its detection. Blood 118:
2395–2404, doi:10.1182/blood-2011-02-303271.
26. Lonnberg M, Lundby C (2013) Detection of EPO injections using a rapid lateral
flow isoform test. Anal Bioanal Chem 405: 9685–9691, doi:10.1007/s00216-
013-6997-8.
Doping and Running Times
PLOS ONE | www.plosone.org
5
November 2014 | Volume 9 | Issue 11 | e112978
27. de Hon O, Kuipers H, van Bottenburg M (2014) Prevalence of doping use in
elite sports: a review of numbers and methods. Sports Med Aug 29 [Epub ahead
of print].
28. Delanghe JR, Joyner MJ (2008) Testing for recombinant human erythropoietin.
J Appl Physiol (1985) 105: 395–396, doi:10.1152/japplphysiol.90746.2008.
29. D’Angelo C, Tamburrini C (2010) Addict to win? A different approach to
doping. J Med Ethics 36: 700–707.
30. Baron DA, Martin DM, Abol Magd S (2007) Doping in sports and its spread to
at-risk populations: an international review. World Psychiatry Off J World
Psychiatr Assoc WPA 6: 118–123.
31. Tokish JM, Kocher MS, Hawkins RJ (2004) Ergogenic aids: a review of basic
science, performance, side effects, and status in sports. Am J Sports Med 32:
1543–1553.
32. Haugen KK (2004) The performance-enhancing drug game. J Sports Econ 5:
67–86.
33. Noakes TD (2004) Tainted glory–doping and athletic performance.
N Engl J Med 351: 847–849.
34. Strelan P, Boeckmann RJ (2006) Why drug testing in elite sport does not work:
perceptual deterrence theory and the role of personal moral beliefs. J Appl Soc
Psychol 36: 2909–2934.
Doping and Running Times
PLOS ONE | www.plosone.org
6
November 2014 | Volume 9 | Issue 11 | e112978
| Speed trends in male distance running. | 11-19-2014 | Kruse, Timothy N,Carter, Rickey E,Rosedahl, Jordan K,Joyner, Michael J | eng |
PMC7663387 | International Journal of
Environmental Research
and Public Health
Review
A Scoping Review of the Relationship between
Running and Mental Health
Freya Oswald 1,*
, Jennifer Campbell 1
, Chloë Williamson 2, Justin Richards 3 and Paul Kelly 2
1
Edinburgh Medical School, The University of Edinburgh, Edinburgh EH16 4TJ, UK; [email protected]
2
Physical Activity for Health Research Centre (PAHRC), University of Edinburgh, Edinburgh EH8 8AQ, UK;
[email protected] (C.W.); [email protected] (P.K.)
3
Faculty of Health, Victoria University Wellington, Wellington 6140, New Zealand; [email protected]
*
Correspondence: [email protected]
Received: 10 October 2020; Accepted: 29 October 2020; Published: 1 November 2020
Abstract: Poor mental health contributes significantly to global morbidity. The evidence regarding
physical benefits of running are well-established. However, the mental health impacts of running
remain unclear. An overview of the relationship between running and mental health has not
been published in the last 30 years. The purpose of this study was to review the literature on the
relationship between running and mental health. Our scoping review used combinations of running
terms (e.g., Run* and Jog*) and mental health terms (general and condition specific). Databases used
were Ovid(Medline), Ovid(Embase), ProQuest and SportDiscus. Quantitative study types reporting
on the relationships between running and mental health were included. Database searches identified
16,401 studies; 273 full-texts were analysed with 116 studies included. Overall, studies suggest that
running bouts of variable lengths and intensities, and running interventions can improve mood and
mental health and that the type of running can lead to differential effects. However, lack of controls
and diversity in participant demographics are limitations that need to be addressed. Cross-sectional
evidence shows not only a range of associations with mental health but also some associations with
adverse mental health (such as exercise addiction). This review identified extensive literature on the
relationship between running and mental health.
Keywords: exercise; mental health; psychology; physical activity; running; jogging
1. Introduction
Poor mental health contributes significantly to the global health burden [1]. The strain of mental
health and behavioural disorders is estimated to account for more years of lived disability than any
other chronic health ailment [1,2]. The global proportion of disability-adjusted life years caused by
mental ill-health has increased from 12.7% to 14% (males) and 13.6% to 14.4% (females) from 2007 to
2017 [3]. Due to the burden and increasing prevalence of mental ill-health, effective management of
mental health disorders is vital [4].
There is substantial evidence to support the relationship between physical activity (PA) and various
mental health outcomes across the lifespan [5–7]. There has been investigation of low-intensity PA on
mental health; for example, Kelly et al. (2018) reported the positive relationships between walking and
mental health in an earlier scoping review [8]. However, a similar synthesis for higher-intensity PA
such as running has not been reported.
While the evidence base for the benefits of running on physical health is well-established,
the mental health changes from running remain unclear. Addressing the gap within this knowledge is
valuable as running is a form of PA popular among many population groups [9]. Inclusive organisations
such as “Couch to 5k” [10], “Girls on the run” [11] and “Parkrun” can support running while promoting
Int. J. Environ. Res. Public Health 2020, 17, 8059; doi:10.3390/ijerph17218059
www.mdpi.com/journal/ijerph
Int. J. Environ. Res. Public Health 2020, 17, 8059
2 of 39
well-being and satisfaction with physical health, facilitating socialisation and community connectedness,
and reducing loneliness [12–14]. In primary care settings, national initiatives such as “Parkrun-Practice”
promote well-being through running [15].
In recent years, there has been a transition within healthcare to focus on disease morbidity rather
than disease mortality, in particular with a drive to improve global mental health [16]. There is
increasing prevalence of mental ill-health; therefore, effective management of mental health disorders
is vital [4]. In order to investigate any differences in mental health effects between high and low
intensities of running, all genres of running must be considered including jogging, sprinting, marathon
running and orienteering.
To the best of the authors’ knowledge, no recent reviews of the relationship between running and
mental health are available. The synthesis provided by this review will enable healthcare practitioners,
psychologists and policy makers to better advise on running for mental health. It will also identify key
gaps in the literature for future research. The aims of this scoping review are the following:
(1)
to provide an overview of what is known regarding the relationship between running and mental
health outcomes in all age groups and populations
(2)
to highlight current knowledge gaps and research priorities
2. Materials and Methods
A scoping review was concluded to be the most appropriate to address the research aims as it
provides an overview of the volume and distribution of the evidence base as well as highlights where
more research is warranted. The review followed the five-stage scoping review framework proposed
by Arksey and O’Malley and was guided by the Preferred Reporting Items for Systematic Reviews and
Meta-Analyses (PRISMA) scoping review extension checklist (Appendix A) [17,18].
2.1. Identify Research Question
Research questions were developed to address the research aims: “What is known about the
effects of running on mental health outcomes?” and “What are the current knowledge gaps?”.
Research question formulation was guided by item 4 in the PRISMA scoping review extension checklist
(Appendix A). The definition of running included jogging, sprinting, marathon running, orienteering
and treadmill running. A wide range of intensities were included as the aim of the scoping review
was to provide an overall picture of the relationship between running (of various intensities) and
mental health.
2.2. Identify Relevant Outcomes
Mental health outcomes were informed by Kelly et al. (2018) [8], who reviewed the relationships
between walking and mental health (Table 1). Measures or disorders of cognitive dysfunction were
considered neurological and thus outside the scope of this review. Eating disorders were included as
they significantly impair physical health or psychosocial functioning. Health-related quality-of-life
was excluded as it was considered to incorporate physical, social, emotional and mental factors.
Int. J. Environ. Res. Public Health 2020, 17, 8059
3 of 39
Table 1. Definitions of the mental health outcomes included within the review: the outcomes were
informed by Kelly et al. (2018) [8].
Outcome
Description
Depression
Depression is a mood disorder with prolonged periods of low mood and a lack of interest and/or pleasure
in normal activities most of the time. This includes major depressive disorder [19].
Anxiety
Anxiety is characterised by uncomfortable or upsetting thoughts and is usually accompanied by agitation,
feelings of tension and activation of the autonomic nervous system. It is important to note the distinction
between transient anxiety symptoms (state anxiety), persistent symptoms (trait anxiety) and anxiety
disorders: a collection of disabling conditions characterised by excessive, chronic anxiety. Examples of
anxiety disorders are specific phobias, social phobia, generalised anxiety disorder, panic disorder,
obsessive–compulsive disorder and post-traumatic stress disorder [20].
Self-efficacy
Self-efficacy is a situation-specific form of self-confidence. Self-efficacy beliefs influence how people think,
feel, motivate themselves and act [21].
Psychological stress
Psychological stress or distress can be defined as the unique discomforting, emotional state experienced
by an individual in response to a specific stressor or demand that results in harm, either temporary or
permanent, to that person [22].
Eating pathology
Eating pathology or disorder can be described as persistent disturbance of eating behaviours or
behaviours intended to control weight, which significantly impairs physical health or psychosocial
functioning. This disturbance should not be secondary to any recognised general medical disorder,
e.g., hypothalamic tumour. This definition includes anorexia nervosa and bulimia nervosa [23].
Self-esteem
Self-esteem is the feelings of value and worth that a person has for oneself. It contributes to overall
self-concept as a construct of mental health [24].
Addiction
Addiction designates a process whereby a behaviour that can function both to produce pleasure and to
provide escape from internal discomfort is employed in a pattern characterized by (1) recurrent failure to
control the behaviour (powerlessness) and (2) continuation of the behaviour despite significant negative
consequences (unmanageability) [25].
Psychological well-being
Psychological well-being links with autonomy, environmental mastery, personal growth, positive relations
with others, purpose in life and self-acceptance. This is often referred to as eudemonic well-being [26].
Self-concept
Self-concept is the organisation of qualities that the individual attributes to themself, which in turn guides
or influences the behaviour of that individual [27].
Mood
Mood is a transient state of a set of feelings, usually involving more than one emotion. Seen as a conscious
summative recognition of feelings that can vary in intensity and duration [28].
2.3. Identify Relevant Studies
Studies were included based on the following criteria:
•
Any geographical location
•
All years between 1970 and 2019
•
Quantitative effects of running on predetermined mental health outcomes
#
Preventive effects (negative)
#
Health promotion effects (positive)
#
Intervention effects
•
Any age group or sex
•
Human studies
•
Designs including primary research (cross-sectional, longitudinal, interventions and natural
experiments with pre-post measures with or without non-running comparisons)
•
Studies that mentioned walking as well as running were included because it is not possible to
differentiate walkers from runners in events such as Parkrun.
Studies were excluded based on the following criteria:
•
Specialist groups including elite, professional or competitive athletes.
•
General physical or aerobic activity, rather than exclusively running
•
Qualitative and ethnographic designs
•
Systematic and scoping reviews (individual studies from identified reviews were included
if relevant)
Int. J. Environ. Res. Public Health 2020, 17, 8059
4 of 39
•
Editorials, opinion pieces, magazine/newspaper articles, case reports and papers without
primary data
•
Focus on secondary mental health within clinical groups with specific physical or mental
conditions that is not the condition being treated with running (e.g., effects on depression in
patients with cancer)
•
Evidence types including guidelines, unpublished and ongoing trials, annual reports, dissertations
and conference proceedings
•
Animal studies
•
Unavailable in English
•
Running intervention was part of a wider study where differentiating the individual effect of
running was not possible (e.g., combined with weight management).
•
Conference abstracts that were not published as full articles
Search Strategy and Databases
Databases searched were Ovid (Medline), Ovid (Embase), ProQuest and SportDiscus. Databases
were searched for titles and abstracts that included at least one running term with one mental health
outcome term. Appropriate truncation symbols were used to account for search term variations.
Common running terms were combined. Search terms and the full search syntax can be found in
Appendix B. Searches were conducted for papers published up to August 2019.
2.4. Study Selection
All identified records were uploaded to Covidence (https://www.covidence.org), and duplicates
were automatically removed. Titles and abstracts were screened, with 20% cross-checked early in the
process to assess agreement between authors. Full texts were reviewed by 2 authors.
2.5. Charting the Data
Data extraction was completed by the lead author (F.O.) with 5% double screened by a second
author (J.R.). The data extraction form was pilot tested with the first 20 studies and informed the
following standardised extraction agreed upon by all authors:
(1)
Author(s), year of publication and geographical location of study
(2)
Mental health conditions examined
(3)
Sample size and population details
(4)
Study design
(5)
Measures used to quantify any change in mental health outcome(s)
(6)
Running dose (if applicable) and compliance (if applicable)
(7)
Whether running was beneficial and the main findings
In studies that used “Profile of Mood States” (POMS) as a measurement of mood state, total mood
disturbance was used in this review if reported by the authors. If the authors only reported one/some
of the POMS subdimensions, these data were extracted instead.
2.6. Collating, Summarising and Reporting Results
Included studies were organized into 3 categories: cross-sectional studies, acute (single, double or
triple) bouts of running, and long-term running interventions. For each of these 3 categories, the results
were presented in two ways: (a) a descriptive numerical analysis to highlight the prevailing domains
of research regarding geographical location, mental health outcomes and research methods and (b) a
narrative summary of the key findings.
Int. J. Environ. Res. Public Health 2020, 17, 8059
5 of 39
3. Results
3.1. Included Studies
From initial searches, 29,851 papers were identified. Following removal of duplicates, 16,401 were
screened at the title and abstract levels and 273 papers were retained for full-text assessments. Ultimately,
116 papers met the inclusion criteria for this review. Figure 1 displays the PRISMA study flowchart.
The results are presented in the following 3 categories: cross-sectional studies, acute bouts of running
and longer-term interventions.
Int. J. Environ. Res. Public Health 2020, 17, x
5 of 66
From initial searches, 29,851 papers were identified. Following removal of duplicates, 16,401
were screened at the title and abstract levels and 273 papers were retained for full-text assessments.
Ultimately, 116 papers met the inclusion criteria for this review. Figure 1 displays the PRISMA study
flowchart. The results are presented in the following 3 categories: cross-sectional studies, acute bouts
of running and longer-term interventions.
Figure 1. Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) depicting
the identification, screening, eligibility and inclusions of texts within the scoping review.
3.2. Category 1: Cross-Sectional Studies
Forty-seven studies utilised cross-sectional designs (with and without non-running comparison
groups) (Table 2) [29–75]. These studies assessed exposure to regular running by questionnaire.
Narrative description of findings of the 47 cross-sectional studies are included within Table S1 within
the supplementary material.
Figure 1. Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) depicting
the identification, screening, eligibility and inclusions of texts within the scoping review.
3.2. Category 1: Cross-Sectional Studies
Forty-seven studies utilised cross-sectional designs (with and without non-running comparison
groups) (Table 2) [29–75]. These studies assessed exposure to regular running by questionnaire.
Narrative description of findings of the 47 cross-sectional studies are included within Table S1 within
the supplementary material.
Int. J. Environ. Res. Public Health 2020, 17, 8059
6 of 39
Table 2. Summary of data extraction from the 47 cross-sectional studies.
Author
Year
Country
Design
Population
Mental Health Outcome
(Measurement)
Study Aim
Main Findings
Wilson et al. [29]
(1980)
Canada
Cross-sectional
n = 30, all male; age range 20–45;
10 marathoners, 10 regular
joggers and 10 non-exercisers
Mood (Profile of Mood States)
Comparing mood states of
marathon runners, regular
joggers and non-exercisers
Marathoners and joggers reported less depression (F(2,28) = 7.51, p < 0.003), anger (F = 10.11,
p < 0.001) and confusion (F = 12.41, p < 0.001) and more vigour (F = 103.21, p < 0.001) than
non-exercisers. Marathoners reported less fatigue (F = 10.26, p < 0.001) and tension (F = 7.51,
p < 0.003) than non-exercisers. Marathoners and joggers did not significantly differ on
reported fatigue and tension; however, marathoners had significantly less depression, anger
and confusion but more vigour than joggers.
Joesting [30]
(1981)
USA
Controlled
cross-sectional
n = 100 runners; 79 males,
mean age 18.36; 21 females,
mean age 16.53
Depression (Depression Adjective
Checklist)
Investigating the relationship
between running and depression
Significantly (p < 0.01) decreased depression in males and female runners compared to
Lubin’s data for nonpsychiatric patients: male and female runners mean depression scores
were 4.59 and 4.33, respectively, while the normative nonpsychiatric sample means were 8.02
and 7.32, respectively.
Jorgenson et al.
[31]
(1981)
USA
Cross-sectional
n = 454 regular runners;
390 males and 64 females;
majority aged 30–39
Emotional well-being (structured
questionnaire consisting of 55 items
designed by the author)
Investigating the relationship
between emotional well-being
and running
Of the runners, 92.3% (n = 419) indicated an increase in emotional well-being (p < 0.01) but
no report on the scale of improvement. Age and emotional well-being were significantly
correlated (gamma value = 0.42, p < 0.001), with the older runner having the greater
perception of emotional well-being resulting from running. There was a significant inverse
relationship between average hours per week running and emotional well-being (gamma
value = −0.43, p < 0.001).
Valliant et al. [32]
(1981)
Canada
Cross-sectional
n = 68 male runners; 30 marathon
runners, mean age 34.4;
38 recreational runners,
mean age 20.6
Self-sufficiency and personality profiles
(a 1-h “Sixteen Personality Factor
Questionnaire”)
Comparing self-sufficiency and
personality profiles in marathon
runners vs. recreational joggers
Marathon runners had a more self-sufficient personality compared to joggers who were less
assertive and more conscientious and had controlled personality types: On average,
marathoners more reserved (F = 17.07, df = 1,66, p < 0.001), intelligent (F = 12.69, df = 1,66,
p < 0.001), tender-minded (F = 11.79, df = 1,66, p < 0.001), imaginative (F = 11.09, df = 1,66,
p < 0.005) and self-sufficient (F = 19.84, df = 1,66, p < 0.001) than joggers. Conversely, joggers
were more happy-go-lucky (F = 10.05, df = 1,66, p < 0.005), apprehensive (F = 10.51, df = 1,66,
p < 0.005) and controlled (F = 7.09, df = 1,66, p < 0.01).
Francis et al. [33]
(1982)
USA
Cross-sectional
n = 44 male participants; mean
age 32; non-running controls who
ran 0 miles weekly (n = 16),
20 miles (n = 10), 30–40 miles
(n = 8) and 50–60 miles (n = 10)
Anxiety, depression and hostility
(State-Trait Anxiety Inventory and the
Multiple Affect Adjective Check List)
Comparing anxiety, depression
and hostility in various groups of
runners vs. sedentary controls
Compared to sedentary controls, runners had lower anxiety (4.2 vs. 7.2, p < 0.01),
depression (8.6 vs. 12.3, p < 0.01) and hostility (4.8 vs. 6.8, p < 0.01).
Hailey et al. [34]
(1982)
USA
Cross-sectional
n = 60 male runners; aged 13–60;
Those who ran for less than 1
year (n = 12), those who ran for
1–4 years (n = 32) and those who
ran for over 4 years (n = 16)
Negative addiction (Negative
addiction scale)
Investigating the relationship
between running and negative
addiction
The more years that a male had been running, the greater the risk of developing negative
addiction (F(2,58) = 3.48, p < 0.05). Runners with a running history of <1 year scored a mean
of 3.84 (scale of 1–14), those running for 1–4 years scored 5.63 and those running for 4+ years
scored 6.38. Addiction scores for runners of 4+ years was greater than the addiction score for
runners of <1 year (t(59) = 2.72, p < 0.005). Likewise, the addiction score for runners of
between 1–4 years was greater than the score for runners <1 year (t(59) = 2.52, p < 0.01).
No statistically significant difference in addiction scores were found between the 1–4-year
group and the 4+ year group.
Callen [35]
(1983)
USA
Cross-sectional
n = 424 non-professional runners
who ran on average more than
28.8 miles per week; 303 males
and 121 females; mean age 34
Mental and emotional aspects (a
questionnaire designed by the author)
Investigating mental and
emotional aspects associated with
long-distance running in
non-professional runners,
including depression, tension,
mood, happiness, self-confidence
and self-image
Ninety-six percent of subjects noticed mental/emotional benefits from running, but none
reported the size of benefits. Benefits included relief of tension (86% of all respondents, n.s.),
improved self-image (77%, n.s.), better mood (66%, p < 0.05), improved self-confidence (64%,
n.s.), relieved depression (56%, p < 0.05) and improved happiness (58%, n.s.). However, 25%
stated they had experienced emotional problems associated with running, with almost every
instance being a problem of depression, anger or frustration associated with not being able to
run due to injury, but no details of size or significance were reported. Sixty-nine percent of
runners experienced an emotional “high” while running.
Galle et al. [36]
(1983)
USA
Controlled
cross-sectional
n = 391 female subjects; aged 15
to 50; runners (n = 102), infertility
patients (n = 103), fertile subjects
(n = 139) and Clomid study
patients whose only infertility
abnormality was ovulation
dysfunction (n = 47)
Anxiety and depression (Hopkins
Symptom Checklist-90)
Comparing psychologic profiles
including anxiety and depression
in runners, infertility patients,
fertile subjects and Clomid study
patients whose only infertility
abnormality was ovulation
dysfunction
Emotional distress scores of runners were not significantly different from the fertile control
subjects (F = 1.19, ns), but both groups of infertility patients showed greater distress on items
in the depression subscale than the runners and fertile control subjects (F = 3.42, p < 0.025).
The only significant difference between runners and fertile control subjects was that control
subjects had higher hostility (p < 0.05). Regarding just runners, there was significant
differences in depression between amenorrhoeic (n = 15) and regular cycling runners (n = 87),
with amenorrhoeic runners scoring higher in the depression factor than regular ovulation
cycle runners (F = 3.0, p < 0.10).
Int. J. Environ. Res. Public Health 2020, 17, 8059
7 of 39
Table 2. Cont.
Author
Year
Country
Design
Population
Mental Health Outcome
(Measurement)
Study Aim
Main Findings
Lobstein et al. [37]
(1983)
USA
Pre-post controlled
between subject
design
n = 22 medically healthy men;
11 physically active men and 11
sedentary men; aged 40–60
Depression (Minnesota Multiphasic
Personality Inventory)
Assessing the impact of a
treadmill run with increasing
gradient on depression
Sedentary men were significantly more depressed than men who ran (mean = 61.36 vs. 50.73,
respectively, p < 0.01), but both groups were within clinical limits for normal, mentally
healthy, middle aged men.
Rudy et al. [38]
(1983)
USA
Cross-sectional
n = 319 female regular runners;
aged between 16 and 60
Anxiety and self-esteem (Rosenberg
Self-esteem Scale and Zuckerman’s
Anxiety Adjective Checklist)
Investigating how levels of
anxiety and self-esteem related to
intensity of jogging
Female runners jogging with great intensity had significantly less anxiety than lower
intensities (x2 = 22.83; p < 0.001). Results indicate that intensity of jogging influences
self-esteem but was not significant: 89% of women scored in the range of high self-esteem,
and in the open-ended answers, 29% of responses stated that they feel better about
themselves, 12% had increased self-confidence and 6% stated a sense of accomplishment.
Goldfarb et al. [39]
(1984)
USA
Cross-sectional
n = 200 distance runners;
136 males and 64 females
Anorexia nervosa traits (Goldfarb Fear
of Fat scale and Activity Vector
Analysis)
Investigating anorexia nervosa
traits within distance runners
Runners had a mean score of 2.91 (on a 10-point scale), indicating a low–normal fear of fat,
and only 29 (14.5%) participants reported a high fear-of-fat score (score between 6 and 10).
Fear-of-fat scores did not correlate significantly with measures of running zealousness:
miles run per week (r = −0.04), number of workouts per week (r = 0.09), number of road
races (r = 0.05), marathons completed (r = −0.05) or degree of importance placed on running
(r = −0.03). Runners who demonstrated the greatest zealousness demonstrated Activity
Vector Analysis profiles that clustered around one particular profile (r = 0.64, p < 0.05)
indicating assertive, obsessive, perfectionistic and anxious individuals. Results do not
support a correlation between running and fear of fat; however, runners most closely
resembling “obligatory runners” exhibited traits characteristic of anorexia nervosa patients.
-Guyot et al. [40]
(1984)
USA
Controlled
cross-sectional
n = 126 participants; 64 runners
(44 males and 20 females) vs. 62
non-runners (37 males and
25 females)
Death anxiety (Death Concern Scale)
Comparing death anxiety in
runners vs. non-runners
Runners experienced more death thoughts (F(1,122) = 4.49, p < 0.05) but less death anxiety
(F(1,122) = 6.35, p < 0.05) than non-runners.
Rape [41]
(1987)
USA
Controlled
cross-sectional
n = 42 male participants; aged
18–25; 21 runners vs. 21
non-exercisers
Depression (Beck Depression
Inventory)
Comparing depression scores in
runners vs. non-exercisers
Runners were significantly less depressed (M = 4.38, SD = 3.88) than the non-exercisers
(M = 9.55, SD = 5.40) (t40 = 3.55, p < 0.001). Overall results suggest that running
reduces depression.
Weight et al. [42]
(1987)
South Africa
Controlled
cross-sectional
n = 135 female participants
consisting of marathon runners
(n = 85) vs. cross country runners
(n = 25) vs. non-running controls
(n = 25); aged 18–56
Eating attitudes and disorders (Eating
Attitudes Test and the Eating Disorder
Inventory)
Comparing eating attitudes and
disorders in marathon runners vs.
cross country runners vs.
non-running controls.
No significant differences were found between groups on any of the Eating Attitudes Test
scores (mean scores = 8.4, 14.3 and 11.8). Eating Disorder Inventory scores also did not follow
a definite pattern (mean scores for marathoners, cross country runners and non-running
controls were 24.8, 27.1 and 32.0, respectively), indicating that abnormal eating attitudes and
the incidence of anorexia was no more common among competitive female runners than
among the general population, with a low incidence of anorexia in the total group (2 out of
135 participants).
Chan et al. [43]
(1988)
USA
Cross-sectional
n = 60 runners who ran at least
3× per week for a minimum of a
year; 28 males and 32 females;
prevented runners n = 30 vs.
continuing runners n = 30;
aged 15–50.
Depression, self-esteem and mood
(Zung depression Scale, Rosenberg
Self-esteem Scale and Profile of Mood
States)
Comparing depression,
self-esteem and mood in
prevented runners vs. continuing
runners
Prevented runners reported significantly greater overall psychological distress (Wilks’s = 0.63,
p < 0.01: X92 = 24.38, p < 0.01), depression (F(1,58) = 11.57, p < 0.01) and overall mood
disturbance (F(1,58) = 11.03, p < 0.01) than continuing runners. Prevented runners reported
significantly lower self-esteem (F(1,58) = 3.17, p < 0.05), less satisfaction with the way their
bodies’ present looks (F(1,58) = 4.17, p < 0.05) and had greater desire to change something
about the way their bodies look (F(1,58) = 4.54, p < 0.05) compared to continuing runners.
Frazier [44]
(1988)
USA
Post only,
nonrandomised
long-term
observational
study
n = 86 regular, distance runners
who had all completed a
marathon; 68 males with mean
age of 33.7 and 18 females with a
mean age of 32.2
Mood (Profile of Mood States)
Investigating the relationship
between running and mood in
regular distance runners
Results suggest that regular, distance running improves mood in both males and females.
Running subjects had lower mean scores on tension, anger, depression, fatigue and confusion
and a higher mean of vigour compared to scores for test norms; however, statistical
significance was not reported here. The only significant difference between males and females
was on the confusion subscale: female mean = 7.8 vs. male = 5.5 (F(1,84) = 5.33, p < 0.05).
Lobstein, Ismail et
al. [45]
(1989)
USA
Controlled
cross-sectional
n = 36 male participants; aged
40–60; runners (n = 21) vs.
sedentary controls (n = 15)
Anxiety and depression (Minnesota
Multiphasic Personality Inventory and
Eysenck Personality Inventory)
Comparing anxiety and
depression levels in runners vs.
sedentary controls
Overall, running reduced anxiety (mean = 48.95 vs. 61.48 respectively, p < 0.05, standardised
canonical coefficient = −1.07) and depression (mean = 50.76 vs. 57.93, respectively, p < 0.05,
standardised canonical coefficient = 0.00) compared to being sedentary.
Lobstein,
Rasmussen et al.
[46]
(1989)
USA
Controlled
cross-sectional
n = 20 psychologically normal,
medically healthy men; aged
40–60; physically active joggers
(n = 10) vs. sedentary (n = 10)
Depression and stress (Eysenck
Personality Inventory and Minnesota
Multiphasic Personality Inventory)
Comparing depression and stress
in sedentary men to physically
active joggers
The findings suggest that regular jogging increases emotional stability (t = −2.84, p < 0.01) and
decreases subjective depression with MMPI subscales of depression and Wiggins depression
both being significantly lower in the joggers (t = 3.70, p < 0.01; t = 2.40, p < 0.05; respectively).
Int. J. Environ. Res. Public Health 2020, 17, 8059
8 of 39
Table 2. Cont.
Author
Year
Country
Design
Population
Mental Health Outcome
(Measurement)
Study Aim
Main Findings
Nouri et al. [47]
(1989)
USA
Cross-sectional
n = 100 male participants; aged
18–62; non-exercisers (n = 28),
drop-out joggers (n = 21),
beginning joggers (n = 15),
intermediate joggers (n = 16) and
20 advanced joggers (n = 20)
Anxiety and addiction/commitment
(State-Trait Anxiety Inventory,
Commitment to Running Scale,
and The Buss–Dutkee Inventory
measuring hostility and aggression)
Investigating the relationship
between various levels of jogging
vs. non-exercising on anxiety and
addiction/commitment
Running reduced anxiety levels compared to physical inactivity (F(4,89) = 4.43, p < 0.01),
with advanced joggers scoring significantly lower on trait anxiety than beginner and
intermediate joggers (1.42 vs. 1.77 vs. 1.69, respectively, p < 0.01) and commitment to running
significantly higher for the joggers than the non-exercisers (F(4,89) = 14.30, p < 0.01).
Chan et al. [48]
(1990)
Hong Kong
Cross-sectional
n = 44 male runners of track clubs
who ran a mean of 57.2 km per
week; mean age 27.8
Depression, stress, tension and
personality profiles (Chinese version of
the Personality Research Form)
Investigating the relationship
between running and depression,
stress, tension and personality
profiles
Running increased mood, happiness and outlook, while relieving anger, depression and
aggression, but none reported the size of changes or significance; 36.4% of participants
reported “improving mental health” as a reason to start running. Emotional benefits from
running reported were more self-confident (59.1% of respondents), happier (56.8%), better
mood (50.0%), relieved tension (45.5%), better self-image (36.4%), relieved depression (36.4%),
more aggression (36.4%), improved outlook (34.1%), more content (31.8%) and better family
relationship (15.9%). However, when participants stopped running, 38.6% experienced low
mood and 25.0% experienced anxiousness. More experienced runners, compared to less
experienced runners, were less aggressive or easily angered (t = 2.92, df = 42, p < 0.01),
less guarded or defensive (t = 2.13, df = 42, p < 0.005) and more likely to present themselves
favourably (t = 2.68, df = 35, p < 0.05).
Chapman et al.
[49]
(1990)
USA
Cross-sectional
n = 47 runners; 32 males aged
34–57 and 15 females aged 35
to 59
Running addiction, psychological
characteristics and running (Running
Addiction Scale, Commitment to
Running Scale, Symptom Checklist
(SCL-90-R) and Levenson’s Locus of
Control Scale)
Investigating the relationship
between running addiction,
psychological characteristics
and running
Results suggest a sex difference in the relationship between addiction and commitment,
in that commitment to running can occur without addiction in females. Running Addiction
Scale (RAS) scores correlated strongly for both sexes with self-rated addiction (p < 0.05) and
moderately with discomfort (p < 0.05). However, the Commitment to Running Scale (CR) did
not significantly correlate with self-rated addiction in females (0.246, ns) while RAS did
(0.753, p < 0.05) (z = 2.00, p < 0.05). Running addiction was associated with a high frequency
of running (p < 0.05) and longer duration of running (males = p < 0.05; females = ns). The CR
score correlated significantly with run frequency for male (0.59, p < 0.05) but not female
runners (0.14, ns), while CR and run duration did not correlate significantly for either sex
(males = 0.16, females = 0.28, n.s.). Duration of running was associated with mood
enhancement, implying that the benefits of running to mood may be obtained without
addiction. Males were above the norm for obsessive–compulsive tendencies (SCL-90 score)
and significantly higher than for females (p < 0.05), with running addiction associated with
male-positive personality characteristics (p < 0.05) but not with mood enhancement.
There were no significant correlations with personality traits for females.
Guyot [50]
(1991)
USA
Cross-sectional
n = 370 regular long-distance
runners; 289 males, mean age 38;
81 females, mean age 35
Addiction and death anxiety (Dickstein
Death Concern Scale and
author-created questionnaires for pain
running, running motives, risk taking
and medical symptoms)
Investigating the relationship
between addiction and death
anxiety between pain runners
and non-pain runners
Of the 370 runners, 56% pushed themselves during running until they felt pain. Compared to
non-pain runners, pain runners were more likely to be male, taller (F(1,366) = 11.45, p < 0.05),
heavier (F(1,366) = 9.19, p < 0.05) and younger (F(1,366) = 5.75, p < 0.05). Overall, results
suggest that runners classified as pain runners experienced significantly more death thoughts
(F(1,364) = 5.04, p < 0.05) and death anxiety (F(1,364) = 8.86, p < 0.05) than non-pain runners.
Maresh et al. [51]
(1991)
USA
Cross-sectional
n = 29 male distance runners;
mean age 40.1
Psychological characteristics including
anxiety, depression and stress
(Myers–Briggs Type Indicator Form
and Multidimensional
Anger-Inventory)
Investigating psychological
characteristics including anxiety,
depression and stress in
distance runners
Results suggest that long-term involvement in running is associated with low levels of
self-reported anxiety (m = 2.5 on a 6-point scale), depression (M = 1.8) and stress (m = 2.5).
Runners’ personality profiles differed from the normative sample, suggesting that running is
associated with more introverted personalities compared to men in the general population.
Compared to a normative sample of male control students, runners were less angry overall,
less frequently angry, and angry across fewer situations. However, 82% of runners reported
withdrawal symptoms when forced to be inactive, with a self-reported addiction average of
4.4 (“moderately” to “very”) on a 6-point scale.
Gleaves et al. [52]
(1992)
USA
Controlled
cross-sectional
n = 60 female participants;
runners (n = 20), bulimia patients
(n = 20) and a non-exercising,
non-dieting control group (n = 20)
Depression, body image and bulimia
nervosa symptomology (Beck’s
Depression Inventory, Body Image
Assessment Procedure, subscales from
the Eating Disorder Inventory,
Automatic thoughts Questionnaire and
dieting/weight loss questionnaire)
Comparing depression, body
image disturbance and bulimia
nervosa symptomology in
runners, bulimia patients and a
non-exercising, non-dieting
control group
No differences were found between runners and controls throughout the study. Bulimics had
significantly higher depression scores than runners and controls (20.65, 3.30 and 4.80
respectively, F = 56.95, p < 0.0001), but runners and controls did not differ from each other.
The same pattern of results was found for the Autonomic Thoughts Questionnaire (F = 45.87,
p < 0.0001) and Eating Disorder Inventory (F = 34.95, p < 0.0001), with bulimics scoring
higher, but no significant difference was found between runners and controls:
ATQ means = 85.40, 41.10 and 41.50, respectively, and EDI means = 12.80, 0.80 and 1.60,
respectively. There were significant group effects for all three variables of body image
(p < 0.01); again, bulimics differed from runners and controls.
Int. J. Environ. Res. Public Health 2020, 17, 8059
9 of 39
Table 2. Cont.
Author
Year
Country
Design
Population
Mental Health Outcome
(Measurement)
Study Aim
Main Findings
Coen et al. [53]
(1993)
USA
Cross-sectional
n = 142 male marathon runners;
mean age 44.07; obligatory
runners (n = 65) vs.
non-obligatory runners (n = 77)
Anxiety, anorexia and self-identity
(Obligatory Exercise Questionnaire,
State-Trait Personality Inventory and
The Ego Identity Scale)
Investigating the relationship
between obligatory running vs.
non-obligatory running on
anxiety, anorexia and self-identity
Compared to the non-obligatory runners, the obligatory group ran significantly more miles
per week,0 spent more time running each week (t(140) = 13.19, p < 0.001) and had significantly
higher levels of anxiety (18.85 vs. 6.45, respectively, (p < 0.01), suggesting that running
represents a successful coping mechanism to reduce anxiety. There was no statistically
significant difference in Ego Identity Scale score (p > 0.05), indicating that neither group
showed a higher developed sense of identity.
Furst et al. [54]
(1993)
USA
Controlled
cross-sectional
n = 188 participants, with n = 98
runners: 72 males and 26 females
vs. n = 90 gym exercisers: 60
males and 30 females; majority
aged 20–29
Negative addiction (Negative
Addiction Scale)
Comparing negative addiction in
runners vs. gym exercisers
A significant association was found between years of participation in physical activity and
addiction scores (F(5,182) = 6.39, p < 0.01) regardless of the type of activity, with no significant
differences in addiction scores between runners and gym exercisers.
Masters et al. [55]
(1993)
USA
Cross-sectional
n = 712 participants in a
marathon; 601 males and 111
females; aged 16–79
Self-esteem and psychological coping
of runners (Motivation of Marathoners
Scales, Sport Orientation
Questionnaire, Marlowe–Crowne
Social Desirability Scale, Attentional
Focusing Questionnaire, and 3 body
satisfaction and composition questions)
Investigating self-esteem and
psychological coping of marathon
runners
Participation in marathon running and training was used as a way to problem solve with
self-distraction for psychological coping (r(66) = 0.54, p < 0.001), improving self-esteem (r(66)
= 0.31, p < 0.01) and life meaning (r(66) = 0.36, p < 0.01). Marathon runners reporting higher
anxiety levels were more likely to endorse psychological motives for marathon running,
indicating that their running helps them avoid or dampen negative emotional experiences:
psychological coping (r(62) = 0.38, p < 0.01) and self-esteem (r(62) = 0.36, p < 0.01).
Women more strongly endorsed weight concern as a reason for involvement in marathons
(t(588) = −3.52, p < 0.001). Personal goal achievement and competition were both positively
related to training miles per week (r(575) = 0.22, p < 0.001 and r(576) = 0.30, p < 0.001,
respectively).
Pierce et al. [56]
(1993)
USA
Cross-sectional
n = 89 male runners; n = 33
non-competitive runners vs.
n = 24 5-km runners vs. n = 32
marathoner runners
Exercise dependence (negative
addiction scale)
Comparing exercise dependence
in recreational (non-competitive)
runners vs. 5-km runners vs.
marathoner runners
Training mileage was significantly correlated with exercise dependence, with marathoners
showing significantly higher (p < 0.05) mean exercise dependence scores (3.78) compared to
5K (2.9) and recreational runners (2.16). There was no significant difference in exercise
dependence scores found between recreational and 5K runners.
Klock et al. [57]
(1995)
USA
Controlled
cross-sectional
n = 22 females who were not
currently pregnant or taking oral
contraceptives; amenorrhoeic
runners (n = 7, mean age 28.0),
eumenorrheic runners (n = 9,
mean age 32.1) and eumenorrheic
sedentary women as controls
(n = 6, mean age = 27.5)
Depression, anorexia nervosa,
excessive exercise and eating disorder
(The modified Body Image
Questionnaire, the Beck Depression
Inventory, the Symptom Checklist-90
and the Eating Disorders Inventory)
Comparing depression, anorexia
nervosa, excessive exercise and
eating disorders in amenorrhoeic
runners, eumenorrheic runners
and eumenorrheic sedentary
women as controls
No significant differences were found between amenorrhoeic runners, eumenorrheic runners
and eumenorrheic sedentary controls on any of the psychological measures; hence,
these results do not suggest that there are psychological similarities between obligatory
runners and anorexics. However, there was a subgroup of amenorrhoeic runners (3 out of 9)
who scored in the clinically depressed range on the BDI, indicating that they were mild to
moderately depressed, and who also had the highest scores in their group on the Eating
Disorder Inventory measures.
Thornton et al.
[58]
(1995)
UK
Cross-sectional
n = 40 long-standing, habitual
male runners who ran on average
4 times per week with a weekly
mileage of 42.5 miles;
mean age 38
Addiction (Rudy and Estok Running
Addiction Scale, the Hailey and Bailey
Running Addiction Scale, and the
Personal Incentives for Exercise
questionnaire)
Investigating a relationship
between habitual running and
addiction
A high level of commitment in runners was found, with 55% classified as moderately
committed (scores 13–20) and 22% classified as highly “addicted” (scores +20), but no
relationship between years of running and addiction scales was found. This contrasts the
significant correlations between both the Estok RAS and frequency of running (rs = 0.38;
p < 0.05) and the Bailey RAS and number of runs per week (rs = 0.55; p < 0.01).
The correlation between the two addiction scales revealed a strong positive relationship
(rs = 0.81; p < 0.001). The primary motivation for running was mastery (mean Personal
Incentives for Exercise score of 4.2), followed by competition (3.93), weight regulation (3.9),
health benefits (3.89), fitness (3.87) and social recognition (3.01).
Powers et al. [59]
(1998)
USA
Controlled
cross-sectional
n = 57; habitual male runners
(n = 20), habitual female runners
(n = 20) and female anorexia
nervosa patients (n = 17)
Psychological profiles (Minnesota
Multiphasic Personality Inventory,
Leyton Obsessional Inventory, Obligate
Running Questionnaire, Becks
Depression Inventory and three body
image tests)
Comparing psychological profiles
of habitual male runners, habitual
female runners and female
anorexia nervosa patients
Significant differences in body image between groups (F = 7.969, p < 0.001) were found,
but no significant differences between female groups were found. Anorexics scored higher
than either group of runners (p < 0.001) for MMPI subscales of depression, hysteria and
psychopathic deviate, while none of the mean scores for either set of runners were considered
clinically significant. Anorexics scored higher on the Becks Depression Inventory than both
male and female runners (F = 68,645, p < 0.0001, mean scores = 23, 2.4 and 3.45, respectively),
but again, there was no significant differences between the runners. While there were
suggestive similarities between female runners and anorexics on body image, the overall
results found few psychological similarities between anorexia patients and habitual runners,
with evidence of significant psychopathology on all psychological measures in the anorexia
group, while both groups of runners were consistently in the normal range.
Int. J. Environ. Res. Public Health 2020, 17, 8059
10 of 39
Table 2. Cont.
Author
Year
Country
Design
Population
Mental Health Outcome
(Measurement)
Study Aim
Main Findings
Slay et al. [60]
(1998)
USA
Cross-sectional
n = 324 regular runners;
240 males and 84 females;
84 classified as obligatory
runners: 63 males and 21 females;
ages 15–71
Eating pathology traits (Eat Attitudes
Test, and Obligatory Running and
Motivations for Running
Questionnaire)
Comparing eating pathology
traits between obligatory and
non-obligatory runners
Obligatory runners, particularly females, are most at risk of eating pathophysiology,
as obligatory runners scored significantly higher on the EAT test, with female obligatory
runners having the highest mean EAT score (r = 0.40, p < 0.0002). At low levels of obligatory
running, women and men scored similarly on the EAT test (F(1,164) = 2.78, p > 0.05); however,
at higher levels of obligatory running, women demonstrated significantly higher EAT scores
than men (F(1,164) = 29.50, p < 0.001).
Ryujin et al. [61]
(1999)
USA
Controlled
cross-sectional
n = 55 female participants;
collegiate distance runners
(n = 20) vs. non-running
undergraduate student controls
(n = 35)
Eating disorder symptomology
(Eating Disorders Inventory 2)
Comparing eating disorder
symptomology in collegiate
distance runners to non-running
undergraduate student controls
Distance runners showed no enhanced symptomatology of eating disorders; instead,
female distance runners exhibited fewer symptoms of eating disorders on all subscales of the
Eating Disorder Inventory-2 except Perfectionism: drive for thinness (t(107) = 3.34, p < 0.005),
bulimia (t(107) = 2.48, p < 0.05) and body dissatisfaction (t(107) = 4.23, p < 0.001).
Leedy [62]
(2000)
USA
Controlled
cross-sectional
n = 276 runners with an average
of 11.5 years of running
experience; 239 men, mean age
37.9; 37 women, mean age 40.5
Depression and anxiety (Diagnostic
and Statistical Manual-IV, and an
author-adapted scale based on the
Running Addiction Scale)
Comparing depression and
anxiety in runners to non-runners
Of the non-runners and runners, 16.2% and 4.6%, respectively, had been diagnosed with an
anxiety disorder or prescribed an anxiolytic medication. These participants had significantly
higher anxiety trait scores than those without a diagnosis: F(1,274) = 18.87, p < 0.0001; 27% of
non-runners and 11.8% of runners reported a diagnosis of depression or were prescribed an
antidepressant. These participants had significantly higher measures of depression traits:
F(1,274) = 22.46, p < 0.0001. Women’s Stress Relief scores were significantly higher than men’s
(F(1,229) = 20.51, p < 0.001). Stress relief scores also varied across race length (F(2,229) = 6.47,
p < 0.005), indicating that runners entered in the 5–10K runs had lower scores than those
running the half/full marathon. Results indicate that highly committed runners (n = 31) had
significantly lower anxiety (F(2,113) = 5.73, p < 0.01) and depression scores (F(2,113) = 8.00,
p < 0.001) than recreational runners (n = 46) and non-runners (n = 39).
Edwards et al. [63]
(2005)
South Africa
Cross-sectional
n = 277 participants; 94 males and
183 females; mean age 25.2;
hockey players (n = 60), runners
(n = 40) and health club gym
members (n = 69) vs. a control
group of non-exercisers (n = 108)
Psychological well-being and physical
self-perception (Ryff’s Short
Standardized 18-item scale of Objective
Psychological Well-being, Fox’s
Physical Self-Perception Profile (PSPP)
and the Physical
Self-Perception Profile)
Comparing psychological
well-being and physical
self-perception in hockey players,
runners and health club gym
members vs. a control group of
non-exercisers.
All three forms of physical activity were associated with significantly higher (p < 0.01) scores
on 11 out of the 15 dimensions of psychological well-being and physical self-perception
scales compared to the control group: autonomy (F = 11.3), personal growth (F = 35.4),
environmental mastery (F = 9.6), purpose in life (F = 149.2), positive relations with others
(F = 81.6), self-acceptance (F = 50.4), sport competence (F = 41.3), conditioning (F = 28.1),
sport importance (F = 11.7), conditioning importance (F = 28.1) and body importance
(F = 31.0).
Schnohr et al. [64]
(2005)
Denmark
Observational
cohort study
n = 12,028 participants;
5479 males and 6549 females;
aged 20–79
Stress (An author-created
questionnaire)
Comparing stress levels between
jogging and various levels of
physical (in)activity in
leisure time
Those who were vigorously physically active (joggers) had the lowest level of stress
compared to those with low activity levels (males, 3.1% vs. 12.8%, respectively; female,
3.3% vs. 19.3%, respectively). With increasing physical activity in leisure time, there was a
decrease in level of stress between sedentary persons and joggers (Odds Ratio (OR) = 0.30)
and a decrease in life dissatisfaction between sedentary persons and joggers (OR = 0.30).
The highest levels of stress and dissatisfaction was seen in sedentary persons who remained
inactive at follow-up, while the group that changed from sedentary to active had an adjusted
OR < 0.50.
Strachan et al. [65]
(2005)
Canada
Prospective
longitudinal study
n = 67 regular runners; 32 were
male and 35 were female; mean
age of 40.6
Self-efficacy and self-identity
(author-created measures of task
self-efficacy and self-regulatory efficacy,
and a 10-item, validated athletic
identity measurement scale)
Investigating the relationship
between running and self-efficacy
and self-identity
Significant comparisons were made between extreme self-identity groups on social cognitive
and behavioural variables (F(5,37) = 4.72, p < 0.002), with those higher in self-identity scoring
higher on task self-efficacy (p < 0.001), scheduling self-efficacy (p < 0.03), running more
frequently (p < 0.001) and running for longer durations (p < 0.005) than those who scored
lowest on self-identity. Both scheduling self-efficacy (R2 change = 0.16, p < 0.001) and barriers
to self-efficacy (R2 change = 0.22, p < 0.001) were correlated with self-identity to prospectively
predict running frequency (F(2,64) = 9.98, p < 0.001; F(2,63) = 12.89, p < 0.001, respectively).
Both task self-efficacy (R2 change = 0.06, p < 0.05) and self-identity (R2 change = 0.06,
p < 0.04) were significant predictors of maintenance duration.
Int. J. Environ. Res. Public Health 2020, 17, 8059
11 of 39
Table 2. Cont.
Author
Year
Country
Design
Population
Mental Health Outcome
(Measurement)
Study Aim
Main Findings
Galper et al. [66]
(2006)
USA
Retrospective
cross-sectional
n = 6728 participants; 5451 males
with a mean age of 49.5 and 1277
females with a mean age of 48.1;
inactive (n = 1454 men and
n = 422 women), insufficiently
active (n = 1892 men and n = 443
women), sufficiently active
(n = 1396 men and n = 283
women) and highly active
(n = 709 men and n = 129 women)
Depression and emotional well-being
(Center for Epidemiological Studies
Scale for Depression and the General
Well-Being Schedule)
Assessing retrospectively if the
level of walking/running
impacted depression and
emotional well-being
Significant inverse association between increased physical activity and reduced depression
scores for both men (F(6, 5306) = 20.93, p < 0.0001) and women (F(6, 1247) = 11.80, p < 0.0001)
and a positive association between increased physical activity and increased well-being
scores in men (F(6, 5306) = 78.65, p < 0.0001) and women (F(6, 1247) = 24.82, p < 0.0001) were
found. These effects peaked at 11–19 miles per week (the sufficiently active category).
Luszcynska et al.
[67]
(2007)
UK
Longitudinal
prospective cohort
study
n = 139 runners; 111 males and 29
females; mean age of 29.5; strong
(n = 72) and weak (n = 66)
maintenance self-efficacy, strong
(n = 72) and weak (n = 61)
recovery self-efficacy, and strong
(n = 87) and weak (n = 45)
intentions
Self-efficacy and running behaviour
(an author-created questionnaire)
Investigating the relationship
between self-efficacy and running
behaviour with data collected
twice with a time gap of 2 years
Participants decreased the frequency of running sessions after 2 years, regardless of baseline
intensions or self-efficacy; however, those with stronger recovery in self-efficacy jogged more
than those with weaker recovery in self efficacy 2 years later. All participants reduced the
number of jogging or running sessions over 2 years (F(1,131) = 43.43, p < 0.001); however,
those with strong baseline recovery self-efficacy ran/jogged more often at 2 years than those
who had weak recovery self-efficacy at baseline (F(1,131) = 6.12, p < 0.05). Participants reduced
the number of running or jogging sessions over the 2 years, regardless of strong or weak
intentions at baseline (F(1,130) = 34.55, p < 0.001) or of strong or weak baseline maintenance of
self-efficacy (F(1,130) = 42.12, p < 0.001). No effects of maintenance self-efficacy were found.
Recovery self-efficacy at T1 predicted recovery self-efficacy (p < 0.05), maintenance
self-efficacy (p < 0.05), and jogging or running behaviour (p < 0.05) assessed 2 years later.
Overall, social-cognitive variables predicted behaviour, whereas behaviour did not predict
social-cognitive variables.
Smith et al. [68].
(2010)
UK
Cross-sectional
n = 93 non-competitive, regular
runners; 47 males and 46 females
Exercise dependence, running
addiction and social physique anxiety
(Exercise Dependence Scale, Running
Addiction Scale and Social Physique
Anxiety Scale)
Comparing exercise dependence,
running addiction and social
physique anxiety in male vs.
female runners
While a significant proportion of runners displayed symptoms of exercise dependence,
results did not find that exercise dependence was linked to social physique anxiety
(F(3.179) = 1.21, p > 0.05) or that there was a difference between men and women (p > 0.05 in
all cases). There was no significant difference between males and females for running
addiction scale (22.64 and 20.91, respectively), social physique anxiety scale (22.30 and 22.61,
respectively) or total exercise dependence scale scores (72.56 and 66.86, respectively).
Gapin et al. [69]
(2011)
USA
Cross-sectional
n = 179 regular runners; 88 males
and 91 females; 91 obligatory and
88 non-obligatory runners
Disordered eating (Eating Disorder
Inventory, Athletic Identity
Measurement Scale and Obligatory
Exercise Questionnaire)
Comparing disordered eating in
obligatory and non-obligatory
runners.
Obligated running (exercising to maintain identification with the running role) may be
associated with pathological eating and training practices, with obligatory runners scoring
significantly higher on all of the Eating Disorder Inventory measures (F(1,166) = 9.75,
p < 0.002),and the Athletic Identity Measure Scale (F(8,161) = 8.85, p < 0.001). Results also
indicated that runners in the obligatory group demonstrated greater concern with dieting,
preoccupation with weight and pursuit of thinness.
Wadas [70]
(2014)
USA
Cross-sectional
n = 68 male, high school cross
country runners; mean age 15.9
Disordered eating behaviours
(questionnaire consisting of The
Exercise Motivation Inventory 2,
the Eating Attitudes Test 26 and the
ATHLETE questionnaire)
Investigating any relationship
between male runners with
disordered eating behaviours and
eating attitudes
Risk factors associated with eating disorders within high school male cross-country runners
were found. Factors that had a significant relationship with disordered eating were weight
management (r = 0.31, p = 0.011), drive for thinness and performance (r = 0.36, p < 0.05),
and feelings about performance/performance perfectionism (r = 0.26, p < 0.05). No significant
relationships were found between disordered eating behaviours and personal body feelings
(r = 0.19, p = 0.109), feelings about eating (r = 0.18, p = 0.137), and feelings about being an
athlete (r = 0.12, p = 0.345); 4.41% (n = 3) of participants scored 20 or higher on the EAT-26,
indicating being at risk for disordered eating and displaying symptoms. An additional 13.2%
(n = 9) met the cutoff score of 14 for disordered eating behaviours.
Samson et al. [71]
(2015)
USA
Cross-sectional
n = 308 marathon runners;
177 males and 191 females;
mean age 41
Self-esteem and psychological coping
(Motivation for Marathons Scale,
The Perceived control questionnaire
and Sport Mental Toughness
Questionnaire)
Investigating the relationship
between self-esteem and
psychological coping with
marathon running
Self-esteem was positively associated with perceived control (r = 0.40) (x27 = 47.08, p < 0.001,
CFI = 0.85 and RMSEA = 0.14) but negatively associated with mental toughness. There was
also a positive relationship between perceived control and psychological coping (r = 0.42)
(x28 = 45.65, p < 0.001, CFI = 0.85 and RMSEA = 0.12). Results of the Motivations of
Marathoners Scales suggested than females were more likely to run to improve psychological
coping (4.8 and 4.42, respectively) and self-esteem (5.22 and 4.62, respectively) than men.
Int. J. Environ. Res. Public Health 2020, 17, 8059
12 of 39
Table 2. Cont.
Author
Year
Country
Design
Population
Mental Health Outcome
(Measurement)
Study Aim
Main Findings
Lucidi et al. [72]
(2016)
Italy
Cross-sectional
prospective field
study
n = 669 runners training for a
marathon; 569 males and 100
females; mean age 42.07
Stress (Perceived Stress Scale,
the Passion Scale and The Italian
version of the Locomotion and
Assessment Scales)
Investigating the relationship
between running and stress in
runners training for a marathon
Results suggest that running does not directly impact stress (β = −0.01; p = 0.75); however,
running increases harmonious passion (β = 0.37; p < 0.001), which in turn reduced athletes’
experience of stress. The indirect effect of running on anticipatory stress perception through
harmonious passion was statistically significant (αβ = −0.10; 95% confidence interval:
from −0.15 to −0.05). Similarly, the indirect effect of assessment on stress through obsessive
passion was statistically significant (αβ = 0.12; 95% confidence interval: from 0.07 to 0.17).
Results also indicated a significant direct effect of assessment on the athletes’ experience of
stress (β = 0.22; p < 0.001).
Batmyagmar et al.
[73]
(2019)
Austria
Prospective
longitudinal study
n = 99 participants; n = 50 elderly
marathon runners, mean age of
66, with 46 men and 4 women vs.
n = 49 non-exercising controls,
mean age of 66, with 44 men and
5 women
Self-reported health and well-being
and quality of life (Short Form Health
Survey-36)
Comparing self-reported health
and well-being,
and quality of life over 4 years in
elderly marathon runners to
non-exercising controls
Findings suggested that extensive high-intensity endurance exercise is linked with improved
subjective health and well-being in elderly persons, with athletes evaluating their health as
better than non-athletes in the following categories: general health perception (F = 14.21,
p < 0.001), vitality (F = 13.37, p < 0.001), social functioning (F = 11.30, p < 0.001), emotional
role functioning (F = 1.42, p < 0.002) and mental health (F = 6.07, p < 0.0016).
Cleland et al. [74]
(2019)
Australia
Cross-sectional
372 participants of “Parkrun”
events; mean age 43.8
Enjoyment, self-efficacy and factors of
participation in Parkrun event
(author-created questionnaires).
Investigating enjoyment,
self-efficacy and factors of
participation in Parkrun event
participants
Overall results suggested that perceived social benefits (B coefficient = 0.43) and self-efficacy
for Parkrun (B coefficient = 0.18) were positively associated with Parkrun participation.
Perceived benefits of Parkrun including enjoyment and social factors (B = 0.70) were
positively associated with participation, as was overall enjoyment (B = 0.30), self-efficacy for
Parkrun (B = 0.46), social support for Parkrun from family (absolute: B = 0.05) and social
support from friends (B = 0.04) related to Parkrun.
Lukacs et al. [75]
(2019)
Hungary
Cross-sectional
n = 257 amateur runners with at
least 2 years of running
experience; 131 males and 126
females; mean age 40.49
Exercise addiction and psychological
features (Exercise Dependence Scale;
a Cantril ladder for Overall life
satisfaction; SCOFF eating disorder
questionnaire; the UCLA 3-item
Loneliness Scale; Body Image Subscale
from Body Investment scale; and the
Depression, Anxiety and Stress
Scale-21).
Investigating the prevalence of
exercise addiction and
psychological features in amateur
runners, including perceived
health, life satisfaction, loneliness,
stress, anxiety, depression, body
shape and eating disorders
Respondents (137) were characterized as nondependent symptomatic, 97 were nondependent
asymptomatic and 23 were at risk of exercise addiction. Results found that five variables
significantly predicted the risk of exercise addiction in runners: weekly time spent running,
childhood physical activity, lower educational attainment, anxiety and loneliness (ranges of B
= 0.47 to 2.06, 95% CI for odds ratio = 1.61 to 7.86, p < 0.001 to p = 0.023).
Int. J. Environ. Res. Public Health 2020, 17, 8059
13 of 39
3.2.1. Runners Versus Non-Running Comparisons
Sixteen of the 47 studies directly compared measures of mental health in runners and non-running
comparisons [29,33,36,37,40–42,45–47,57,61–64,73]. They found that runners had lower depression and
anxiety [33,36,37,40,41,45–47,62], lower stress [64], higher psychological well-being [63,73], and better
mood [29] compared to sedentary controls. In these studies, there was no evidence of increased
prevalence of eating psychopathology in non-elite runners [42,57,61].
3.2.2. Runners Only
Nineteen studies only included runners [30,31,34,35,39,44,48,49,51,55,58,65–67,70,74–76] and
compared different levels and types of running. Some studies found a positive association with
higher self-identity runners and low levels of depression and high self-efficacy [30,65–67,74].
Studies investigating marathon training found a positive relationship of marathon training with
self-esteem and psychological coping [55,71]. Two questionnaires of long-distance runners found a
correlation between long-distance running and disordered eating behaviours, with obligatory runners
(obsessive runners who sacrificed commitments and relationships for running and suffered withdrawal
symptoms if they missed a run) exhibiting traits characteristic of anorexia nervosa patients [39] and risk
factors for eating disorders identified within male high school cross-country runners [70]. One study of
runners training for a marathon suggested that running did not directly impact stress [72]. There were
conflicting results from papers investigating negative addiction; one indicated that with more years
spent running came a greater risk of negative addiction [34], while another found no relationship
between years of running and addiction [58] and another found a sex difference in that commitment to
running can occur without addiction in female runners but not in males [49]. Another paper found
that five variables significantly predicted risk of exercise addiction in runners: weekly time spent
running, childhood PA, lower educational attainment, anxiety and loneliness [75]. The remaining
four cross-sectional studies of runners only found that, since participating in running, they had better
emotional well-being, relief of tension, self-image and self-confidence, mood, depression, aggression
and anger, anxiety and happiness, but not all reported significance or effect size [31,35,44,48,51].
A further eight studies compared groups of runners [32,38,50,53,56,60,68,69]. One paper found
that females jogging with greater intensity had significantly less anxiety than those jogging at lower
intensities [38]. The results from these studies showed that obligatory runners had significantly
higher anxiety [53] and eating disorder measures [60,69] than non-obligatory runners and that female
obligatory runners are most at risk of eating pathophysiology [60]. Non-elite marathoners showed
significantly higher exercise dependence scores [56] but had more self-sufficient personalities compared
to recreational runners who did not run marathons [32].
One paper did not find that exercise
dependence was linked to social physique anxiety [68], while another found that runners classified as
pain runners (pushed themselves until they felt pain) experienced significantly more death thoughts
and death anxiety than non-pain runners [50].
3.2.3. Runners Compared to Individuals with Eating Disorders
Two studies compared runners to individuals with diagnosed eating disorders but neither indicated
that habitual running led to development of disordered eating or body-image problems [52,59].
3.2.4. Prevented Runners
One study found that habitual runners prevented from running by illness or injury had significantly
greater overall psychological distress, depression and mood disturbance than continuing runners as
well as significantly lower self-esteem and body-image [43].
Int. J. Environ. Res. Public Health 2020, 17, 8059
14 of 39
3.2.5. Runners Compared to Gym Exercisers
A study comparing negative addiction in runners versus gym exercisers found significant
association between years of participation in running and gym exercise with negative addiction,
regardless of activity type [54].
3.2.6. Summary of Cross-Sectional Evidence
Consistent evidence was found for a positive association between positive mental health outcomes
and habitual or long-term recreational running compared to non-runners. In contrast, there was
evidence that high or extreme levels of running (high frequency and long distance including marathon
running) were associated with markers of running ill-health compared to levels of moderate running.
3.3. Category 2: Acute Bouts of Running
Narrative description of findings of the 35 studies with an acute bout of running are included
within Tables S2–S4 within the supplementary material.
3.3.1. Single Bouts
Twenty-three studies incorporated a design using a single bout of running to compare pre-post
measurements of mood and short-term measures of mental health (Table 3) [77–99]. Twenty-two of
these found positive improvement in measures of mental health (including anxiety, depression and
mood); however, one found a decrease in self-efficacy of children following participation in gymnasium
PACER (progressive aerobic cardiovascular endurance run) running challenge [95].
Eleven studies used a single bout of treadmill running, and all found positive pre-post differences in
mental health outcomes [84–86,88–93,97,99]. Results found significant reductions in state-trait anxiety;
total mood disturbance; and POMS subscales of anxiety, depression and confusion. A single bout
of treadmill running also significantly improved self-esteem; psychological well-being; children and
adolescent self-efficacy; state anxiety, depression and totally mood disturbance; adult self-efficacy;
and general affective response. One study found that mood improvements were not evident until
40 min of running [88], while another found that depressed individuals participating in a treadmill run
with increasing gradient improved depressed mood immediately post-run but that depressed mood
increased at 30-min postexercise [93].
Three studies used a single bout of track running and found significant decreases in anxiety [78,87]
and total mood disturbance [81]. Two studies found that a single outdoor run significantly improved
depression scores and that even a 10-min jog caused significant mood enhancement [80,94]. Two studies
found that a single bout of self-paced running significantly reduced all but one of the POMS subscales
and had significant positive changes in all measures of states of affect [82,96].
There were significant improvements for self-esteem, stress and total mood disturbance following
a 5-km Parkrun [98], while a 3-mile “fun-run” increased positive mood and decreased negative
mood [83]. Two studies used longer runs as exposures: one found that a 1-h run significantly reduced
anxiety and nonsignificantly reduced depression [79], while the other found that a 12.5-mile jog
significantly improved pleasantness; decreased trait anxiety; nonsignificantly increased activation; and
reduced state-anxiety, sadness, anxiety, depression and relaxation subscales [77].
Int. J. Environ. Res. Public Health 2020, 17, 8059
15 of 39
Table 3. Summary of data extraction from the 23 single-bout studies.
Author
Year
Country
Design
Population
Mental Health Outcome
(Measurement)
Study Aim
Main Findings
Nowlis et al. [77]
(1979)
Canada
Pre-post non-controlled
study
n = 18, experienced joggers; 5 females
and 13 males; age range 17 to 55
Mood and anxiety (Mood
Adjective Checklist and State
Trait Anxiety Inventory)
Impact of a 12.5-mile jog on mood
and anxiety
Significant improvement in measures of pleasantness (2.00 to 2.67, p < 0.01);
a significant decrease in trait anxiety (34.81 to 33.31, p < 0.10); a nonsignificant
increase in activation; a reduction in state-anxiety; and a reduction of sadness,
anxiety, depression and relaxation subscales
Wilson et al. [78]
(1981)
Canada
Pre-post controlled study
n = 42; 20 runners, 12 aerobics class
exercisers vs. 10 people having lunch;
23 females and 19 males; age range 21
to 28
Anxiety (State-Trait Anxiety
Inventory)
Impact of a solo indoor track run
on anxiety
Significant decrease in anxiety post-activity (F(1,39) = 15.63, p < 0.003)
Markoff et al. [79]
(1982)
Hawaii
Pre-post non-controlled
n = 15, all had run at least 1 marathon;
11 males and 4 females; aged 23–45
Mood (Profile of Mood States)
Impact of a 1-h run on mood
Significant reduction in anxiety (t = 2.72, p < 0.01) and a nonsignificant
reduction in depression (t = 1.80, n.s.)
Thaxton et al. [80]
(1982)
USA
Non-randomised
controlled trial
n = 33, regular runners; 24 males and 9
females; mean age 36; 4 groups: outdoor
running test (n = 6), pre-test but no
running test (n = 9), no pre-test but
running test (n = 11), and no pre-test and
no running test (n = 7).
Mood (Profile of Mood States)
Impact of 30 min outdoor
running on mood
Significant differences in the depression scores between the 30 min outdoor
running group and abstaining groups (F(1,29) = 4.8, p < 0.05) but no significant
differences between anxiety, vigour and fatigue scores
McGowan et al.
[81]
(1991)
USA
Non-randomised
controlled trial
n = 72, college students; 25 joggers vs. 11
karate vs. 26 weight lifters vs. 10 science
lecture class members
Mood (Profile of Mood States)
Impact of 75 min of jogging on an
outdoor track on mood
Significant decrease in total mood disturbance from pre- (35.68) to post- (24.16)
test (t24 = 2.84, p < 0.009) following 75 min of jogging on a track
Goode et al. [82]
(1993)
USA
Pre-post non-controlled
n = 150, regular runners; 104 males,
36 females; mean age 31.7
Mood (Profile of Mood States)
Impact of own training for
running on mood
Significant alterations in all but one (vigor) of the POMS scales, with a
significant reduction post-run in tension/anxiety (mean change of −3.1, p < 0.1),
depression (mean change of −1.5, p < 0.1), confusion (mean change of −1.1,
p < 0.1) and anger (mean change of −1.8, p < 0.1) and a significant increase in
fatigue post-run (mean change of +1.8, p < 0.1).
Morris et al. [83]
(1994)
UK
Pre-post non-controlled
n = 165, members of a road runners club;
98 males and 67 females; mean age 34
Mood (author-devised adjective
checklist based on POMS)
Impact of a 3 mile “fun-run” on
mood
Increase in positive mood (F(1,163) = 68.18, p < 0.001), decrease in negative mood
(F(1.163) = 47.62, p < 0.001) and greater improvements in mood in women than in
men that was not significant (p > 0.1).
Rudolph et al. [84]
(1996)
USA
Randomised
non-controlled trial
n = 36, moderately active female
university students; n = 12 for 10-, 15-
and 20-min interventions; mean age 20.6
Self-efficacy (Exercise-Efficacy
Scale)
Impact of various timings of
treadmill running on self-efficacy
(10, 15 and 20 min)
Significant increase in mean scores of self-efficacy in all 3 groups, from pre to
postexercise (F(1, 33) = 74.57, p < 0.001), and moderate effect sizes in the 15-
(ES = 0.36) and 10- (ES = 0.49) minute conditions although the largest effect size
(ES) occurred in the 20-min condition (ES = 0.68)
Cox et al. [85]
(2001)
USA
Randomised controlled
trial
n = 24, physically active male university
students; mean age of 28.3
Psychological affect and
well-being (Subjective Exercise
Experiences Scale)
Impact of 30 min of treadmill
jogging at either 50% or 75%
predicted VO2 max on
psychological affect and
well-being
Significant improvement in psychological well-being following an acute bout of
aerobic exercise (p = 0.037, η2p = 0.07)
O’Halloran et al.
[86]
(2002)
Australia
Pre-post non-controlled
n = 50, regular runners; 25 males and 25
females; mean age 26.6
Mood (Profile of Mood States and
Beliefs Concerning Mood
Improvements Associated with
Running Scale)
Impact of a 60-min treadmill run
on mood
Significant reductions in anxiety (composed-anxious POMS scale = 25.6 to 29.12,
p < 0.05, i.e., more composed-less anxious) and depression (elated-depressed
POMS scale = 24.56 to 27.10, p < 0.01, i.e., more elated-less depressed)
Szabo et al. [87]
(2003)
UK
Pre-post non-controlled
time series
quasi-experimental
n = 39, sports science university students;
22 males and 17 females; aged 20–23
Anxiety, positive well-being and
psychological distress
(Spielberger State Anxiety
Inventory and Exercise induced
Feeling Inventory)
Impact of 20 min of track running
on anxiety and feelings
Significant reduction in state anxiety (F(1.5, 58.3) = 5.32, p < 0.01) and a positive
effect on psychological distress and positive well-being
O’Halloran et al.
[88]
(2004)
Australia
Randomised controlled
trial
n = 160 regular runners; 80 did run vs. 80
no running; 80 males and 80 females;
aged 18–40
Mood (Profile of Mood States and
Beliefs Concerning Mood
Improvements Associated with
Running Scale).
Impact of a 60-min treadmill run
on mood
Improvements in composure, energy, elation and mental clarity during the run;
in the energetic-tired subscale, evident improvements at 25 min (F(1,156) = 10.09,
p = 0.002); in the rest, non-evident mood improvements until 40 min of running;
and more composure (less anxious) (F(1,156) = 9.47, p = 0.002) and more clear
headedness (less confused) (F(1,156) = 5.57, p = 0.02) in runners
Robbins et al. [89]
(2004)
USA
Pre-post non-controlled
n = 168, inactive children and
adolescents; 86 males and 82 females;
mean age 12.6
Self-esteem using the Walking
Efficacy Scale
Impact of a 20-min treadmill run
on self-efficacy in children and
adolescents
Significant increase in children and adolescents’ self-efficacy postexercise
(F(1, 158) = 84.31, p < 0.001) but significantly lower pre-activity self-efficacy in
African American girls reported than the other three race-gender groups (F(3,164)
= 5.55, p < 0.01)
Int. J. Environ. Res. Public Health 2020, 17, 8059
16 of 39
Table 3. Cont.
Author
Year
Country
Design
Population
Mental Health Outcome
(Measurement)
Study Aim
Main Findings
Pretty et al. [90]
(2005)
UK
Randomised controlled
trial
n = 100; 45 males and 55 females; mean
age 24.6
Mood and self-esteem (Profile of
Mood States and Rosenberg
Self-Esteem Questionnaire)
Impact of a 20-min treadmill run
with rural vs. urban stimuli on
mood and self-esteem
Significant increase in self-esteem (from 19.4 to 18.1 on the Rosenberg
Self-Esteem Questionnaire, p < 0.001), with rural and urban pleasant stimuli
producing a significantly greater positive effect on self-esteem than exercise
alone, while both rural and urban unpleasant scenes reduced the positive effects
of exercise on self-esteem
Hoffman et al. [91]
(2008)
USA
Pre-post pre-experimental
study
n = 32; 16 regular exercisers and 16
non-exercisers; 8 males and 8 females in
each group
Mood (Profile of Mood States)
Impact of a 30-min treadmill run
on mood
Decreased total mood disturbance in a 30-min treadmill run in both regular
exercisers (−16 points, 95% CI = 7–24) and non-exercisers (−9 points, 95%
CI = 1–18) but almost double the effect in exercisers.
Kwan et al. [92]
(2010)
USA
Pre-post non-controlled
n = 129; 62 males and 67 females; mean
age 22
General affective response
(Physical Activity Affect scale)
Impact of a 30-min treadmill run
on general affective response
Positive effect of the run on general affective response during exercise (b = 0.52,
SE = 0.09, p < 0.0001) and 15 min postexercise (b = 0.73, CI.95 = 0.56, 0.89,
t(126) = 8.63, p < 0.0001).
Weinstein et al.
[93]
(2010)
USA
Pre-post controlled study
n = 30; 15 males and 15 females; 2 with
minor depressive disorder, 12 with major
depressive disorder and 16 as controls;
mean age 39.8
Mood and depression (Becks
depression Inventory scale and
Profile of Mood States)
Impact of 25 min of increasing
graded treadmill running on
mood and depression
Not only improvements in depressed mood immediately following exercise
(p = 0.02) of 25 min of increasing graded treadmill running but also increased
depressed mood at 30 min postexercise (F(1,27) = 3.98; p = 0.05; ηp2 = 0.13) and
significant relation between the severity of depression and increases in
depressed mood (r = 0.60, p = 0.001) at 30 min postexercise
Anderson et al.
[94]
(2011)
UK
Randomised controlled
trial 2 × 2 mixed design
n = 40, from various sports clubs; aged
18–25
Mood (“Incredibly Short Profile
of Mood States”)
Impact of a light 10-min outdoor
jog on mood
Significant mood enhancement even with a light 10=min jog on a grass playing
field (F(1,38) = 24.18, p < 0.001, n2p = 0.39) compared with a 10-min cognitive task
Kane et al. [95]
(2013)
USA
Pre-post non-controlled
n = 34 school children; 16 males and 18
females; aged 11–14
Self-efficacy (Self-efficacy
questionnaire adapted for
children)
Impact of the running PACER
challenge (20 m sprints with
increasing pace inside a
gymnasium) on self-efficacy
in children
Decrease in self-efficacy following participation in the run (from 2.7 to 2.3
following exercise, t = 4.6, p < 0.001, large effect size of d = 0.79) but positive
correlation between PACER laps and pre- and post-measures of exercise
self-efficacy
Szabo et al. [96]
(2013)
Hungary
Pre-post non-controlled
n = 50 recreational runners; 37 males and
13 females mean age 29.02
States of affect using the Exercise
Induced Feeling Inventory
Impact of a 5 km self-paced run
along a public running path on
states of affect
Significant positive changes in all 4 measures of states of affect following a 5-km
self-paced run: revitalisation (F(1,48) = 145.93, p < 0.001, partial n2 = 0.75,
ES = 2.0), positive engagement (F(1,48) = 97.11, p < 0.001, partial n2 = 0.67,
ES = 1.6), tranquillity (F(1,48) = 85.02, p < 0.001, partial n2 = 0.64, ES = 1.5) and
exhaustion (F(1,48) = 32.25, p < 0.001, partial n2 = 0.40, ES = 1.0)
McDowell et al.
[97]
(2016)
Ireland
Randomised controlled
trial
n = 53; 27 males and 26 females;
mean age of 21.2
Mood and anxiety (Profile and
Mood States and State-Trait
Anxiety Inventory)
Impact of a 30-min treadmill run
on mood and anxiety
Significantly improved state anxiety (F1,92 = 12.52, p < 0.001), feelings of
depression (F1,86 = 5.05, p < 0.027) and total mood disturbance (F = 36.91,
p < 0.001) compared to 30 min of seated quiet rest
Rogerson et al.
[98]
(2016)
UK
Pre-post non-controlled
mixed between-within
n = 331 Parkrun attendees; 180 males and
151 females; mean age 40.8
Psychological well-being
(Questionnaire containing parts
of the Profile of Mood States,
Rosenberg Self-esteem scale and
Perceived Stress Scale)
Impact of a 5-km park run on
psychological well-being
Significant (p < 0.001) improvements post-run for self-esteem (7.7%
improvement; F(1, 324) = 100.58, η2 = 0.24), stress (18.4% improvement;
F(1, 315) = 50.78, η2 p = 0.139) and total mood disturbance (14.2% improvement;
F(1, 278) = 22.15, η2p = 0.07)
Edwards et al. [99]
(2017)
USA
Randomised controlled
trial
n = 27; 8 joggers vs. 9 walkers vs. 10
stretchers; aged 18–35
Stress and anxiety
(Exercise-Induced Feeling
Inventory and Affective
Circumplex Scale, and the
Strait-Trait Anxiety Inventory)
Impact of a 15-min treadmill jog
on stress and anxiety
Emotionally protective effect from a 15-min bout of treadmill jogging (n = 8)
compared to an equivalent amount of time walking (n = 9) or stretching (n = 10)
after exposure to a film clip intended to elicit a negative emotional response,
with reduced anxiousness from baseline to post-jog (28.8 to 13.1, p = 0.06) on the
State-Trait Anxiety Inventory, and increased anger score from baseline to
post-film clip in the stretching group (1.2 to 26.0, p = 0.048) unlike the walking
(p = 0.11) and jogging (11.3 to 9.4, p = 0.19) groups
Int. J. Environ. Res. Public Health 2020, 17, 8059
17 of 39
3.3.2. Double Bouts
There were nine studies that had a double-bout design [100–108] (Table 4). Eight of the nine
studies were primarily designed to compare conditions rather than to compare the impact of running
on mental health, including green/park versus urban, solo versus group, different pacing and different
durations of running [101–108]. Seven of the eight studies found that markers of mental health
improved significantly after running [101–107]. Only one study was designed to primarily assess the
impact of running on mental health, and although there was no control, they found higher mood and
feelings of pleasantness post-run but these “did not reach significance” [100].
Four studies compared park/rural versus urban running, and all found measures of mental health
including anxiety, depression, mood and self-esteem improved post-run [103–105,107]. No paper
reported a statistically significant difference in emotional benefit between park and urban conditions.
Two studies compared solo versus group running: one found that anxiety reduced following both group
and solo running [101], while the other found that children’s anxiety levels increased nonsignificantly
following individual and group running [108]. One study compared 10- and 15-min runs and found
that they produced similar psychological benefits to mood [102]. Another compared a self-paced
versus prescribed-paced run and found higher self-efficacy before the prescribed-paced run compared
to the self-paced run [106].
3.3.3. Triple Bouts
Three studies used three bouts of running (Table 5) [109–111]. One study found that, while two
indoor runs had a positive effect on mood, the outdoor run had an even greater benefit to mood
with subjects feeling less anxious, depressed, hostile and fatigued and feeling more invigorated [109].
Another study also used 3 runs of varying intensities and found significant overall mood benefits
postexercise but no significant differences between intensities [110]. One study compared 3 intensities
of treadmill exercise to a sedentary control condition and found that state anxiety improved following
running at 5% below and at the lactate threshold but that anxiety increased after running at 5% above
the lactate threshold [111]. Overall, these studies suggest that running improves mood, that outdoor
running has a greater benefit to mood and that most intensities of running improve mood, with the
exception of an intensity markedly above the lactate threshold. However, only one study included a
control condition [111].
3.3.4. Summary of Acute Bouts
Overall, these studies suggest that acute bouts of running can improve mental health and that the
type of running can lead to differential effects. The evidence suggests that acute bouts of treadmill,
track, outdoor and social running (2.5–20 km and 10–60 min) all result in improved mental health
outcomes. There were few differences between high and low intensities. Studies consistently show that
any running improves acute/short-term mood markers, but the lack of inactive comparison conditions
is a limitation to the strength of the evidence. Little variation in the demographics of participants and
small sample sizes limit generalizability and precision of findings.
Int. J. Environ. Res. Public Health 2020, 17, 8059
18 of 39
Table 4. Summary of data extraction from the 9 double-bout studies.
Author
Year
Country
Design
Population
Mental health Outcome
(Measurement)
Study Aim
Main Findings
Krotee [108]
(1980)
USA
Pre-post pre-experimental
non-controlled
n = 78, children aged 7–12.
Anxiety (State-Trait Inventory for
Children)
Impact of 50-metre group vs.
solo run on anxiety
Children’s anxiety levels increased nonsignificantly following a run in
either an individual (30.54 to 32.72, n.s.) or a group setting (30.67 to 31.83,
n.s.).
Wildmann et al.
[100]
(1986)
Germany
Pre-post non-controlled
n = 21, male long-distance
runners; mean age of 29.8
Feelings of pleasantness and
changes of mood
(Eigenschaftsworterliste scale
adjective checklist)
Impact of 2 identical 10-km
runs (1 week apart) on
feelings of pleasantness and
change of mood
Higher scores of good mood and feelings of pleasantness were found
following the runs (mean increase of the two runs for all subjects was 2.79
from a total of 19 items, but the increase did not reach significance).
O’Connor et al.
[101]
(1991)
USA
Pre-post non-controlled
n = 17, members of local
running clubs; 10 males
and 7 females;
mean age 25
Anxiety and body awareness
(State-Trait Anxiety Inventory
and Body Awareness Scale)
Impact of a 5-mile outdoor
group vs. solo run on anxiety
Anxiety levels were reduced following both a group (mean baseline = 34.0
vs. pre-exercise = 42.5 vs. postexercise = 27.5, p < 0.05) and solo run (mean
baseline = 34.0 vs. pre-exercise = 40.0 vs. postexercise = 30.0, p < 0.05).
Nabetani et al.
[102]
(2001)
Japan
Pre-post non-controlled
n = 15, healthy,
moderately active male
graduate students
Mood (Mood Checklist
Short-form 1)
Impact of a 10-min vs. a
15-min treadmill run on mood
Following the 10-min trial, anxiety significantly decreased (ES = 0.61,
p < 0.01), whilst there was no significant difference in pleasantness
(ES = 0.86) and relaxation (ES = 0.33). Following the 15-min trial, anxiety
(ES = 0.51) and pleasantness (ES = 0.62) significantly decreased (p < 0.01),
but relaxation (ES = 0.07) had no significant pre-post difference.
Bodin et al. [103]
(2003)
Sweden
Pre-post non-controlled
within-subjects
n = 12, regular runners. 6
male and 6 female. Mean
age of 39.7.
Emotional restoration ie.
depression/anxiety
(Exercise-Induced Feeling
Inventory and the Negative
Mood Scale).
Impact of 1 h park vs. urban
run on depression and
anxiety.
Runners preferred the park to the urban environment and perceived it as
more psychologically restorative; there was no statistical difference in
results for park vs. urban settings, with running in both settings causing a
significant decline in anxiety/depression (F(1,10) = 16.2, p < 0.002, r = 0.78,
effect size = 0.30).
Butryn et al. [104]
(2003)
USA
Pre-post non-controlled
within-subjects
n = 30, non-elite female
distance runners; mean
age 31
Mood, feeling states and
cognition states (Profile of Mood
States, Exercise-Induced Feeling
Inventory and Thoughts During
Running Scale)
Impact of a 4-mile park vs.
urban run on mood
Total mood disturbance scores decreased by 8.97 (p < 0.001), with a similar
effect following the urban run: total mood disturbance scores decreased by
9.13 (p < 0.001).
Kerr et al. [105]
(2006)
Japan
Pre-post non-controlled
n = 22, recreational
runners; mean age 22.7
Stress and emotions (Tension and
Effort Stress Inventory)
Impact of indoor vs. outdoor
5-km run on stress and
emotions
Significant pre-post effects irrespective of running condition were found,
with an increase in relaxation (F(1, 21) = 5.60, p < 0.05) and excitement
(F(1, 21) = 24.65, p < 0.001) and a decrease in anxiety (F(1, 21) = 9.90, p < 0.01).
Rose et al. [106]
(2012)
New Zealand
Pre-post controlled
n = 32, all females; 17
sedentary and 15 active;
mean age 45
Self-efficacy (Self-Efficacy for
Exercise Scale)
Impact of self-paced vs.
prescribed pace 30-min
treadmill run on self-efficacy
Higher self-efficacy was observed before the prescribed paced run
compared to the self-paced run (F1,28 = 5.81; p < 0.023; n2 = 0.17).
Reed et al. [107]
(2013)
UK
Pre-post non-controlled
n = 75, children aged 11
and 12
Self-esteem (Rosenberg Self
Esteem Scale)
Impact of rural vs. urban
1.5-mile run on self-esteem
Significant increase in self-esteem (F(1,74) = 12.2, p < 0.001) was found,
but no significant difference between the urban or green exercise condition
(F(1,74) = 0.13, p = 0.72) or any significant difference between boys and girls
were found.
Int. J. Environ. Res. Public Health 2020, 17, 8059
19 of 39
Table 5. Summary of data extraction from the 3 triple-bout studies.
Author
Year
Country
Design
Population
Mental Health Outcome
(Measurement)
Study Aim
Main Findings
Harte et al. [109]
(1995)
Australia
Pre-post non-randomised
controlled-repeated
measure design
n = 10, male amateur triathletes or
marathon runners with a mean age
of 27.1
Mood (Profile of Mood
States)
Impact of a 12-km outdoor run vs.
indoor treadmill run with
external stimuli vs. an indoor
treadmill run with internal
stimuli on mood
While the two indoor runs had a positive effect on
mood, outdoor running had an even greater benefit to
mood with subjects less anxious (F(3,35) = 14.12,
p < 0.005), less depressed (F(3,35) = 4.16, p < 0.01),
less hostile (F(3,35) = 13.13, p < 0.005), less fatigued
(F(3,35) = 15.09, p < 0.005) and more invigorated
F(3,35) = 13.01, p < 0.005).
Berger, Owen +
Motl [110]
(1998)
USA
Pre-post non- controlled
study
Study 1: n = 71 college students
(32 males and 39 females) with a mean
age of 21.39; study 2: n = 68 college
students (28 males and 40 females)
with a mean age of 22.22
Mood (Profile of Mood
States)
Impact of three 15-min runs of
varying intensities (50, 65 or 80%
age-adjusted HR max) on mood
Significant overall mood benefits postexercise
(F(6.57) = 6.43, p < 0.0001) for all intensities but no
significant differences between intensities were found.
Markowitz et al.
[111]
(2010)
USA
Pre-post controlled trial
n = 28, college-aged students; 14 active
vs. 14 sedentary controls; mean age 21
Anxiety (State-Trait
Anxiety Inventory)
Impact of three 20-min treadmill
runs of varying intensities (5%
below, 5% above and directly at
lactate threshold) on anxiety
This was the only triple-bout study with a sedentary
control condition. State anxiety improved postexercise
at 5% below (effect size = −0.38, p < 0.001) and after
exercise at the lactate threshold (effect size = −0.20,
p < 0.001), but anxiety increased at 5% above the lactate
threshold (effect size = +0.13, p = 0.0030).
Int. J. Environ. Res. Public Health 2020, 17, 8059
20 of 39
3.4. Category 3: Longer-Term Interventions
Thirty-four studies investigated the effects of more than three bouts of running on measures
of mental health ranging from 2-week interventions to 1-year marathon training programmes
(Table 6) [112–144].
Narrative description of 34 studies are available in Table S5 within the
supplementary material.
Eight studies used 2–8 week running interventions [121,122,125,127,128,132,137,139]. Male regular
runners deprived of running for 2 weeks had increased anxiety and depression symptoms compared
to continuing runners [125].
Two 3-week interventions both found that mood improved while
amateur runners had lesser anxiety on running days compared to non-running days; perceived stress
in adolescents did not significantly change [132,137]. A 4-week intervention of regular treadmill
running at set paces in moderately trained male runners found that an increase in intensity of runs
was associated with significant increase in total mood disturbance while running at a pace with
more economical values was associated with more positive mental health profiles [127]. A 7-week
non-controlled intervention of weekly 40-min fixed distance outdoor rural runs increased mood in
both male and female regular exercising university students, with faster runners scoring higher than
slower runners [128]. An 8-week intervention of a combination of weekly group and solo jogging in
middle-aged chronically stressed, sedentary women found lower anxiety and greater self-efficacy than
baseline and compared to relaxation group controls [121]. Two studies used a 8-week intervention
of walking/running with non-treatment controls and found significant improvements in mood and
decrease of depression, including in outpatients diagnosed with mild to severe depression [122,139].
Eleven studies used 10–20 week running interventions [114–116,119,123,126,129–131,140,143].
Three 10-week walking/jogging interventions found reductions in anxiety measures, improvement of
well-being and conflicting results for changes in depression measures compared to controls [115,119,129].
Another 10-week running intervention found that depression, trait anxiety and state anxiety all decreased
significantly while mood improved significantly [114]. A further 10-week running intervention found
that, although the exercise group was more likely to use exercise to cope with stress, there were
no significant differences in stress or coping measurements between the running and comparison
group [123]. Three 12-week interventions found significantly reduced stress and improvements in mood
in college students compared to controls, with more mood improvement in males and in females with
higher masculinity [126,130,143]. One 12-week intervention of self-directed running in recreational
runners found that well-being was significantly higher during weeks when individuals ran further and
ran more often while self-efficacy was related to distance ran but not to frequency of running [143].
Running interventions of 14–20 weeks improved mood and self-esteem and lowered emotional stress
reactivity in college/university students compared with controls [116,131,140].
Int. J. Environ. Res. Public Health 2020, 17, 8059
21 of 39
Table 6. Summary of data extraction from the 34 longer-term intervention studies.
Author
Year
Country
Design
Population
Mental Health Outcome
(Measurement)
Study Aim
Main Findings
Lion [112]
(1978)
USA
Randomised
controlled trial
n = 6, chronic psychiatric patients;
2 males and 4 females; 3 had the
running intervention and 3 were
controls; middle aged
Anxiety and body image
(State-Trait Anxiety Inventory
and Rorschach Inkblot Test for
body-boundary image)
Impact of running a mile 3 times per
week for 2 months on anxiety and
body image in chronic psychiatric
patients
Post-test anxiety was significantly reduced in the jogging group vs. control
group (t = 3.2, df = 4, p < 0.05). Joggers showed an average drop of 9
points on the STAI (39.3 to 30.3) from pre- to post-test, while controls
showed an average rise of 4 points (32.6 to 36.6). No statistical difference
was found between groups for post-test body image scores.
Blue [113]
(1979)
USA
Pre-post
non-controlled
n = 2 former inpatients of a
psychiatric hospital; 1 male aged
37 and 1 female aged 32
Depression (Zung depression
scale)
Impact of 3 runs per week for 9
weeks on depression
Following running intervention, both patients’ depression scores reduced
from “moderately depressed” to “mildly depressed” (decrease of 18 and
15 points on the Zung Depression Scale).
Young [114]
(1979)
USA
Pre-post
non-controlled
n = 32 adults; 4 groups: young
males (n = 8, mean age 30.13),
middle-aged males (n = 8, mean
age 53.0), young females (n = 8,
mean age 28.25) and middle aged
females (n = 8, mean age 50.25)
Anxiety and depression using the
Multiple Affect Adjective
Checklist
Impact of a 10-week walking/jogging
programme consisting of 1 h 3× per
week on anxiety and depression
Significant reductions in pre- to post-test anxiety (F(1,28) = 6.01, p < 0.05)
were found. Results for anxiety and depression both showed significant
age differences in favour of older subjects ((F(1,28) = 5.37, p < 0.05) and
(F(1,28) = 5.21, p < 0.05), respectively). However, there was no significant
improvement with subject depression scores.
Blumenthal et al.
[115]
(1982)
USA
Non-randomised
controlled cohort
n = 16 healthy adults; 5 males and
11 females; mean age 45.1
Anxiety and mood (Profile of
Mood States and the State-Trait
Anxiety Inventory)
Impact of 3 times weekly
walking-jogging programme for 10
weeks vs. 10 weeks of sedentary
controls on anxiety and mood
The exercise group exhibited less tension (F(1,30) = 4.49, p < 0.04),
depression (F(1,15) = 4.82, p < 0.04), fatigue (F(1,30) = 3.88, p < 0.05) and
confusion (F(1,15) = 4.40, p < 0.05) but more vigour (F(1,15) = 3.28, p < 0.09)
than sedentary controls. No change was observed for either group on the
POMS anger subscale. After the 10-week programme, exercisers also
exhibited less state anxiety (F(1,26) = 4.15, p < 0.05) and less trait anxiety
(F(1,26) = 6.05, p < 0.02).
Trujillo [116]
(1983)
USA
Randomised
controlled trial
n = 35 female college students;
13 weight trainers, 12 runners
and 10 controls
Self-esteem (Tennessee
Self-concept Scale and the Bem
Sex Role Inventory)
Impact of a 16-week running
programme vs. weight training vs. a
control on self-esteem
Both the running and weight training groups showed a significant increase
in self-esteem from pre- to post-programme (t(11) = 2.11, p < 0.05), while
the control group showed a nonsignificant loss in self-esteem (t(9) = 0.55,
p > 0.05).
Tuckman et al.
[117]
(1986)
USA
Randomised
non-controlled
trial
n = 154 children; aged 9–11
Psychological affects such as
creativity, perceptual function,
behaviour and self-concept
(Alternate Uses Test,
Bender–Gestalt Test, Devereaux
Elementary School Behaviour
Rating Scale and Piers–Harris
Children’s Self-Concept Scale)
Impact of three 30-min runs per week
on an outdoor running track for 12
weeks on psychological affects in
children (creativity, perceptual
function, behaviour and self-concept,
compared to 12 weeks of the school’s
regular physical education
Running significantly improved creativity of school children compared to
regular physical education participants (F ratio = 17.00, p < 0.001) but had
no significant effect on classroom behaviour, perceptual functioning or
self-concept.
Doyne et al. [118]
(1987)
USA
Randomised
controlled trial
n = 40 women; all with a
diagnosis of minor or major
depression; mean age of 28.52
Depression (Beck’s Depression
Inventory, Hamilton Rating Scale
for Depression and Depression
Adjective Checklists)
Impact of 3 runs per week on an
indoor track for 8 weeks on
depression in women diagnosed with
depression, compared to 8 weeks of
weight lifting vs. control
Running statistically and clinically significantly decreased depression
scores (F(4,138) = 14.98, p < 0.01) relative to the wait-list control group,
with improvements reasonably well maintained at 1 year follow-up.
Fremont et al.
[119]
(1987)
USA
Randomised
non-controlled
trial
n = 49; 13 males and 36 females;
aged 19–62
Depression, anxiety and mood
state (Beck’s Depression
inventory, State-Trait Anxiety
Inventory and The Profile of
Mood States)
Impact of 3 runs per week for 10
weeks on depression, anxiety and
mood vs. 10 weeks of counselling vs.
10 weeks of a combination of running
and counselling
Depression, trait anxiety and state anxiety all decreased significantly
((F(4,184) = 50.3, p < 0.0001), (F(1, 46) = 27.1, p < 0.0001), (F(1,46) = 21.9,
p < 0.0001), respectively), while mood improved significantly over the 10
weeks (F(18,378) = 4.5, p < 0.001).
Hannaford et al.
[120]
(1988)
USA
Randomised
controlled trial
n = 27 male psychiatric patients
with major psychiatric disorders;
age range 25–60; 9 runners, 9 in
corrective therapy for 8 weeks
and 9 waiting list controls
Depression and anxiety (Zung
Self Rating Depression Scale and
State Trait Anxiety Index)
Impact of three 30-min runs per week
for 8 weeks on depression and
anxiety in psychiatric patients with
major psychiatric disorders
Significant reductions were observed in depression scores (F(2,23) = 3.61,
p = 0.043) compared to the waiting list controls, and nonsignificant
reductions were observed in anxiety scores (F(2,23) = 1.085, p = 0.354)
compared to the waiting list control group.
Int. J. Environ. Res. Public Health 2020, 17, 8059
22 of 39
Table 6. Cont.
Author
Year
Country
Design
Population
Mental Health Outcome
(Measurement)
Study Aim
Main Findings
Long et al. [121]
(1988)
Canada
Randomised
non-controlled
trial
n = 39 chronically stressed,
sedentary working women; mean
age 40; 18 joggers vs. 21
relaxation intervention
Stress, anxiety and self-efficacy
(Trait Anxiety Inventory, Sherer et
al.’s Inventory for Self-Efficacy
and a modified version of the
Ways of Coping Checklist)
Impact of an 8-week running
programme consisting of a weekly
group session plus twice weekly solo
jogs on stress, anxiety and
self-efficacy
Runners had significantly less anxiety and greater self-efficacy than
baseline; 24% of subjects reached clinically significant improvements at the
end of treatment, and 36% reached clinically significant improvements at
14-month follow-up. The jogging group exhibited higher self-efficacy,
and the time effect for the pre to the post/follow-up average was significant
for both self-efficacy and trait anxiety (F(2, 36) = 15.38, p < 0.001), while
total coping scores did not change (F(2, 35) = 2.88, p < 0.07) from pre to
post/follow-up.
Simons et al. [122]
(1988)
USA
Non-randomised
controlled trial
n = 128; 53 experimental subjects
(24 male, 30 female, mean age
44.9); 75 control subjects (28 male,
47 female, mean age of 42.0)
Mood (Profile of Mood States,
Nowicki–Strickland
Internal–External Control Scale
for Adults and Marlowe–Crowne
Social Desirability Scale)
Impact of two 30 min walk/runs per
week for 8 weeks on mood,
compared to a weekly 30-min fitness
lecture for 8 weeks
Significant improvement in mood pre- to post-test intervention compared
to non-treatment controls (F(1,126) = 4.46, p < 0.05) as well as significant
improvement in pre- to follow-up mood change scores (F(1,98) = 7.63,
p < 0.01) were observed.
Moses et al. [123]
(1989)
UK
Randomised
controlled trial
n = 75 sedentary adult volunteers;
mean age 38.8; four 10-week
conditions: high-intensity aerobic
walk-jog programme (n = 18);
moderate intensity walk-jog
programme (n = 19);
attention-placebo including
strength, mobility and flexibility
exercises (n = 18); or waiting list
control (n = 20).
Mood and mental well-being
(Profile of Mood States and the
Hospital Anxiety and Depression
Scale)
Impact of varying intensity 10 week
walk-jog programmes on mood and
mental well-being
Significant reductions in tension/anxiety (F(3,71) = 2.94, p < 0.05) were
reported only by subjects in the moderate exercise condition. Significant
differences in the confusion subscale were found over time (F(1,71) = 3.70,
p < 0.06), with greater decreases in the moderate exercise group than in the
high exercise, attention-placebo or waiting list conditions. No significant
effects were found on the perceived coping scales, but there was significant
improvement on the physical well-being scale in the exercisers
(F(3,71) = 3.82, p < 0.01) after 10 weeks, while the waiting list group ratings
decreased. At follow-up, only subjects in the moderate exercise condition
reported decreased ratings of depression/dejection (F(2,55) = 3.00, p < 0.06)
and positive changes that approached significance for the perceived
coping assets scale (F(2,55) = 2.56, p < 0.08), but this was not the case for the
high exercise or attention-placebo conditions.
Ossip-Klein et al.
[124]
(1989)
USA
Randomised
controlled trial
n = 32 clinically depressed
women; mean age 28.52
Self-concept (Beck Self-Concept
Test)
Impact of running on an indoor track
4 times per week for 8 weeks on
self-concept in clinically depressed
women compared to weight lifting 4
times weekly vs. a delayed treatment
(assessment only) control
Self-concept significantly improved in the clinically depressed women
compared to controls (F(3,99) = 7.62, p < 0.0001). Self-concept scores were
also significantly higher in those in the running condition compared to the
wait-list condition at post-treatment (F(2, 33) = 4.69, p < 0.05), with
improvements also reasonably well-maintained over time.
Morris et al. [125]
(1990)
UK
Pre-post study
with randomised
comparison
n = 30 male regular runners;
mean age 37; 20 participants
stopped running for 2 weeks vs.
20 continued running as normal
Anxiety and depression (General
Health Questionnaire and short
forms of the Zung Anxiety and
Zung Depression scales)
Impact of stopping running for 2
weeks on anxiety and depression
Somatic symptoms, anxiety/insomnia and social dysfunction, symptoms of
depression (p < 0.05), were all significantly greater in deprived than in
continuing runners, and Zung depression (F(1,37) = 22.64, p < 0.001) and
anxiety (F(1,37) = 11.51, p < 0.01) scores were significantly higher after the
two weeks. Significantly more deprived than non-deprived subjects
exceeded the suggested cutoff score for a psychiatric case after both weeks
of deprivation (x2 = 5.38 and 4.51, respectively, df = 1, p < 0.05), but there
was no statistical difference between groups once the deprived group
resumed running.
Int. J. Environ. Res. Public Health 2020, 17, 8059
23 of 39
Table 6. Cont.
Author
Year
Country
Design
Population
Mental Health Outcome
(Measurement)
Study Aim
Main Findings
Friedman et al.
[126]
(1991)
USA
Randomised
controlled trial
n = 387 students; 177 males and
188 females; mean age 20;
84 joggers, 96 relaxation, 100
group interaction and 107
lecture-control
Stress and mood (Profile of Mood
States and Bem Sex Role
Inventory)
Impact of 12 weeks of jogging on
stress and mood
High masculinity male and female joggers reported significantly more
mood improvement than those with low masculinity (p < 0.004).
All women joggers reported significant reductions in depression after
jogging, but those with high psychological masculinity experienced
significantly greater reductions than low masculinity joggers (p < 0.04).
Femininity had a significant effect on combined POMS scores
(F(6,297) = 2.79, p < 0.02), with higher psychological femininity associated
with higher tension, depression and fatigue and with lower vigour and
confusion scores compared to those low in femininity. There were
significant pre-post session × technique interactions for high and low
masculinity women (F(18,843.36) = 2.47, p < 0.0007; F(18,843.36) = 2.49,
p < 0.0006, respectively). Short-term improvements in POMS scores
depended upon masculinity for women joggers and participants in group
interaction.
Williams et al.
[127]
(1991)
USA
Pre-post
non-controlled
within subject
design
n = 10 moderately trained male
runners; mean age 25.6
Mood (Profile of Mood States)
Impact of 4 weeks of treadmill
running 5 times per week at set paces
reflecting 50, 60 and 70% VO2 max on
mood
Regarding within-subject data, an increase in mean VO2 was associated
with a significant increase in total mood disturbance (r = 0.88, p < 0.01),
while running at a pace with more economical values was associated with
more positive mental health profiles. However, when considered as a
group, there was no relationship between running efficiency in moderately
trained male runners and total mood disturbance.
Kerr et al. [128]
(1993)
Holland
Pre-post
non-controlled
n = 32 regular exercising
university students (18 males and
14 females) aged 18–22
Mood (Stress-Arousal Checklist
and Telic State Measure)
Impact of a weekly 40-min fixed
distance run (5 km for females, 6.6
km for males) through a wooded area
for 7 weeks on mood
In both males and females, there were significant increases from pre- to
post-running intervention in telic state measure felt arousal scores
(F(1,16) = 52.37, p = 0.0001 and F(1,12) = 16.16, p = 0.002, respectively),
stress-arousal checklist arousal scores (F(1,16) = 15.34, p = 0.001 and
F(1,12) = 25.19, p = 0.0001, respectively) and telic state measure preferred
arousal scores (F(1,16) = 4.49, p = 0.05 and F(1,12) = 11.82, p = 0.005,
respectively). In contrast, telic state measure arousal discrepancy scores
decreased significantly for males (F(1,16) = 6.74, p = 0.02) and females
(F(1,12) = 11.86, p = 0.005) pre- to post-running.
Long [129]
(1993)
Canada
Randomised
controlled trial
n = 35; 14 males and 21 females;
mean age 35.6; 12 runners,
9 stress inoculation and 14
wait-list control
Anxiety and stress (Cornell
Medical Symptom Checklist)
Impact of 3 runs per week for 10
weeks on anxiety and stress
Although the exercise group was more likely to report using exercise to
cope with stress, there was no significant differences found between
groups on stress or coping classifications. There was also no significant
difference in scores of the Cornell Medical Symptom Checklist between the
running group and the stress inoculation treatment groups (F < 1; M = 87.4
vs. M = 86.2, respectively).
Berger and
Friedman [130]
(1998)
USA
Randomised
controlled trial
n = 387 undergraduate college
students; 117 males and 188
females; mean age 20.0; 84
joggers vs. 96 relaxation response
vs. 100 in discussion groups vs.
107 in the control group
Stress and mood (Profile of Mood
States)
Impact of three jogs per week for a
minimum of 20 min over 12 weeks on
stress and mood
Jogging was significantly more effective in reducing stress than the control
activity (F(18,280) = 1.79 to 1.85, p < 0.03), and joggers reported larger and
more numerous reductions in tension, depression and anger than the
control group; however, changes in vigour, fatigue and confusion were
sporadic. There were no long-term benefits observed.
Berger and Owen
[131]
(1998)
USA
Pre-post with
comparison
n = 91 college students; n = 67 in
weekly walking/jogging (32 males
and 35 females) vs. n = 24 in a
weekly health science class
(9 males and 15 females)
Mood and anxiety (Profile of
Mood States and State-Trait
Anxiety Inventory)
Impact of twice weekly
walking/jogging for 14 weeks on
mood and anxiety
No significant interaction between exercise intensity and pre-post mood
benefits was observed. Joggers reported significant short-term mood
benefits following running regardless of exercise intensities (F(6,56) = 4.87,
p < 0.0005). Joggers reported significant pre-post exercise changes on all
POMS subscales: tension (F = 15.67, p < 0.0002), depression (F = 15.64,
p < 0.0002), anger (F = 12.77, p < 0.0007), vigour (F = 22.29, p < 0.00005),
fatigue (F = 20.14, p < 0.00005) and confusion (F = 26.34, p < 0.00005).
Int. J. Environ. Res. Public Health 2020, 17, 8059
24 of 39
Table 6. Cont.
Author
Year
Country
Design
Population
Mental Health Outcome
(Measurement)
Study Aim
Main Findings
Szabo et al. [132]
(1998)
UK
Pre-post
non-controlled
observational
cohort study
n = 40 members of an amateur
running club; 30 males, mean age
40.5, and 10 females, mean age 37
Anxiety and mood, i.e.,
exhaustion, tranquillity, positive
engagement and revitalization
(Commitment to running scale,
Exercise-induced Feeling
Inventory and Spielberger State
Anxiety Inventory)
Impact of running vs. non-running
days on anxiety and mood over 21
consecutive days
Reported differences (effect sizes ranging from 0.07 to 0.56, all p < 0.05) all
favour running days over non-running days, concluding that, on running
days, runners experienced less anxiety (F(1,38) = 5.22, p < 0.03) and better
subscales of mood: exhaustion (F(1,38) = 4.34, p < 0.04), tranquillity
(F(1,38) = 5.56, p < 0.02), revitalisation (F(1,38) = 18.32, p < 0.001) and
positive engagement (F(1,38) = 11.79, p < 0.001).
Broman-Fulks et al.
[133]
(2004)
USA
Randomised
non-controlled
trial
n = 54 participants with elevated
anxiety sensitivity scores;
13 males and 41 females; mean
age 21.17; 29 high-intensity
aerobic exercisers vs. 25
low-intensity aerobic exercisers
Anxiety sensitivity (Anxiety
Sensitivity Index, State-trait
Anxiety Inventory and Body
Sensations Questionnaire)
Impact of six 20-min treadmill
sessions of either high/low-intensity
aerobic exercise across 2 weeks on
anxiety sensitivity in participants
with elevated anxiety
sensitivity scores
Six 20-min treadmill sessions of both high-intensity and low-intensity
running across 2 weeks reduced anxiety sensitivity (F(2,56) = 42.50,
p < 0.001, n2 = 0.60; F(2, 48) = 13.72, p < 0.001, n2 = 0.36; respectively).
State anxiety also decreased from pre-post high-intensity running (35.10 to
32.03); however, it increased following low-intensity running (42.72 to
42.32), but neither of these effects were significant.
Haffmans et al.
[134]
(2006)
Holland
Randomised
controlled trial
n = 60 psychiatric patients all
suffering from a depressive
disorder; 19 males and 41 females;
mean age 39; 20 runners vs. 21 in
physiotherapy training vs.
19 controls
Depression and self-efficacy
(Hamilton Rating Scale for
Depression, Becks Depression
Inventory, Self-Efficacy Scale and
Physical Self-Efficacy Scale)
Impact of running therapy for 3 days
per week for 12 weeks on depression
and self-efficacy in psychiatric
patients all suffering from depression
While after 6 weeks of running, self-efficacy was significantly higher
(p = 0.03), after the full 12 weeks of running, there was no significant
difference in depression (26.7 to 25.5, n.s.) or self-efficacy (46.6 to 49.1, n.s.)
scores from baseline.
Thornton et al.
[135]
(2008)
USA
Repeated
measures design
n = 50 runners over age 18
Anxiety (Beck Anxiety Inventory)
The relationship between anxiety and
marathon
Marathon training decreased Beck Anxiety Inventory scores (0.9) initially
from baseline pre-training levels compared to 2 months prior to marathon
day (0.7; 72% had no change from baseline, 22% were less anxious and 6%
were more anxious). However, anxiety scores increased as race day
approached: at 1 month prior to race day (1.4; 46% had no change from
baseline, 19% were less anxious and 35% were more anxious than baseline)
and 1 week prior to race (2.6; 22% had no change from baseline, 14% were
less anxious and 64% were more anxious than baseline, respectively).
Scholz et al. [136]
(2008)
Switzerland
Pre-post
non-controlled
non-experimental
longitudinal study
n = 30 untrained participants;
4 males and 26 females; mean age
41.2
Self-efficacy (4-part
author-created measurement)
Impact of a 1-year marathon training
programme on self-efficacy
The trend between running and self-efficacy had substantial correlation
but was not significant. No statically significant differences was observed
in the baseline level, trend or fluctuation of self-efficacy between the
participants who successfully completed the marathon and those who did
not, but the baseline level of self-efficacy was positively associated with
the baseline level in running (correlation analyses = 0.27; p < 0.05; 95%
CI = 0.00; 0.53) and fluctuation in self-efficacy correlated positively with
fluctuation in running (0.39; p < 0.05; 95% CI = 0.03; 0.74). As this was a
non-experimental longitudinal study, no causal statements can be drawn.
Kalak et al. [137]
(2012)
Switzerland
Randomised
controlled trial
n = 51 adolescents; 24 males and
27 females; mean age 18.3;
27 runners vs. 24 controls
Mood and stress (A-a daily mood
log, a questionnaire assessing
positive and negative comping
strategies, and Perceived
Stress Scale)
Impact of daily 30-min morning runs
on weekdays for 3 weeks (i.e., 3 × 5
runs) on stress and mood
Perceived stress did not differ significantly between running and control
groups over time (F(1,49) = 1.71, n2 = 0.034, n.s.), while mood in the
morning increased significantly over time in the running group compared
with controls (F(5,245) = 16.08, n2 = 0.247, p < 0.05). However, irrespective
of group, mood in the evening improved, and there was no significant
difference of mood in the evening between groups.
Inoue et al. [138]
(2013)
USA
Pre-post
non-controlled
n = 148 homeless people;
134 males and 14 females;
mean age 29.9
Self-sufficiency (author-created
scale)
Impact of 10 organised runs on
self-sufficiency in homeless people
Running involvement had a significant positive correlation with perceived
self-sufficiency (r = 0.30, p < 0.01). Results suggested that participants
gained higher levels of perceived self-sufficiency as they became more
involved with running during the program (F = 3.39, p < 0.01,
Adjusted R2 = 0.08), and increases in running involvement were the sole
significant predictor of the outcome (β = 0.29, t = 3.57, p < 0.01).
Samson et al. [76]
(2013)
USA
Pre-post
non-controlled
n = 39 university students who all
had running experience; 11 males
and 28 females; mean age 20.5
General affect and self-efficacy
(Positive and Negative Affect
Scale and author-created
measurements for self-efficacy)
Impact of a 15-week marathon
training program of 3 group training
days per week and one run of 8–20
miles on the weekend on general
affect and self-efficacy
Self-efficacy significantly increased over the training programme
(F(12,444) = 5.81, p < 0.01), but there was a significant decrease of positive
affect over time (F(12,444) = 8.35, p < 0.01) and no significant change was
found for negative affect over the programme.
Int. J. Environ. Res. Public Health 2020, 17, 8059
25 of 39
Table 6. Cont.
Author
Year
Country
Design
Population
Mental Health Outcome
(Measurement)
Study Aim
Main Findings
Doose et al. [139]
(2015)
Germany
Randomised
controlled trial
n = 46 outpatients diagnosed
with mild to severe depression;
aged 18–65; 30 walker/runner vs.
16 controls
Depression (Hamilton Rating
Scale and Beck Depression
Inventory)
Impact of group walking/running 3
times per week for 8 weeks on
depression
Depression clinically significantly decreased on the Hamilton Rating Scale
(Cohen’s d = 1.8; mean change = 8.24; p = <0.0001), and while there were
reductions, they were without clinical significance (Cohen’s d = 0.50;
mean change = 4.66; p = 0.09) in the Becks Depression Inventory scores.
Von Haaren et al.
[140]
(2015)
Germany
Randomised
controlled trial,
within subject
design
n = 61 inactive male university
students; mean age 21.4
Stress and mood (a shorten mood
scale based on the
Multidimensional Mood
Questionnaire and a one-item test
for perceived control and stress)
Impact of a 20-week running training
course on stress and mood during
academic examinations, compared to
waiting list controls
Significant emotional stress reactivity was observed in both groups during
academic assessment episodes; participants in aerobic training showed
lower emotional stress reactivity compared with the control participants
after the 20-week training programme, with perceived stress of the aerobic
group remaining similar during both exam periods (2.27 to 2.24), while it
increased further in the control group (2.43 to 2.51).
Kahan et al. [141]
(2018)
USA
Pre-post with
comparison
n = 11 children; 9 males and 2
females; aged 9 and 10
Self-esteem and self-efficacy
(50-item, author-created
questionnaire)
Impact of 20 running sessions
alternating between game vs. lap
running on self-esteem and
self-efficacy in children
Means for self-esteem and task-efficacy were 3.63 and 4.16, respectively,
on a 5-point scale, while the mean for task-efficacy was 4.16 on a 5-point
scale, and high inherent-interest participants (i.e., higher
moderate–vigorous physical activity in the running laps condition) had
statistically significant higher scores than low inherent-interest
participants on recognition (p = 0.01), ego orientation (p = 0.03) and
expectancy beliefs (p = 0.03) subscales. There were no direct comparisons
of self-esteem and self-efficacy in game vs. lap running.
Keating et al. [142]
(2018)
Canada
Pre-post
non-controlled
n = 46 participants with complex
mood disorders; 11 males and 35
females; 29 youths (mean age
22.1) and 17 adults
(mean age 45.2)
Stress, anxiety and depression
(Cohen’s Perceived Stress Scale,
Becks Depression Inventory,
Becks Anxiety Inventory and
Short Form Survey)
Impact of 12 weeks of twice weekly
running in a group setting that offers
social support supervised by clinical
professionals on stress, anxiety and
depression
Significant decreases in depression (F(11,201) = 4.5, p < 0.0001),
anxiety (F(11,186) = 4.8, p < 0.0001) and stress (F(11,186) = 2.3, p = 0.01) from
baseline was observed. Following intervention, mean depression scores
decreased by 39% in adults from high to low levels and by 27% in youths
from moderate to reduced moderate levels. Younger participant age,
younger age at onset of illness and higher perceived levels of friendship
with other running group members (ps ≤ 0.04) were associated with lower
depression, anxiety and stress scores. Higher attendance was linked with
decreasing depression and anxiety (ps ≤ 0.01) scores over time.
Nezlek et al. [143]
(2018)
Poland
Pre-post
observational
cohort study over
3 months with no
control
n = 244 recreational runners; 127
males and 117 females; mean age
32.5
Psychological well-being,
self-esteem, self-efficacy and
affect (Rosenberg Self-esteem
Scale, Satisfaction with Life Scale,
and a circumplex model that
distinguishes the valence and
arousal of affect)
Impact of 3 months of self-prescribed
running on psychological well-being,
self-esteem, self-efficacy and affect
Positive within-person relationships between how much people ran each
week and self-reports of well-being were observed, with well-being
significantly higher during weeks when individuals ran more often and
further. Self-efficacy was related to distance run but not to frequency.
For the km that people ran each week, significant moderation was found
for weekly Satisfaction with Life Scale (γ11 = −0.0002, p = 0.013),
self-esteem (γ11 = −0.0002, p = 0.015), positive activated affect
(γ11 = −0.0003, p < 0.001), positive deactivated affect (γ11 = −0.0008,
p < 0.01), negative activated affect (γ11 = 0.0002, p = 0.046) and negative
deactivated affect (γ11 = 0.0003, p = 0.01).
Kruisdijk et al.
[144]
(2019)
Holland
Randomised
controlled trial
n = 48 participants with major
depressive disorder; mean age
42.6; 25 runner-walkers vs. 23
controls
Depression (Hamilton
Depression Scale)
Impact of 6 months of
running-walking for one hour twice a
week on depression in subjects with
major depressive disorder
No significant difference or effect on depression in favour of the
intervention group (Cohen’s d < 0.2, F = 0.13, p = 0.73) with only 9
participants (19%) completing the study was found, with low statistical
power and lack of follow-up at six and 12 months.
Int. J. Environ. Res. Public Health 2020, 17, 8059
26 of 39
A number of studies looked at specific populations. One investigated the impact of 10 organised
runs on homeless people and found significant positive correlation with perceived self-sufficiency [138].
Two investigated the effects in children and found that running significantly improved creativity and
higher self-esteem subscales [117,141]. Three looked at marathon training programmes: one found
a positive correlation between the trend in running and self-efficacy but was not significant [136],
while another found a significant increase in self-efficacy over the programme [76]. The remaining
study used participants who were already self-enrolled in a marathon, and researchers found that,
while anxiety decreased initially during training, anxiety increased as marathon day approached [135].
Nine studies used subjects with known psychiatric disorders and found that longer-term
interventions generally improved markers of mental health in psychiatric populations, particularly
markers of depression [112,113,118,120,124,133,134,142,144]. Running interventions from 2 to 12 weeks
all resulted in significant positive effects on mental health [112,118,120,124,133,142,144]. While an
anti-depressive effect of exercise was apparent in patients with minor to moderate psychiatric problems,
one study found that this was not reflected in patients with major depressive disorder due to issues
with compliance and motivation towards the intervention [144].
Summary of Longer-Term Interventions
Overall, running interventions of 2–20 weeks generally show improved markers of a range of
mental health outcomes compared to non-running controls, including mental health outcomes in
psychiatric and homeless populations. The risk of longer-term running interventions on adverse
mental health outcomes remains unclear.
3.5. Summary of Key Findings
The key findings of the each of the three categories of studies are summarised in Table 7.
Table 7. Summary of key findings within each of the three categories.
Study Type
Number of Studies
Summary of Evidence
Cross-sectional
47 studies
Consistent evidence was found for a positive association between mental health and
habitual or long-term recreational running compared to non-runners. In contrast,
there was evidence that high or extreme levels of running were associated with
markers of running ill-health compared to levels of moderate running.
Acute: single/double/triple bout
35 studies
Overall, these studies suggest that acute bouts of running can improve mental health
and that the type of running can lead to differential effects. Evidence suggests that
acute bouts of treadmill, track, outdoor and social running (2.5–20 km and 10–60 min)
all result in improved mental health outcomes. There were few differences between
high and low intensities. Studies consistently show that any running improves
acute/short-term mood markers but that lack of inactive comparisons limits the
strength of evidence. Little variation in the demographics of participants and small
sample sizes limit generalizability and precision of findings.
Interventions (2 weeks or more)
34 studies
Overall, running interventions of 2–20 weeks generally show improved markers of a
range of mental health outcomes compared to non-running controls, including mental
health outcomes in psychiatric and homeless populations. The risk of longer-term
running interventions on adverse mental health outcomes remains unclear.
3.6. Evidence Gaps
As well as reporting the available evidence, this review also aimed to identify key gaps in the
evidence base for running and mental health. Consideration of sample demographics in the n = 116
included studies resulted in the following gaps being identified:
•
lack of studies in those aged under 18 (Only four acute bout studies [89,95,107,108] and two longer
term interventions [117,141] looked directly at children under age 15, while a further 2 studies
looked specifically at adolescents [70,137]);
•
lack of studies in those aged over 45;
•
lack of gender-specific approaches;
•
few studies investigating clinical populations; and
•
limited diversity in patient demographics.
Int. J. Environ. Res. Public Health 2020, 17, 8059
27 of 39
4. Discussion
4.1. Principal Findings
There is a growing body of literature exploring the relationships of running on certain mental health
outcomes. There were variations in methods and outcomes studied, but there were similar overall
beneficial trends. Generally, evidence supported positive effects of a range of lengths and intensities
of running on mental health. However, there was limited diversity in participant demographics.
Attribution was also compromised by the limited number of studies with comparisons/control groups.
Synthesis of quantified effects is made challenging by large variations in reporting methods. Consistency
and appropriateness of mental health measures was also varied throughout the literature.
The review identified a smaller evidence-base focused on clinical populations. Behaviour change
and compliance can be challenging in populations with clinical depressive disorders [145], and there
is limited evidence regarding the long-term impact of PA in the treatment of depression [7,146,147].
Further investigations of the effects of running in populations with prior diagnoses of mental health
disorders may help to address the global burden of mental illness.
4.2. Plausible Explanations for Findings
Our findings suggest that, throughout cross-sectional evidence, acute bouts of running and
longer-term running interventions are associated with improvements in a range of mental health
outcomes. This is likely explained by running supplying a sufficient dose of moderate to vigorous
PA to stimulate the known mental health benefits associated with PA. These benefits are thought to
be mediated by neurobiological, psychosocial and behavioural mechanisms, all of which an effective
running intervention of any genre has the potential to influence [148]. The differential effects of these
mechanisms remain unclear and may explain the variation in findings by running duration, intensity,
setting, and social or individual participation.
4.3. Comparison to Literature
This review does not present running and mental health as a novel idea. As early as 1979,
scholars discussed the relationship between psychotherapy and running [149]. An early review by
Vezina et al. (1980) reported that regular running causes positive mood changes, increases self-esteem
and decreases anxiety [150]. Another review by Hinkle (1992) found positive psychological effects
in both adults and children including reductions in depressive mood and anxiety, and enhanced
self-esteem [151]. However, a review by Weinstein et al. (1983) found that the volume of literature
examining running and depression was scarce, and while running appeared to improve a sense of
well-being, there was minimal evidence to strongly support reductions in depression and anxiety [152].
Studies from 1986 [153] and 1991 [154] warned that long-distance running had the potential to
trigger development of eating disorders in people who were psychologically or biologically at risk.
Early research also highlighted that runners should be aware of the possibility of addiction [155] and
that women may be linked more strongly to negative addiction than men [156].
This review agrees with these earlier findings but is the first to use systematic scoping review
methods. This means that it presents a transparent search and inclusion strategy and is less prone to
bias in terms of included studies and resulting findings. As such, this review has contributed to the
evidence base by demonstrating that the weight of evidence up to 2019 favours positive mental health
relationships with running.
4.4. Strengths and Limitations
The authors acknowledge the limitation that this review was designed to assess the behaviour
of running but that there are fields of studies including treadmill-based exercise which our review
may not have picked up. However, the strength of this review is that the review does not focus on
laboratory-based exercise but instead on what a healthcare professional may recommend to a free-living
Int. J. Environ. Res. Public Health 2020, 17, 8059
28 of 39
patient or the general public for mental health benefits. However, subjective measures of running
intensity were not considered in detail, which may impact the conclusions of the review. The authors
acknowledge that the results were not separated by means of running type due to the method of
prioritization used to report the results, and thus, this remains a research gap. As with any scoping
review, it is possible that the search and inclusion strategy led to omission of some key research.
Synthesis of quantified effects was also made challenging by the large range of reporting methods
used within the studies. This scoping review did not attempt to undertake quality appraisal of the
included studies. The wide range of study designs and methods included within the review does not
allow a statistical synthesis of the effectiveness of the studies.
4.5. Implications
Pharmacological management is often used as a first-line of defence for mental health
disorders [157]; however, it is not always effective due to poor adherence and relapse [158].
Ineffective management adds to the global burden of poor mental health [159], With increasing
pressures on healthcare budgets, PA offers an augmentative therapeutic option for mental health
management [160]. It is likely that using a cost-effective therapy such as running to improve mental
health would prove economical as well. An integrated lifestyle intervention (i.e., iterative process) may
be more feasible than a single add-on exercise intervention (i.e., addition of an individual behaviour)
for patients with major depressive disorder who are deemed suitable for running therapy by clinicians.
This review presents the effects of running on mental health and can inform healthcare professionals
and psychologists who advise on management of mental health conditions. The authors’ interpretation
of the evidence base is that, with appropriate clinical judgement, practitioners may identify patients
with an interest in running or previous history of running as an ideal candidate for running as a form of
psychotherapy. Findings from this review indicate that characteristics of running to be recommended
may include self-pacing, distance and time feasibility to the individual, and being within the lactate
threshold. There were consistent trends within findings despite a variety of running interventions,
which suggest that it would be appropriate to recommend track running, outdoor urban and rural
running, and treadmill running to improve mental health. However, a large number of studies used
healthy, active college-aged participants, which may limit the relevance of these recommendations
to other population groups. It is acknowledged that running will not be a suitable recommendation
for everyone and that prescription of running is not as simple as just instructing people to run; it will
require clinical expertise with regard to mental health in the way it is prescribed [161].
4.6. Future Research
This review identifies research gaps regarding patient demographics, but we have further
recommendations about increasing sample sizes, quantitative study design and more coherent mental
health outcomes. There was great variability in mental health outcome measures, particularly within
the acute bout studies, where short-term measures of mental health could have equally been defined
as mood and affect. We recommend that future research seeks more clarity on appropriate outcome
measures. A comparison of types, settings and intensities of running is needed to better inform running
and mental health recommendations.
Recommendations for future research include addressing the effect of running on mental health
of those under 18, those over 50s and clinical populations. A meta-analysis of the subset of study
types such as interventions should be carried out. While the appropriateness of running interventions
in those over 50 may be questioned, there is evidence that older adults do also benefit from the
anti-depressive effect of exercise [162]. We know that children running can be used as a population
intervention, for example, in “The Daily Mile” [163], which signifies the importance of addressing this
gap around the mental health impact of running in those under 18. Future systematic reviews and
meta-analyses are needed to quantify the benefits of running on specific outcomes.
Int. J. Environ. Res. Public Health 2020, 17, 8059
29 of 39
5. Conclusions
This review is the most recent to comprehensively report the breadth of literature on the relationship
between running and mental health. We conclude that running has important positive implications
for mental health, particularly depression and anxiety disorders, but synthesis of quantified effects is
made challenging by variation in reporting methods and remains a gap. This scoping review may have
consequences for researchers, practitioners and relevant organisations and may inform the practice
of healthcare professionals. Knowledge gaps concerning running on the mental health of children,
older adults and clinical populations provide guidance for future research
Supplementary Materials: The following are available online at http://www.mdpi.com/1660-4601/17/21/8059/s1,
Table S1: Narrative description of findings of the 47 cross-sectional studies. Table S2–S4: Narrative description
of findings of the 35 studies with an acute bout of running. Table S5: Narrative description of findings of the
34 studies with a longer-term intervention of running.
Author Contributions: P.K., J.R. and F.O. conceived the study. P.K., C.W. and F.O. designed the search strategy.
F.O. conducted searching of databases. J.C., P.K., F.O. and C.W. screened the records. F.O. and P.K. screened the
full texts. F.O. completed all data extraction, and J.R. conducted quality checks. F.O. drafted the full manuscript,
and all authors reviewed and approved final submission. All authors have read and agreed to the published
version of the manuscript.
Funding: This research received no external funding.
Acknowledgments: The authors would like to thank a number of people for their assistance during the scoping
review: Thelma Dugmore for her support and administration help within the Edinburgh University Physical
Activity for Health Research Centre (PAHRC), Marshall Dozier for her assistance setting up a Covidence account
for the project to run through, all the staff at PAHRC for being so welcoming and interested in the project and,
finally, Colin Oswald who was a source of great encouragement and support throughout the project.
Conflicts of Interest: The authors declare no conflict of interest.
Appendix A
Table A1. Preferred Reporting Items for Systematic reviews and Meta-Analyses extension for Scoping
Reviews (PRISMA-ScR) Checklist [18].
Section
Item
PRISMA-ScR CHECKLIST ITEM
Reported on Page
Number
Title
Title
1
Identify the report as a scoping review.
1
Abstract
Structured summary
2
Provide a structured summary that includes (as applicable)
background, objectives, eligibility criteria, sources of evidence,
charting methods, results and conclusions that relate to the
review questions and objectives.
1
Introduction
Rationale
3
Describe the rationale for the review in the context of what is
already known. Explain why the review questions/objectives
lend themselves to a scoping review approach.
1–2
Objectives
4
Provide an explicit statement of the questions and objectives
being addressed with reference to their key elements (e.g.,
population or participants, concepts and context) or other
relevant key elements used to conceptualize the review
questions and/or objectives.
1–2
Methods
Protocol and registration
5
Indicate whether a review protocol exists; state if and where it
can be accessed (e.g., a Web address), and if available, provide
registration information, including the registration number.
2
Eligibility criteria
6
Specify characteristics of the sources of evidence used as
eligibility criteria (e.g., years considered, language and
publication status), and provide a rationale.
3
Information sources *
7
Describe all information sources in the search (e.g., databases
with dates of coverage and contact with authors to identify
additional sources) as well as the date that the most recent
search was executed.
4
Search
8
Present the full electronic search strategy for at least 1
database, including any limits used, such that it could be
repeated.
Appendix B
Selection of sources of
evidence †
9
State the process for selecting sources of evidence (i.e.,
screening and eligibility) included in the scoping review.
4
Int. J. Environ. Res. Public Health 2020, 17, 8059
30 of 39
Table A1. Cont.
Section
Item
PRISMA-ScR CHECKLIST ITEM
Reported on Page
Number
Data charting process ‡
10
Describe the methods of charting data from the included
sources of evidence (e.g., calibrated forms or forms that have
been tested by the team before their use and whether data
charting was done independently or in duplicate) and any
processes for obtaining and confirming data from
investigators.
4
Data items
11
List and define all variables for which data were sought and
any assumptions and simplifications made.
4
Critical appraisal of
individual sources of
evidence §
12
If done, provide a rationale for conducting a critical appraisal
of included sources of evidence; describe the methods used
and how this information was used in any data synthesis (if
appropriate).
N/A
Synthesis of results
13
Describe the methods of handling and summarizing the data
that were charted.
4
Results
Selection of sources of
evidence
14
Give numbers of sources of evidence screened, assessed for
eligibility and included in the review, with reasons for
exclusions at each stage, ideally using a flow diagram.
4
Characteristics of sources
of evidence
15
For each source of evidence, present characteristics for which
data were charted and provide the citations.
Tables 2–6
Critical appraisal within
sources of evidence
16
If done, present data on critical appraisal of included sources
of evidence (see item 12).
N/A
Results of individual
sources of evidence
17
For each included source of evidence, present the relevant
data that were charted that relate to the review questions and
objectives.
Tables 2–6
Synthesis of results
18
Summarize and/or present the charting results as they relate to
the review questions and objectives.
5–40
Discussion
Summary of evidence
19
Summarize the main results (including an overview of
concepts, themes and types of evidence available), link to the
review questions and objectives and consider the relevance to
key groups.
40–41
Limitations
20
Discuss the limitations of the scoping review process.
41
Conclusions
21
Provide a general interpretation of the results with respect to
the review questions and objectives as well as potential
implications and/or next steps.
42
Funding
Funding
22
Describe sources of funding for the included sources of
evidence as well as sources of funding for the scoping review.
Describe the role of the funders of the scoping review.
43
JBI = Joanna Briggs Institute; PRISMA-ScR = Preferred Reporting Items for Systematic reviews and Meta-Analyses
extension for Scoping Reviews. * From where sources of evidence (see second footnote) are compiled, such as
bibliographic databases, social media platforms and websites. † A more inclusive/heterogeneous term used to
account for the different types of evidence or data sources (e.g., quantitative and/or qualitative research, expert
opinion and policy documents) that may be eligible in a scoping review as opposed to only studies. This is not to be
confused with information sources (see first footnote). ‡ The frameworks by Arksey and O’Malley (1) and Levac and
colleagues (2) and the JBI guidance (3,4) refer to the process of data extraction in a scoping review as data charting.
§ The process of systematically examining research evidence to assess its validity, results and relevance before using
it to inform a decision. This term is used for items 12 and 19 instead of “risk of bias” (which is more applicable to
systematic reviews of interventions) to include and acknowledge the various sources of evidence that may be used
in a scoping review (e.g., quantitative and/or qualitative research, expert opinion and policy document).
Appendix References
(1)
Arksey, H.; O’Malley, L. Scoping studies: Towards a methodological framework. Int. J. Soc.
Res. Methodol. 2005, 8, 19–32.
(2)
Levac, D.; Colquhoun, H; O’Brien, K.K. Scoping studies:
Advancing the methodology.
Implement Sci. 2010, 5, 69, doi:10.1186/1748-5908-5-69.
(3)
Peters, M.D.; Godfrey, C.M.; Khalil, H.; McInerney, P.; Parker, D.; Soares, C.B. Guidance for
conducting systematic scoping reviews.
Int.
J. Evid.
Based Healthc.
2015, 13, 141–146,
doi:10.1097/XEB.0000000000000050.
(4)
Peters, M.D.J.; Godfrey, C.; McInerney, P.; Baldini Soares, C.; Khalil, H.; Parker, D. Scoping reviews.
In Joanna Briggs Institute Reviewer’s Manual; Aromataris, E., Munn, Z., Eds.; Joanna Briggs Inst:
Adelaide, Australia, 2017.
Int. J. Environ. Res. Public Health 2020, 17, 8059
31 of 39
Appendix B
Table A2. Details of the search strategy used in all 4 databases.
Notes for Database
Ovid (Embase)
Ovid (Medline)
Sport DISCUS (Ebscohost)
ProQuest Social Science Journals
Ti,ab Searches Title & Abstract
Ti,ab Searches Title & Abstract
AB Searches Abstract
Ab Searches Abstract
Running Search Terms
Run*, Jog*, Sprint*, Park-run, Orienteer,
Orienteering, Marathon, Marathon-running,
Treadmill
Mental Health Search Terms
Mental Health, Mental illness, Mental state,
Emotions, Emotional, Depression, Depressive
disorder, depressive therapy, Postnatal depression,
Postpartum depression, Seasonal affective disorder,
Situational depression, Atypical depression,
Persistent depressive disorder, anxiety, loneliness,
stress, mood, self-efficacy, sleep, psychological,
psychological characteristics, psychology, eating
disorder, disordered eating, anorexia, bulimia,
exercise, health status disparities, quality of life,
motivation, adjustment disorder, sick role.
relaxation, lifestyle, exercise therapy, social support
NOT search Terms
Rodent, mouse, rat, bovine, pig, animal*, horses,
mice, ecology, dermatology, epigenetics, gene*,
molecule*, cell*, phenotype, drug*, hormone*, food,
nutrient*, glucose, imaging, football, tennis,
swimming, heart, troponin, cardiology, lung,
respiratory, bone, cesarean, newborn, breast-feeding,
HIV, cough, rectal, protocol, procedure, surgery,
operation, stroke, sacroiliitis, COPD, asthma,
Apnoea, angina, allergy, railway, falling
Search Syntax
(remember the ‘adj’ function)
(remember the ‘adj’ function)
(remember the ‘adj’ function)
(remember the ‘adj’ function)
(((“mental-health” or “mental-illness” or “mental-state” or emotions
or emotional or depression or “depressive-disorder” or
“depressive-therapy” or “postpartum-depression” or
“seasonal-affective-disorder” or “situational-depression” or
“atypical-depression” or “persistent-depressive-disorder” or anxiety
or loneliness or stress or mood or “self-efficacy” or sleep or
psychological or “psychological-characteristics” or psychology or
“eating-disorder” or “disordered-eating” or anorexia or Bulimia or
exercise or “health-status-disparities” or “quality-of-life” or
motivation or “adjustment-disorder” or “sick-role” or relaxation or
lifestyle or “exercise-therapy” or “social-support”)) AND (run* or
Jog* or sprint* or “park-run” or orienteer or orienteering or marathon
or “Marathon-running” or treadmill) NOT (dermatology OR
epigenetics OR gene* OR drug* OR surgery OR hormone* OR food
OR imaging OR animal* OR football OR tennis OR swimming OR
rodent OR mouse OR rat OR pig OR bovine OR phenotype or Heart
or cardiology OR lung or bone OR caesarean OR HIV OR troponin
OR cough OR protocol OR breast-feeding OR cell* OR sacroiliitis OR
rectal or procedure OR COPD or respiratory OR nutrient* OR glucose
or newborn OR stroke OR asthma OR operation OR horses OR falling
OR railway OR molecule* OR apn?ea OR angina OR allergy OR mice
OR ecology)).ab,ti.
(((“mental-health” or “mental-illness” or “mental-state” or emotions
or emotional or depression or “depressive-disorder” or
“depressive-therapy” or “postpartum-depression” or
“seasonal-affective-disorder” or “situational-depression” or
“atypical-depression” or “persistent-depressive-disorder” or anxiety
or loneliness or stress or mood or “self-efficacy” or sleep or
psychological or “psychological-characteristics” or psychology or
“eating-disorder” or “disordered-eating” or anorexia or Bulimia or
exercise or “health-status-disparities” or “quality-of-life” or
motivation or “adjustment-disorder” or “sick-role” or relaxation or
lifestyle or “exercise-therapy” or “social-support”) and (run* or Jog*
or sprint* or “park-run” or orienteer or orienteering or marathon or
“Marathon-running” or treadmill)) not (dermatology or epigenetics
or gene* or drug* or surgery or hormone* or food or imaging or
animal* or football or tennis or swimming or rodent or mouse or rat
or pig or bovine or phenotype or Heart or cardiology or lung or bone
or caesarean or HIV or troponin or cough or protocol or
breast-feeding or cell* or sacroiliitis or rectal or procedure or COPD
or respiratory or nutrient* or glucose or newborn or stroke or asthma
or operation or horses or falling or railway or molecule* or apn?ea or
angina or allergy or mice or ecology)).ab,ti.
(AB(run* OR jog* OR sprint OR “park run” OR orienteer OR
orienteering OR marathon OR “marathon-running” OR treadmill)
AND AB(“mental health” OR “mental illness” OR “mental state” OR
emotions OR emotional OR depression OR “depressive disorder” OR
“depressive therapy” OR “postpartum depression” OR “seasonal
affective disorder” OR “situational depression” OR “atypical
depression” OR “persistent depressive disorder” OR anxiety OR
loneliness OR stress OR mood OR “self-efficacy” OR sleep OR
psychological OR “psychological characteristics” OR psychology OR
“eating disorder” OR “disordered eating” OR anorexia OR bulimia
OR exercise OR “health status disparities” OR “quality-of-life” OR
motivation OR “adjustment disorder” OR “sick role” OR relaxation
OR lifestyle OR “exercise therapy” OR “social-support”)) NOT
(dermatology OR epigenetics OR gene* OR drug* OR surgery OR
hormone* OR food OR imaging OR animal* OR football OR tennis
OR swimming OR rodent OR mouse OR rat OR pig OR bovine OR
phenotype or Heart or cardiology OR lung or bone OR caesarean OR
HIV OR troponin OR cough OR protocol OR breast-feeding OR cell*
OR sacroiliitis OR rectal or procedure OR COPD or respiratory OR
nutrient* OR glucose or newborn OR stroke OR asthma OR operation
OR horses OR falling OR railway OR molecule* OR apnoea OR
angina OR allergy OR mice OR ecology)
(ab((run* OR jog* OR sprint OR “park run” OR orienteer OR
orienteering OR marathon OR “marathon-running” OR treadmill))
AND ab((“mental health” OR “mental illness” OR “mental state” OR
emotions OR emotional OR depression OR “depressive disorder” OR
“depressive therapy” OR “postpartum depression” OR “seasonal
affective disorder” OR “situational depression” OR “atypical
depression” OR “persistent depressive disorder” OR anxiety OR
loneliness OR stress OR mood OR “self-efficacy” OR sleep OR
psychological OR “psychological characteristics” OR psychology OR
“eating disorder” OR “disordered eating” OR anorexia OR bulimia
OR exercise OR “health status disparities” OR “quality-of-life” OR
motivation OR “adjustment disorder” OR “sick role” OR relaxation
OR lifestyle OR “exercise therapy” OR “social-support”)) NOT
ab((dermatology OR epigenetics OR gene* OR drug* OR surgery OR
hormone* OR food OR imaging OR animal* OR football OR tennis
OR swimming OR rodent OR mouse OR rat OR pig OR bovine OR
phenotype OR Heart OR cardiology OR lung OR bone OR caesarean
OR HIV OR troponin OR cough OR protocol OR breast-feeding OR
cell* OR sacroiliitis OR rectal OR procedure OR COPD OR respiratory
OR nutrient* OR glucose OR newborn OR stroke OR asthma OR
operation OR horses OR falling OR railway OR molecule* OR apnoea
OR angina OR allergy OR mice OR ecology))) AND
(stype.exact(“Scholarly Journals”) AND la.exact(“ENG”))
Search complete?
Yes
Yes
Yes
Yes
Search saved?
Yes
Yes
Yes
Yes
Saved under:
Embase RunningMH
Medline RunningMH
Sport Discus Running MH
ProQuest RunningMH
Number of hits:
10,131 Text results (this had a limit of only human studies, as well as
a limit for articles and articles in press applied to the search)
10,154 text results (this had a limit of human studies applied to the
search)
3461 (this had a limit of English studies only, and academic journal
only applied to the search)
5933 (this search was carried out within the sports medicine and
education index database and in the social sciences database)
Uploaded to Covidence?
Yes
Yes
Yes
Yes
Int. J. Environ. Res. Public Health 2020, 17, 8059
32 of 39
References
1.
Whiteford, H.A.; Ferrari, A.J.; Degenhardt, L.; Feigin, V.; Vos, T. The global burden of mental, neurological
and substance use disorders: An analysis from the Global Burden of Disease Study 2010. PLoS ONE 2015,
10, e0116820. [CrossRef] [PubMed]
2.
Lopez, A.D.; Murray, C.C.J.L. The global burden of disease, 1990–2020. Nat. Med. 1998, 4, 1241–1243.
[CrossRef] [PubMed]
3.
GBD 2017 DALYs and HALE Collaborators. ‘Global, regional, and national disability-adjusted life-years
(DALYs) for 359 diseases and injuries and healthy life expectancy (HALE) for 195 countries and territories,
1990-2017: A systematic analysis for the Global Burden of Disease Study 2017. Lancet Lond. Engl. 2018, 392,
1859–1922. [CrossRef]
4.
Demyttenaere, K.; Bruffaerts, R.; Posada-Villa, J.; Gasquet, I.; Kovess, V.; Lepine, J.P.; Angermeyer, M.C.;
Bernert, S.; De Girolamo, G.; Morosini, P.; et al. Prevalence, severity, and unmet need for treatment of mental
disorders in the World Health Organization World Mental Health Surveys. JAMA 2004, 291, 2581–2590.
[CrossRef] [PubMed]
5.
Chan, J.S.Y.; Liu, G.; Liang, D.; Deng, K.; Wu, J.; Yan, J.H. Special Issue-Therapeutic Benefits of Physical
Activity for Mood: A Systematic Review on the Effects of Exercise Intensity, Duration, and Modality. J. Psychol.
2019, 153, 102–125. [CrossRef]
6.
Richards, J.; Jiang, X.; Kelly, P.; Chau, J.; Bauman, A.; Ding, D. Don’t worry, be happy: Cross-sectional
associations between physical activity and happiness in 15 European countries. BMC Public Health 2015,
15, 53. [CrossRef]
7.
Schuch, F.B.; Vancampfort, D.; Richards, J.; Rosenbaum, S.; Ward, P.B.; Stubbs, B. Exercise as a treatment for
depression: A meta-analysis adjusting for publication bias. J. Psychiatr. Res. 2016, 77, 42–51. [CrossRef]
8.
Kelly, P.; Williamson, C.; Niven, A.G.; Hunter, R.; Mutrie, N.; Richards, J. Walking on sunshine: Scoping
review of the evidence for walking and mental health. Br. J. Sports Med. 2018, 52, 800–806. [CrossRef]
9.
Active Lives. Sport England. Available online: https://www.sportengland.org/know-your-audience/data/
active-lives (accessed on 27 September 2020).
10.
Couch to 5K: Week by Week-NHS’. Available online: https://www.nhs.uk/live-well/exercise/couch-to-5k-
week-by-week/ (accessed on 28 November 2019).
11.
Nonprofit Girls Empowerment Program | Girls on the Run.
GOTR. Available online: https://www.
girlsontherun.org/ (accessed on 28 November 2019).
12.
Parkrun Eases the Loneliness of the Long-Distance Runner | British Journal of General Practice’.
Available online: https://bjgp.org/content/64/625/408 (accessed on 28 November 2019).
13.
Sifers, S.K.; Shea, D.N. Evaluations of Girls on the Run/Girls on Track to Enhance Self-Esteem and Well-Being.
J. Clin. Sport Psychol. 2013, 7, 77–85. [CrossRef]
14.
Grunseit, A.; Richards, J.; Merom, D. Running on a high: Parkrun and personal well-being. BMC Public Health
2017, 18, 59. [CrossRef]
15.
Parkrun Practice. Available online: https://r1.dotdigital-pages.com/p/49LX-52M/parkrunpractice (accessed on
28 November 2019).
16.
Patel, V.; Garrison, P.; Mari, J.D.; Minas, H.; Prince, M.; Saxena, S. The Lancet’s Series on Global Mental
Health: 1 year on. Lancet 2008, 372, 1354–1357. [CrossRef]
17.
Scoping Studies:
Towards a Methodological Framework:
International Journal of Social Research
Methodology:
Vol 8,
No 1’.
Available online:
https://www.tandfonline.com/doi/full/10.1080/
1364557032000119616 (accessed on 27 September 2019).
18.
Tricco, A.C.; Lillie, E.; Zarin, W.; O’Brien, K.K.; Colquhoun, H.; Levac, D.; Moher, D.; Peters, M.D.; Horsley, T.;
Weeks, L.; et al.
PRISMA Extension for Scoping Reviews (PRISMA-ScR): Checklist and Explanation.
Ann. Intern. Med. 2018, 169, 467. [CrossRef]
19.
American Psychological Association.
Diagnostic and Statistical Manual of Mental Disorders (DSM-5®);
American Psychiatric Pub: Washington, DC, USA, 2013.
20.
Physical Activity Guidelines Advisory Committee Report, 2008 to the Secretary of Health and Human Services:
(525442010-001); American Psychological Association: Worcester, MA, USA, 2008. [CrossRef]
21.
Zulkosky, K. Self-Efficacy: A Concept Analysis. Nurs. Forum (Auckl.) 2009, 44, 93–102. [CrossRef]
22.
Ridner, S.H. Psychological distress: Concept analysis. J. Adv. Nurs. 2004, 45, 536–545. [CrossRef] [PubMed]
Int. J. Environ. Res. Public Health 2020, 17, 8059
33 of 39
23.
Fairburn, C.G.; Brownell, K.D. Eating Disorders and Obesity: A Comprehensive Handbook; Guilford Press:
New York, NY, USA, 2005.
24.
Pressman, S.D.; Cohen, S. Does positive affect influence health? Psychol. Bull. 2005, 131, 925–971. [CrossRef]
25.
Goodman, A. Addiction: Definition and implications. Br. J. Addict. 1990, 85, 1403–1408. [CrossRef]
26.
Ryff, C.D. Eudaimonic well-being and health: Mapping consequences of self-realization. In The Best within
Us: Positive Psychology Perspectives on Eudaimonia; American Psychological Association: Washington, DC,
USA, 2013; pp. 77–98.
27.
Kinch, J.W. A Formalized Theory of the Self-Concept. Am. J. Sociol. 1963, 68, 481–486. [CrossRef]
28.
Lane, A.M.; Terry, P.C. The Nature of Mood: Development of a Conceptual Model with a Focus on Depression.
J. Appl. Sport Psychol. 2000, 12, 16–33. [CrossRef]
29.
Wilson, V.E.; Morley, N.C.; Bird, E.I. Mood profiles of marathon runners, joggers and non-exercisers.
Percept. Mot. Skills 1980, 50, 117–118. [CrossRef]
30.
Joesting, J. Running and Depression. Percept. Mot. Skills 1981, 52, 442. [CrossRef]
31.
Jorgenson, D.E.; Jorgenson, C.B. Perceived Effects of Running/Jogging: A Social Survey of Three Clubs.
Int. Rev. Sport Sociol. 1981, 16, 75–85. [CrossRef]
32.
Valliant, P.M.; Bennie, F.A.; Valiant, J.J. Do marathoners differ from joggers in personality profile: A sports
psychology approach. J. Sports Med. Phys. Fit. 1981, 21, 62–67.
33.
Francis, K.T.; Carter, R. Psychological characteristic of joggers. J. Sports Med. Phys. Fit. 1982, 22, 386–391.
34.
Hailey, B.J.; Bailey, L.A. Negative addiction in runners: A quantitative approach. J. Sport Behav. 1982, 5,
150–154.
35.
Callen, K.E. Mental and emotional aspects of long-distance running. Psychosomatics 1983, 24, 133–134.
[CrossRef]
36.
Galle, P.C.; Freeman, E.W.; Galle, M.G.; Huggins, G.R.; Sondheimer, S.J. Physiologic and psychologic profiles
in a survey of women runners. Fertil. Steril. 1983, 39, 633–639. [CrossRef]
37.
Lobstein, D.D.; Mosbacher, B.J.; Ismail, A.H. Depression as a powerful discriminator between physically
active and sedentary middle-aged men. J. Psychosom. Res. 1983, 27, 69–76. [CrossRef]
38.
Rudy, E.B.; Estok, P.J. Intensity of jogging: Its relationship to selected physical and psychosocial variables in
women. West. J. Nurs. Res. 1983, 5, 325–336. [CrossRef]
39.
Goldfarb, L.A.; Plante, T.G. Fear of fat in runners: An examination of the connection between anorexia
nervosa and distance running. Psychol. Rep. 1984, 55, 296. [CrossRef]
40.
Guyot, G.W.; Fairchild, L.; Nickens, J. Death concerns of runners and nonrunners. J. Sports Med. Phys. Fit.
1984, 24, 139–143.
41.
Rape, R.N. Running and Depression. Percept. Mot. Skills 1987, 64 (Suppl. 3), 1303–1310. [CrossRef]
42.
Weight, L.M.; Noakes, T.D. Is running an analog of anorexia?: A survey of the incidence of eating disorders
in female distance runners. Med. Sci. Sports Exerc. 1987, 19, 213–217. [CrossRef] [PubMed]
43.
Chan, C.S.; Grossman, H.Y. Psychological effects of running loss on consistent runners. Percept. Mot. Skills
1988, 66, 875–883. [CrossRef] [PubMed]
44.
Frazier, S.E. Mood state profiles of chronic exercisers with differing abilities. Int. J. Sport Psychol. 1988, 19,
65–71.
45.
Lobstein, D.D.; Ismail, A.H.; Rasmussen, C.L. Beta-endorphin and components of emotionality discriminate
between physically active and sedentary men. Biol. Psychiatry 1989, 26, 3–14. [CrossRef]
46.
Lobstein, D.D.; Rasmussen, C.L.; Dunphy, G.E.; Dunphy, M.J. Beta-endorphin and components of depression
as powerful discriminators between joggers and sedentary middle-aged men. J. Psychosom. Res. 1989, 33,
293–305. [CrossRef]
47.
Nouri, S.; Beer, J. Relations of moderate physical exercise to scores on hostility, aggression, and aggression,
and trait-anxiety. Percept. Mot. Skills 1989, 68 Pt 2, 1191–1194. [CrossRef]
48.
Chan, D.W.; Lai, B. Psychological aspects of long-distance running among Chinese male runners in Hong
Kong. Int. J. Psychosom. Off. Publ. Int. Psychosom. Inst. 1990, 37, 30–34.
49.
Chapman, C.L.; de Castro, J.M. Running addiction: Measurement and associated psychological characteristics.
J. Sports Med. Phys. Fit. 1990, 30, 283–290.
50.
Guyot, W.G. Psychological and medical factors associated with pain running. J. Sports Med. Phys. Fit. 1991,
31, 452–460.
Int. J. Environ. Res. Public Health 2020, 17, 8059
34 of 39
51.
Maresh, C.M.; Sheckley, B.G.; Allen, G.J.; Camaione, D.N.; Sinatra, S.T. Middle age male distance runners:
Physiological and psychological profiles. J. Sports Med. Phys. Fit. 1991, 31, 461–469.
52.
Gleaves, D.H.; Williamson, D.A.; Fuller, R.D. Bulimia nervosa symptomatology and body image disturbance
associated with distance running and weight loss. Br. J. Sports Med. 1992, 26, 157–160. [CrossRef] [PubMed]
53.
Coen, S.P.; Ogles, B.M. Psychological characteristics of the obligatory runner: A critical examination of the
anorexia analogue hypothesis. J. Sport Exerc. Psychol. 1993, 15, 338–354. [CrossRef]
54.
Furst, D.M.; Germone, K. Negative addiction in male and female runners and exercisers. Percept. Mot. Skills
1993, 77, 192–194. [CrossRef]
55.
Masters, K.S.; Ogles, B.M.; Jolton, J.A. The development of an instrument to measure motivation for marathon
running: The Motivations of Marathoners Scales (MOMS). Res. Q. Exerc. Sport 1993, 64, 134–143. [CrossRef]
56.
Pierce, E.F.; McGowan, R.W.; Lynn, T.D. Exercise dependence in relation to competitive orientation of runners.
J. Sports Med. Phys. Fitness 1993, 33, 189–193.
57.
Klock, S.C.; DeSouza, M.J. Eating disorder characteristics and psychiatric symptomatology of eumenorrheic
and amenorrheic runners. Int. J. Eat. Disord. 1995, 17, 161–166. [CrossRef]
58.
Thornton, E.W.; Scott, S.E. Motivation in the committed runner: Correlations between self-report scales and
behaviour. Health Promot. Int. 1995, 10, 177–184. [CrossRef]
59.
Powers, P.S.; Schocken, D.D.; Boyd, F.R. Comparison of habitual runners and anorexia nervosa patients.
Int. J. Eat. Disord. 1998, 23, 133–143. [CrossRef]
60.
Slay, H.A.; Hayaki, J.; Napolitano, M.A.; Brownell, K.D. Motivations for running and eating attitudes in
obligatory versus nonobligatory runners. Int. J. Eat. Disord. 1998, 23, 267–275. [CrossRef]
61.
Ryujin, D.H.; Breaux, C.; Marks, A.D. Symptoms of eating disorders among female distance runners: Can the
inconsistencies be unraveled? Women Health 1999, 30, 71–83. [CrossRef] [PubMed]
62.
Leedy, G. Commitment to Distance Running: Coping Mechanisms or Addiction. J. Sport Behav. Mob. Ala
2000, 23, 255–270.
63.
Edwards, S.D.; Ngcobo, H.S.; Edwards, D.J.; Palavar, K. Exploring the relationship between physical activity,
psychological well-being and physical self- perception in different exercise groups. South Afr. J. Res. Sport
Phys. Educ. Recreat. 2005, 27, 59–74. [CrossRef]
64.
Schnohr, P.; Kristensen, T.S.; Prescott, E.; Scharling, H. Stress and life dissatisfaction are inversely associated
with jogging and other types of physical activity in leisure time–The Copenhagen City Heart Study. Scand. J.
Med. Sci. Sports 2005, 15, 107–112. [CrossRef] [PubMed]
65.
Strachan, S.M.; Woodgate, J.; Brawley, L.R.; Tse, A. The Relationship of Self-Efficacy and Self-Identity to
Long-Term Maintenance of Vigorous Physical Activity. J. Appl. Biobehav. Res. 2005, 10, 98–112. [CrossRef]
66.
Galper, D.I.; Trivedi, M.H.; Barlow, C.E.; Dunn, A.L.; Kampert, J.B. Inverse association between physical
inactivity and mental health in men and women. Med. Sci. Sports Exerc. 2006, 38, 173–178. [CrossRef]
67.
Luszczynska, A.; Mazurkiewicz, M.; Ziegelmann, J.P.; Schwarzer, R. Recovery self-efficacy and intention
as predictors of running or jogging behavior: A cross-lagged panel analysis over a two-year period.
Psychol. Sport Exerc. 2007, 8, 247–260. [CrossRef]
68.
Smith, D.; Wright, C.; Winrow, D. Exercise dependence and social physique anxiety in competitive and
non-competitive runners. Int. J. Sport Exerc. Psychol. 2010, 8, 61–69. [CrossRef]
69.
Gapin, J.I.; Petruzzello, S.J. Athletic identity and disordered eating in obligatory and non-obligatory runners.
J. Sports Sci. 2011, 29, 1001–1010. [CrossRef]
70.
Wadas, G.; DeBeliso, M. Disordered eating, eating attitudes, and reasons for exercise among male high school
cross country runners. Sport J. 2014, 17. Available online: https://www.cabdirect.org/cabdirect/abstract/
20153072464 (accessed on 15 July 2020).
71.
Download Citation of The Relationship between Motivations, Perceived Control, and Mental Toughness
Among Marathon Runners. ResearchGate. Available online: https://www.researchgate.net/publication/
325793259_The_relationship_between_motivations_perceived_control_and_mental_toughness_among_
marathon_runners (accessed on 15 July 2020).
72.
Lucidi, F.; Pica, G.; Mallia, L.; Castrucci, E.; Manganelli, S.; Bélanger, J.J.; Pierro, A. Running away from stress:
How regulatory modes prospectively affect athletes’ stress through passion. Scand. J. Med. Sci. Sports 2016,
26, 703–711. [CrossRef] [PubMed]
Int. J. Environ. Res. Public Health 2020, 17, 8059
35 of 39
73.
Batmyagmar, D.; Kundi, M.; Ponocny-Seliger, E.; Lukas, I.; Lehrner, J.; Haslacher, H.; Winker, R. High intensity
endurance training is associated with better quality of life, but not with improved cognitive functions in
elderly marathon runners. Sci. Rep. 2019, 9, 4629. [CrossRef]
74.
Cleland, V.; Nash, M.; Sharman, M.J.; Claflin, S. Exploring the Health-Promoting Potential of the “parkrun”
Phenomenon: What Factors are Associated With Higher Levels of Participation? Am. J. Health Promot. AJHP
2019, 33, 13–23. [CrossRef]
75.
Lukács, A.; Sasvári, P.; Varga, B.; Mayer, K. Exercise addiction and its related factors in amateur runners.
J. Behav. Addict. 2019, 8, 343–349.
76.
Solmon, M.; Stewart, L. Changes in self-efficacy and affect during a 15-week marathon training program.
Int. J. Sport Psychol. 2013, 44, 55–68.
77.
Nowlis, D.P.; Greenberg, N. Empirical description of effects of exercise on mood. Percept. Mot. Skills 1979, 49,
1001–1002. [CrossRef]
78.
Wilson, V.E.; Berger, B.G.; Bird, E.I. Effects of running and of an exercise class on anxiety. Percept. Mot. Skills
1981, 53, 472–474. [CrossRef]
79.
Markoff, R.A.; Ryan, P.; Young, T. Endorphins and mood changes in long-distance running. Med. Sci.
Sports Exerc. 1982, 14, 11–15. [CrossRef] [PubMed]
80.
Thaxton, L. Physiological and Psychological Effects of Short-term Exercise Addiction on Habitual Runners.
J. Sport Exerc. Psychol. 1982, 4, 73–80. [CrossRef]
81.
McGowan, R.W.; Pierce, E.F.; Jordan, D. Mood alterations with a single bout of physical activity.
Percept. Mot. Skills 1991, 72 Pt 2, 1203–1209. [CrossRef]
82.
Goode, K.T.; Roth, D.L. Factor Analysis of Cognitions during Running: Association with Mood Change.
J. Sport Exerc. Psychol. 1993, 15, 375–389. [CrossRef]
83.
Morris, M.; Salmon, P. Qualitative and quantitative effects of running on mood. J. Sports Med. Phys. Fit. 1994,
34, 284–291.
84.
Rudolph, D.L.; Butki, B.D. Self-efficacy and affective responses to short bouts of exercise. J. Appl. Sport Psychol.
1998, 10, 268–280. [CrossRef]
85.
Cox, R.H.; Thomas, T.R.; Davis, J.E. Positive and negative affect associated with an acute bout of aerobic
exercise. J. Exerc. Physiol. Online 2001, 4, 13–20.
86.
O’Halloran, P.D.; Murphy, G.C.; Webster, K.E. Measure of beliefs about improvements in mood associated
with exercise. Psychol. Rep. 2002, 90 Pt 1, 834–840. [CrossRef]
87.
Szabo, A. The Acute Effects of Humor and Exercise on Mood and Anxiety. J. Leis. Res. 2003, 35, 152–162.
[CrossRef]
88.
O’Halloran, P.D.; Murphy, G.C.; Webster, K. Mood during a 60-minute treadmill run: Timing and type of
mood change. Int. J. Sport Psychol. 2004, 35, 309–327.
89.
Robbins, L.B.; Pender, N.J.; Ronis, D.L.; Kazanis, A.S.; Pis, M.B. Physical activity, self-efficacy, and perceived
exertion among adolescents. Res. Nurs. Health 2004, 27, 435–446. [CrossRef] [PubMed]
90.
Pretty, J.; Peacock, J.; Sellens, M.; Griffin, M. The mental and physical health outcomes of green exercise.
Int. J. Environ. Health Res. 2005, 15, 319–337. [CrossRef]
91.
Hoffman, M.D.; Hoffman, D.R. Exercisers achieve greater acute exercise-induced mood enhancement than
nonexercisers. Arch. Phys. Med. Rehabil. 2008, 89, 358–363. [CrossRef]
92.
Kwan, B.M.; Bryan, A.D. Affective response to exercise as a component of exercise motivation: Attitudes,
norms, self-efficacy, and temporal stability of intentions. Psychol. Sport Exerc. 2010, 11, 71–79. [CrossRef]
[PubMed]
93.
Weinstein, A.A.; Deuster, P.A.; Francis, J.L.; Beadling, C.; Kop, W.J. The Role of Depression in Short-Term
Mood and Fatigue Responses to Acute Exercise. Int. J. Behav. Med. 2010, 17, 51–57. [CrossRef]
94.
Anderson, R.J.; Brice, S. The mood-enhancing benefits of exercise: Memory biases augment the effect.
Psychol. Sport Exerc. 2011, 12, 79–82. [CrossRef]
95.
Kane, I.; Robertson, R.; Fertman, C.; Nagle, E.; McConnaha, W.; Rabin, B. Self-efficacy and enjoyment
of middle school children performing the Progressive Aerobic Cardiovascular Endurance Run (PACER).
Percept. Mot. Skills 2013, 117, 470–483. [CrossRef]
96.
Szabo, A.; Abrahám, J. The psychological benefits of recreational running: A field study. Psychol. Health Med.
2013, 18, 251–261. [CrossRef]
Int. J. Environ. Res. Public Health 2020, 17, 8059
36 of 39
97.
McDowell, C.P.; Campbell, M.J.; Herring, M.P. Sex-Related Differences in Mood Responses to Acute Aerobic
Exercise. Med. Sci. Sports Exerc. 2016, 48, 1798–1802. [CrossRef]
98.
Rogerson, M.; Brown, D.K.; Sandercock, G.; Wooller, J.-J.; Barton, J. A comparison of four typical green
exercise environments and prediction of psychological health outcomes. Perspect. Public Health 2016, 136,
171–180. [CrossRef]
99.
Edwards, M.K.; Rhodes, R.E.; Loprinzi, P.D. A Randomized Control Intervention Investigating the Effects of
Acute Exercise on Emotional Regulation. Am. J. Health Behav. 2017, 41, 534–543. [CrossRef]
100. Wildmann, J.; Krüger, A.; Schmole, M.; Niemann, J.; Matthaei, H. Increase of circulating beta-endorphin-like
immunoreactivity correlates with the change in feeling of pleasantness after running. Life Sci. 1986, 38,
997–1003. [CrossRef]
101. O’Connor, P.J.; Carda, R.D.; Graf, B.K. Anxiety and intense running exercise in the presence and absence of
interpersonal competition. Int. J. Sports Med. 1991, 12, 423–426. [CrossRef]
102. Nabetani, T.; Tokunaga, M. The effect of short-term (10- and 15-min) running at self-selected intensity on
mood alteration. J. Physiol. Anthropol. Appl. Hum. Sci. 2001, 20, 231–239. [CrossRef] [PubMed]
103. Bodin, M.; Hartig, T. Does the outdoor environment matter for psychological restoration gained through
running? Psychol. Sport Exerc. 2003, 4, 141–153. [CrossRef]
104. Butryn, T.M.; Furst, D.M. The effects of park and urban settings on the moods and cognitive strategies of
female runners. J. Sport Behav. 2003, 26, 335–355.
105. Kerr, J.H.; Fujiyama, H.; Sugano, A.; Okamura, T.; Chang, M.; Onouha, F. Psychological responses to
exercising in laboratory and natural environments. Psychol. Sport Exerc. 2006, 7, 345–359. [CrossRef]
106. Rose, E.A.; Parfitt, G. Exercise experience influences affective and motivational outcomes of prescribed and
self-selected intensity exercise. Scand. J. Med. Sci. Sports 2012, 22, 265–277. [CrossRef]
107. Reed, K.; Wood, C.; Barton, J.; Pretty, J.N.; Cohen, D.; Sandercock, G.R.H. A repeated measures experiment of
green exercise to improve self-esteem in UK school children. PLoS ONE 2013, 8, e69176. [CrossRef]
108. Krotee, M.L. The Effects of Various Physical Activity Situational Settings on the Anxiety Level of Children.
J. Sport Behav. Mob. Ala 1980, 3, 158–164.
109. Harte, J.L.; Eifert, G.H. The effects of running, environment, and attentional focus on athletes’ catecholamine
and cortisol levels and mood. Psychophysiology 1995, 32, 49–54. [CrossRef]
110. Berger, B.; Owen, D.; Motl, R.; Parks, L. Relationship between expectancy of psychological benefits and
mood alteration in joggers. Int. J. Sport Psychol. 1998, 29, 1–16.
111. Markowitz, S.M.; Arent, S.M. The exercise and affect relationship: Evidence for the dual-mode model and a
modified opponent process theory. J. Sport Exerc. Psychol. 2010, 32, 711–730. [CrossRef]
112. Lion, L.S. Psychological effects of jogging: A preliminary study. Percept. Mot. Skills 1978, 47 Pt 2, 1215–1218.
[CrossRef]
113. Blue, F.R. Aerobic running as a treatment for moderate depression. Percept. Mot. Skills 1979, 48, 228.
[CrossRef] [PubMed]
114. Young, R.J. The effect of regular exercise on cognitive functioning and personality. Br. J. Sports Med. 1979, 13,
110–117. [CrossRef]
115. Blumenthal, J.A.; Williams, R.S.; Needels, T.L.; Wallace, A.G. Psychological changes accompany aerobic
exercise in healthy middle-aged adults. Psychosom. Med. 1982, 44, 529–536. [CrossRef] [PubMed]
116. Trujillo, C.M. The effect of weight training and running exercise intervention programs on the self-esteem of
college women. Int. J. Sport Psychol. 1983, 14, 162–173.
117. Tuckman, B.W.; Hinkle, J.S. An experimental study of the physical and psychological effects of aerobic
exercise on schoolchildren. Health Psychol. Off. J. Div. Health Psychol. Am. Psychol. Assoc. 1986, 5, 197–207.
[CrossRef]
118. Doyne, E.J.; Ossip-Klein, D.J.; Bowman, E.D.; Osborn, K.M.; McDougall-Wilson, I.B.; Neimeyer, R.A.
Running versus weight lifting in the treatment of depression. J. Consult. Clin. Psychol. 1987, 55, 748–754.
[CrossRef] [PubMed]
119. Fremont, J.; Craighead, L.W. Aerobic exercise and cognitive therapy in the treatment of dysphoric moods.
Cogn. Ther. Res. 1987, 11, 241–251. [CrossRef]
120. Hannaford, C.P.; Harrell, E.H.; Cox, K. Psychophysiological Effects of a Running Program on Depression and
Anxiety in a Psychiatric Population. Psychol. Rec. 1988, 38, 37–48. [CrossRef]
Int. J. Environ. Res. Public Health 2020, 17, 8059
37 of 39
121. Long, B.C.; Haney, C.J. Long-Term Follow-up of Stressed Working Women: A Comparison of Aerobic
Exercise and Progressive Relaxation. J. Sport Exerc. Psychol. 1988, 10, 461–470. [CrossRef]
122. Simons, C.W.; Birkimer, J.C. An exploration of factors predicting the effects of aerobic conditioning on mood
state. J. Psychosom. Res. 1988, 32, 63–75. [CrossRef]
123. Moses, J.; Steptoe, A.; Mathews, A.; Edwards, S. The effects of exercise training on mental well-being in the
normal population: A controlled trial. J. Psychosom. Res. 1989, 33, 47–61. [CrossRef]
124. Ossip-Klein, D.J.; Doyne, E.J.; Bowman, E.D.; Osborn, K.M.; McDougall-Wilson, I.B.; Neimeyer, R.A. Effects of
running or weight lifting on self-concept in clinically depressed women. J. Consult. Clin. Psychol. 1989, 57,
158–161. [CrossRef] [PubMed]
125. Morris, M.; Steinberg, H.; Sykes, E.A.; Salmon, P. Effects of temporary withdrawal from regular running.
J. Psychosom. Res. 1990, 34, 493–500. [CrossRef]
126. Friedman, E.; Berger, B.G. Influence of gender, masculinity, and femininity on the effectiveness of three stress
reduction techniques: Jogging, relaxation response, and group interaction. J. Appl. Sport Psychol. 1991, 3,
61–86. [CrossRef]
127. Williams, T.J.; Krahenbuhl, G.S.; Morgan, D.W. Mood state and running economy in moderately trained
male runners. Med. Sci. Sports Exerc. 1991, 23, 727–731. [CrossRef]
128. Kerr, J.H.; Vlaswinkel, E.H. Self-reported mood and running under natural conditions. Work Stress 1993, 7,
161–177. [CrossRef]
129. Long, B.C. Aerobic conditioning (jogging) and stress inoculation interventions: An exploratory study of
coping. Int. J. Sport Psychol. 1993, 24, 94–109.
130. Berger, B.G.; Friedman, E. Comparison of Jogging, the Relaxation Response, and Group Interaction for Stress
Reduction. J. Sport Exerc. Psychol. 1988, 10, 431–447. [CrossRef]
131. Berger, B.G.; Owen, D.R. Relation of low and moderate intensity exercise with acute mood change in college
joggers. Percept. Mot. Skills 1998, 87, 611–621. [CrossRef]
132. Szabo, A.; Frenkl, R.; Janek, G.; Kálmán, L.; Lászay, D. Runners’ anxiety and mood on running and
non-running days: An in situ daily monitoring study. Psychol. Health Med. 1998, 3, 193–199. [CrossRef]
133. Broman-Fulks, J.J.; Berman, M.E.; Rabian, B.A.; Webster, M.J. Effects of aerobic exercise on anxiety sensitivity.
Behav. Res. Ther. 2004, 42, 125–136. [CrossRef]
134. Haffmans, P.M.J.; Kleinsman, A.C.M.; van Weelden, C.; Huijbrechts, I.P.A.M.; Hoencamp, E. Comparing running
therapy with physiotraining therapy in the treatment of mood disorders. Acta Neuropsychiatr. 2006, 18, 173–176.
[CrossRef] [PubMed]
135. Thornton, E.; Cronholm, P.; McCray, L.; Webner, D. Does Marathon Training Adversely Affect Baseline
Anxiety Levels? AMAA J. 2008, 21, 5–9.
136. Scholz, U.; Nagy, G.; Schüz, B.; Ziegelmann, J.P. The role of motivational and volitional factors for self-regulated
running training: Associations on the between- and within- person level. Br. J. Soc. Psychol. 2008, 47 Pt 3,
421–439. [CrossRef]
137. Kalak, N.; Gerber, M.; Kirov, R.; Mikoteit, T.; Yordanova, J.; Pühse, U. Daily morning running for 3 weeks
improved sleep and psychological functioning in healthy adolescents compared with controls. J. Adolesc.
Health Off. Publ. Soc. Adolesc. Med. 2012, 51, 615–622. [CrossRef]
138. Inoue, Y.; Funk, D.; Jordan, J.S. The role of running involvement in creating self-sufficiency for homeless
individuals through a community-based running program. J. Sport Manag. 2013, 27, 439–452. [CrossRef]
139. Doose, M.; Ziegenbein, M.; Hoos, O.; Reim, D.; Stengert, W.; Hoffer, N.; Vogel, C.; Ziert, Y.; Sieberer, M.
Self-selected intensity exercise in the treatment of major depression: A pragmatic RCT. Int. J. Psychiatry Clin.
Pract. 2015, 19, 266–275. [CrossRef]
140. von Haaren, B.; Haertel, S.; Stumpp, J.; Hey, S.; Ebner-Priemer, U. Reduced emotional stress reactivity to
a real-life academic examination stressor in students participating in a 20-week aerobic exercise training:
A randomised controlled trial using Ambulatory Assessment. Psychol. Sport Exerc. 2015, 20, 67–75. [CrossRef]
141. Kahan, D.; McKenzie, T.L. Physical Activity and Psychological Correlates During an After-School Running
Club. Am. J. Health Educ. 2018, 49, 113–123. [CrossRef]
142. Keating, L.E.; Becker, S.; McCabe, K.; Whattam, J.; Garrick, L.; Sassi, R.B.; Frey, B.N.; McKinnon, M.C.
Effects of a 12-week running programme in youth and adults with complex mood disorders. BMJ Open Sport
Exerc. Med. 2018, 4, e000314. [CrossRef] [PubMed]
Int. J. Environ. Res. Public Health 2020, 17, 8059
38 of 39
143. Nezlek, J.B.; Cyprya´nska, M.; Cyprya´nski, P.; Chlebosz, K.; Jenczylik, K.; Sztacha´nska, J.; Zalewska, A.M.
Within-Person Relationships Between Recreational Running and Psychological Well-Being. J. Sport Exerc.
Psychol. 2018, 40, 146–152. [CrossRef] [PubMed]
144. Kruisdijk, F.; Hopman-Rock, M.; Beekman, A.T.F.; Hendriksen, I. EFFORT-D: Results of a randomised
controlled trial testing the EFFect of running therapy on depression. BMC Psychiatry 2019, 19, 170. [CrossRef]
145. Kruisdijk, F.; Hendriksen, I.; Tak, E.; Beekman, A.-J.; Hopman-Rock, M. EFFORT-D study process evaluation:
Challenges in conducting a trial into the effects of running therapy in patients with major depressive disorder.
Ann. Gen. Psychiatry 2018, 17, 10. [CrossRef]
146. Krogh, J.; Nordentoft, M.; Sterne, J.A.C.; Lawlor, D.A. The effect of exercise in clinically depressed adults:
Systematic review and meta-analysis of randomized controlled trials. J. Clin. Psychiatry 2011, 72, 529–538.
[CrossRef]
147. Stanton, R.; Reaburn, P. Exercise and the treatment of depression: A review of the exercise program variables.
J. Sci. Med. Sport 2014, 17, 177–182. [CrossRef]
148. Lubans, D.; Richards, J.; Hillman, C.; Faulkner, G.; Beauchamp, M.; Nilsson, M.; Kelly, P.; Smith, J.; Raine, L.;
Biddle, S. Physical Activity for Cognitive and Mental Health in Youth: A Systematic Review of Mechanisms.
Pediatrics 2016, 138, e20161642. [CrossRef]
149. Greist, J.H.; Klein, M.H.; Eischens, R.R.; Faris, J.; Gurman, A.S.; Morgan, W.P. Running as treatment for
depression. Compr. Psychiatry 1979, 20, 41–54. [CrossRef]
150. Vezina, M.L.; Ruegger, R.H. THE PSYCHOLOGY OF RUNNING: Implications for nursing and health.
Nurs. Forum (Auckl.) 1980, 19, 108–121. [CrossRef] [PubMed]
151. Hinkle, J.S. Aerobic running behaviour and psychotherapeutics: Implications for sports counseling and
psychology. J. Sport Behav. 1992, 15, 263–277.
152. Weinstein, W.S.; Meyers, A.W. Running as Treatment for Depression: Is It Worth It? J. Sport Exerc. Psychol.
1983, 5, 288–301. [CrossRef]
153. Katz, J.L. Long-distance running, anorexia nervosa, and bulimia: A report of two cases. Compr. Psychiatry
1986, 27, 74–78. [CrossRef]
154. Prussin, R.A.; Harvey, P.D. Depression, dietary restraint, and binge eating in female runners. Addict. Behav.
1991, 16, 295–301. [CrossRef]
155. Yates, A.; Shisslak, C.M.; Allender, J.; Crago, M.; Leehey, K. Comparing obligatory to nonobligatory runners.
Psychosomatics 1992, 33, 180–189. [CrossRef]
156. Pierce, E.F.; Rohaly, K.A.; Fritchley, B. Sex differences on exercise dependence for men and women in a
marathon road race. Percept. Mot. Skills 1997, 84 Pt 1, 991–994. [CrossRef]
157. Yatham, L.N.; Kennedy, S.H.; Parikh, S.V.; Schaffer, A.; Bond, D.J.; Frey, B.N.; Sharma, V.; Goldstein, B.I.;
Rej, S.; Beaulieu, S.; et al. Canadian Network for Mood and Anxiety Treatments (CANMAT) and International
Society for Bipolar Disorders (ISBD) 2018 guidelines for the management of patients with bipolar disorder.
Bipolar Disord. 2018, 20, 97–170. [CrossRef]
158. Scott, J.; Pope, M. Nonadherence with mood stabilizers: Prevalence and predictors. J. Clin. Psychiatry 2002,
63, 384–390. [CrossRef]
159. Keller, M.B.; Lavori, P.W.; Mueller, T.I.; Endicott, J.; Coryell, W.; Hirschfeld, R.M.; Shea, T. Time to recovery,
chronicity, and levels of psychopathology in major depression. A 5-year prospective follow-up of 431 subjects.
Arch. Gen. Psychiatry 1992, 49, 809–816. [CrossRef]
160. Kennedy, S.H.; Lam, R.W.; McIntyre, R.S.; Tourjman, S.V.; Bhat, V.; Blier, P.; Hasnain, M.; Jollant, F.; Levitt, A.J.;
MacQueen, G.M.; et al. Canadian Network for Mood and Anxiety Treatments (CANMAT) 2016 Clinical
Guidelines for the Management of Adults with Major Depressive Disorder: Section 3. Pharmacological
Treatments. Can. J. Psychiatry Rev. Can. Psychiatry 2016, 61, 540–560. [CrossRef]
161. Thomas, J.; Thirlaway, K.; Bowes, N.; Meyers, R. Effects of combining physical activity with psychotherapy
on mental health and well-being: A systematic review. J. Affect. Disord. 2020, 265, 475–485. [CrossRef]
162. Schuch, F.B.; Vancampfort, D.; Rosenbaum, S.; Richards, J.; Ward, P.B.; Veronese, N.; Solmi, M.; Cadore, E.L.;
Stubbs, B. Exercise for depression in older adults: A meta-analysis of randomized controlled trials adjusting
for publication bias. Rev. Bras. Psiquiatr. 2016, 38, 247–254. [CrossRef] [PubMed]
Int. J. Environ. Res. Public Health 2020, 17, 8059
39 of 39
163. Children Fit for Life.
The Daily Mile.
Available online:
https://thedailymile.co.uk (accessed on
18 September 2020).
Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional
affiliations.
© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access
article distributed under the terms and conditions of the Creative Commons Attribution
(CC BY) license (http://creativecommons.org/licenses/by/4.0/).
| A Scoping Review of the Relationship between Running and Mental Health. | 11-01-2020 | Oswald, Freya,Campbell, Jennifer,Williamson, Chloë,Richards, Justin,Kelly, Paul | eng |
PMC7739736 | BRIEF RESEARCH REPORT
published: 04 September 2019
doi: 10.3389/fspor.2019.00017
Frontiers in Sports and Active Living | www.frontiersin.org
1
September 2019 | Volume 1 | Article 17
Edited by:
Giuseppe D’Antona,
University of Pavia, Italy
Reviewed by:
Lars R. McNaughton,
Edge Hill University, United Kingdom
Jeremy Coquart,
Université de Rouen, France
*Correspondence:
Chen Fleischmann
chen.fl[email protected]
Specialty section:
This article was submitted to
Exercise Physiology,
a section of the journal
Frontiers in Sports and Active Living
Received: 13 June 2019
Accepted: 16 August 2019
Published: 04 September 2019
Citation:
Fleischmann C, Horowitz M,
Yanovich R, Raz H and Heled Y (2019)
Asthaxanthin Improves Aerobic
Exercise Recovery Without Affecting
Heat Tolerance in Humans.
Front. Sports Act. Living 1:17.
doi: 10.3389/fspor.2019.00017
Asthaxanthin Improves Aerobic
Exercise Recovery Without Affecting
Heat Tolerance in Humans
Chen Fleischmann 1,2,3*, Michal Horowitz 2, Ran Yanovich 1,2,4, Hany Raz 5 and Yuval Heled 2
1 Institute of Military Physiology, IDF Medical Corps, Tel-Hashomer, Israel, 2 Heller Institute of Medical Research, Sheba
Medical Center, Ramat Gan, Israel, 3 Laboratory of Environmental Physiology, Dentistry Faculty, Hebrew University of
Jerusalem, Jerusalem, Israel, 4 The Academic College at Wingate, Wingate Institute, Netanya, Israel, 5 The Faculty of
Agriculture, Food and Environment, Hebrew University, Rechovot, Israel
Objectives: To examine the supplementation effects of the xanthophyll carotenoid
Astaxanthin on physical performance and exertional heat strain in humans.
Design: A randomized double blind placebo controlled trial.
Methods:
Twenty two male participants (Age: 23.14 ± 3.5 y, height: 175 ± 6 cm,
body mass: 69.6 ± 8.7 kg, % body fat: 16.8 ± 3.8) received placebo (PLA, n = 10)
or Astaxanthin (ATX, n = 12) 12 mg/day Per os (P.O), for 30 days, and were tested pre
and post-supplementation with a maximal oxygen uptake (VO2 Max) test and the heat
tolerance test (HTT) (2 h walk at 40◦C, 40% relative humidity (RH), 5 kph, 2% incline).
NIH database registration no. NCT02088242. Gas exchange, Heart rate (HR), Relative
perceived exertion (RPE), and blood lactate were measured during the VO2 Max test.
Heart rate (HR), rectal (Trec), and skin (Tskin) temperatures, RPE, and sweat rate (SR)
were monitored in the HTT. Serum heat shock protein 72 (HSP72), Creatine phospho-
kinase (CPK), C-reactive protein (CRP), and lipid profile were measured before and after
the test.
Results:
The rise in blood lactate caused by the VO2 Max test was significantly
diminished in the ATX group (9.4 ± 3.1 and 13.0 ± 3.1 mmole∗l−1 in the ATX and PLA
groups, respectively P < 0.02), as was the change in oxygen uptake during recovery
(−2.02 ± 0.64 and 0.83 ± 0.79% of VO2 Max in the ATX and PLA group, respectively,
p = 0.001). No significant differences were observed in the anaerobic threshold or VO2
Max. In the HTT, no significant physiological or biochemical differences were observed
(HR <120 bpm, Trec rose by ∼1◦C to <38◦C, no difference in SR).
Conclusions: Astaxanthin supplementation improved exercise recovery. No benefit was
observed for ATX over PLA in response to heat stress. Further examination of Astaxanthin
in higher exertional heat strain is required.
Keywords: astaxanthin, supplementation, exercise nutritional physiology, aerobic exercise, exercise-recovery,
heat tolerance
Fleischmann et al.
Astaxanthin Exercise Recovery and Heat Tolerance
INTRODUCTION
Astaxanthin is a xanthophyll carotenoid food supplement
prevalent in marine organisms (Kidd, 2011). This potent
antioxidant (Kidd, 2011) affects the Insulin\Insulin growth
factor I (IGF1) and the nuclear kinase mitogen and stress-
activated protein kinase-1 (MSK1) signaling pathways, which
were found to be implicated in preconditioning, survival and
longevity in vitro in human keratinocytes (Terazawa et al.,
2012) and in vivo in Caenorhabditis elegans (Yazaki et al., 2011)
Other in vivo experiments in animals have shown Astaxanthin
is associated with reductions in C reactive protein and DNA
damage and improvement of the cell-mediated and humoral
immune responses (Park et al., 2011) and with improvement
in cardiovascular parameters (Fassett and Coombes, 2012). In
exercising mice Astaxanthin induced diminished fatigue, and
reductions in blood lactate, oxidative damage to lipids and
DNA, and muscle injury (Aoi et al., 2003; Ikeuchi et al.,
2006). However, exercise experiments in humans were equivocal,
showing improved endurance as time trial performance in
competitive cyclists (Earnest et al., 2011), vs. no significant
improvement in well-trained cyclists (Res et al., 2013) and soccer
players (Djordjevic et al., 2012). During exercise, some evidence
from animal experiments supports enhanced fat utilization over
carbohydrates (Ikeuchi et al., 2006; Aoi et al., 2008), yet no
supplementation effect in endurance exercise and recovery was
established (Brown et al., 2018).
Exertional heat injury is a life threatening condition inflicting
many young, healthy individuals, commonly affecting highly
motivated, physically active populations such as military
personnel and athletes (Carter et al., 2005; Casa et al., 2005). Risk
reduction of heat injury includes avoidance of strenuous physical
activity in severe heat load conditions, application of heat
acclimation protocols, and the use of external cooling methods
(Epstein et al., 2000). Preparation for planned military and
athletic activities could potentially enhance resilience to extreme
physical and environmental conditions, reduce the chance of
heat injury and improve injury response and recovery. Several
exogenous agents have been studied either as prophylactic to heat
stress exposure or as post injury treatment yet none were effective
(Moran et al., 1999; Kuennen et al., 2011).
Heat
load
is
a
significant
stressor
during
exercise.
Astaxanthin’s activity against stressor induced generation of
reactive oxygen and nitrogen species (RONS) and inflammatory
cytokines (Brown et al., 2018) may thus be beneficial in heat
stress conditions. Accordingly, Do et al. demonstrated that
Abbreviations: AT, Anaerobic threshold; ATX, Astaxanthin; BMI, Body mass
index; BSA, Body surface area; CPET, Cardio pulmonary exercise test; CPK,
Creatine phospho kinase; CRP, C- reactive protein; ELISA, Enzyme-Linked
Immunosorbent Assay; FDA, U.S. Food and Drug Administration; GI, Gastro
intestinal; HR, Heart rate; HSP72, Heat shock protein 72; HSR, Heat shock
response; HTT, Heat tolerance test; IGF-1, Insulin-like growth factor 1; MAP,
Mean arterial pressure; MPO, Myeloperoxidase; MSK-1, Mitogen- and stress-
activated protein kinase-1; OBLA, Onset of blood lactate accumulation; OD,
Optical density; PLA, Placebo; PP, Pulse pressure; RER, Respiratory exchange ratio;
RH, relative humidity; RONS, Reactive oxygen and nitrogen species; ROS, Reactive
oxygen species; RPE, Relative perceived exertion; SR, Sweat rate; TCR, Thermal
comfort rate; VO2, Oxygen uptake; VO2 Max, Maximal oxygen uptake.
during development, treatment with Astaxanthin increased
protection of porcine oocytes against heat shock, along with
increased resilience to oxidative stress (Do et al., 2015). In
rodents, Astaxanthin enhanced protection against heat related
damages and oxidative stress (Preuss et al., 2009) and resilience
against heat stress combined with gravitational unloading
(Yoshihara et al., 2018). In a preliminary experiment our
group demonstrated improved heat tolerance with elevated
cardiac tissue concentration of HSP72 protein and HSP70
mRNA in rats (Horowitz M. & Abbas A., unpublished data).
In yellow catfish, Astaxanthin pretreatment improved overall
stress resistance, while elevating hepatic heat shock protein 70
(HSP70) mRNA levels, with increased antioxidant capacity,
and decreased expression of the stress related hormone cortisol
and glucose levels (Liu et al., 2016). Pufferfish fed with a diet
containing Astaxanthin produced less reactive oxygen species
(ROS) when exposed to heat stress, and increased production of
superoxide dismutase (SOD), catalase (CAT), and HSP70 mRNA
under high temperature stress, in comparison with the control
(Cheng et al., 2018). Elevation of HSP70 mRNA and HSP72
protein is an important part of the heat shock response (HSR),
representing innate cellular defense mechanisms against heat
related damage (Horowitz, 1998). Overall, across several animal
models, Astaxanthin treatment enhances cellular protection
against heat with corresponding increased levels of HSP70,
possibly by priming key components of the HSR for activation,
and by acting as a potent antioxidant via protection against the
heat stress induced generation of RONS.
Based on the aforementioned knowledge, and in the absence
of known safe substances applicable as preemptive measures
for anticipated heat stress exposure, Astaxanthin emerges as
a potential candidate for enhancing heat resilience through
increased cellular protection, pertinent to both heat and exercise
exposure, as it is a safe food supplement, which may potentially
be consumed chronically without adversely affecting active
populations. Therefore, we set out to determine whether
Astaxanthin supplementation, as a preemptive strategy, could
have an influence on performance in heat stress combined
with exercise scenarios in humans, and potentially serve as an
added line of defense against heat related injury for individuals
anticipating exposure to heat and exercise. We also chose to
separately evaluate the influence of Astaxanthin supplementation
on aerobic fitness, since it is a key contributing component to
endurance in the heat (Mclellan et al., 2012) and to determine
whether the potential added cellular protection might influence
aerobic performance, independently of heat exposure.
The study goals were to determine whether Astaxanthin pre-
supplementation could influence performance in exercise alone
or in combination with heat stress.
METHODS
In
order
to
evaluate
the
influence
of
Astaxanthin
supplementation on heat tolerance and on aerobic capacity,
we employed a double blind placebo controlled randomized
trial. The heat tolerance test (HTT), which involves exposure
Frontiers in Sports and Active Living | www.frontiersin.org
2
September 2019 | Volume 1 | Article 17
Fleischmann et al.
Astaxanthin Exercise Recovery and Heat Tolerance
to mild physical activity in controlled heat load conditions
and the maximal oxygen uptake (VO2 Max) test were used
before supplementation and repeated after 1 month of daily
supplementation. The study was approved by the ethical review
boards of the Sheba medical center (reg. no. 1295-13) and of
the IDF Medical Corps (reg. no. 0471-13) and was registered in
the NIH database (reg. no. NCT02088242). Data collection took
place between March of 2015 and March of 2016 at the Heller
institute of medical research located in the Sheba medical center,
Tel-Hashomer, Israel.
Participants
Twenty two young healthy male volunteers, free from illness and
not consuming medications or dietary supplements, completed
their participation in the study after giving their informed
consent and being examined by the study physician. Participants
were interviewed by a nutritionist to ensure an Astaxanthin
free diet and instructed to avoid changing their exercise
routine for the duration of the study, and refrain from
consuming Astaxanthin containing foods, as well as any dietary
supplements for 2 weeks prior to participating in the physical
tests and throughout the duration of the study. They were
randomly assigned to either the supplementation group, who
received 12 mg of Astaxanthin P.O daily as 3 soft gel capsules
of Astapure R⃝ (10% Oleoresin) 4 mg or a placebo identical
in appearance and taste, which contained no Astaxanthin
(Algatech, Ktora, Israel). The Supplement and placebo capsules
were purchased directly from the manufacturer, to guarantee
production of a placebo identical in every way to the supplement,
apart from the presence of the active ingredient. Certificates
of analysis were issued for each purchased batch, ensuring a
95% purity at least of the active ingredient (Astaxanthin) in the
oleoresin contained in the soft gel capsules.
Treatment
The dose (12 mg) was chosen in accordance with the highest
daily dose approved for human consumption by the U.S. Food
and Drug Administration (FDA) at the time of study approval
and with literature evidence from human experimentation,
demonstrating safety and efficacy at this and higher doses
(Kupcinskas et al., 2008; Yoshida et al., 2010; Choi et al., 2011;
Nakagawa et al., 2011). Supplementation duration (over 30 days)
was chosen in order to ensure adequate time for achieving a
supplemented state and initiating the necessary long term effects,
based on other human experiments involving exercise related
aspects, without any known threat to the subjects’ health and
well-being (Spiller and Dewell, 2003; Bloomer et al., 2005; Earnest
et al., 2011; Miyazawa et al., 2011).
Randomization and assignment to the Astaxanthin or placebo
group was performed by an independent party (the clinical
research division of the Sheba medical center pharmaceutical
services), which also individually dispensed the study product
to the participants. Treatment allocation was disclosed to the
researchers only after study completion. In order to ensure
maximal gastro-intestinal (GI) absorption, participants were
instructed to ingest the supplement or placebo with a meal
containing 15 grams of fat. Supplementation lasted for 30
days, immediately followed by an additional supplementation
period of 5–10 days, during which the physical tests (HTT
and VO2 Max) were repeated, on separate days, in-order to
maintain an effective concentration of the supplement and
ensure the tests were performed under a supplemented state.
Treatment adherence by participants was monitored by keeping
a supplementation log and sending a daily text message after
supplement consumption. A dietary log was also kept for 3 days
before each physical examination day.
Experimental Design and Procedures
Twenty two participants completed the study, after being
randomly assigned to either the Astaxanthin (ATX, n = 12,
age: 22.3 ± 4.0 years) or placebo (PLA, n = 10, age: 24.1 ±
2.60) groups in a double blind manner. Participants in both
groups were of average anthropometrics (Height=173.95 ±
4.0 cm, and 1.75 ± 7.6 cm; Body mass = 68.46 ± 8.0, and
70.96 ± 9.8; BMI = 22.6 ± 2.33, and 23.02 ± 2.40; %body
fat = 13.32 ± 4.15% vs. 17.33 ± 3.41%, in the ATX and PLA
groups, respectively, no significant difference between treatment
groups). Supplementation began after completion of the initial
HTT and VO2 Max tests, and lasted a total of 35–40 days.
The HTT and VO2 Max tests where repeated after 30 days
under ongoing supplementation. Figure 1 is a flow diagram
of the study, detailing the process of participant recruitment,
assignment and testing.
Anthropometric measurement (height, body mass, body fat
from a four points skinfold measurement) was followed by
evaluation of aerobic capacity and heat tolerance which were
conducted on separate days, at least 48 h apart, and followed by
commencement of daily supplementation. Aerobic capacity and
heat tolerance assessment were repeated during the 31–40 day
period of supplementation, while still consuming the supplement
or placebo.
Anthropometry included height (roll-up stadiometer, model
206, Seca medical measuring systems and scales, Germany), body
mass (electronic scales), and determination of body composition
by the four sites (biceps brachii, triceps brachii, suprailiac,
subscapular) skinfold measurement (Lange skinfold caliper, Beta
technology, Santa Cruz, CA) and calculation of fat content and
lean body mass, based on an equation suited to the participant’s
age (Durnin and Womersley, 1974).
The heat tolerance test (HTT) was described by Moran et al.
(2007). Participants were dressed in shorts and tennis shoes and
exposed to 2 h of extreme heat stress (40◦C, 40% RH) in a climatic
chamber, while walking on a motor-driven treadmill (5 kph,
2% incline). Rectal temp. (Trec), skin temp. (Tsk), and heart
rate (HR) were continuously monitored. Fluid consumption
(cold water) was provided ad-libitum from pre-weighed drinking
cans. Trec was measured with a rectal thermistor (YSI-401,
Yellow Springs Incorporated, USA) inserted 10 cm beyond the
anal sphincter. Skin temp. (Tsk) at the chest, upper arm and
calf, was measured using a skin thermistor (YSI-409B, Yellow
Springs Incorporated, USA). Mean Tsk was calculated by Burton’s
equation (Burton, 1935). All temperatures were continuously
recorded (MP150 and Acqknowledge software, version 3.9,
Biopac systems, USA). Heart rate (HR) was continuously
Frontiers in Sports and Active Living | www.frontiersin.org
3
September 2019 | Volume 1 | Article 17
Fleischmann et al.
Astaxanthin Exercise Recovery and Heat Tolerance
FIGURE 1 | Study flow diagram.
monitored by a heart rate monitor (model: RS800CX, POLAR,
Finland). Blood pressure (BP) was monitored at pre-set time
points (before the test, after 1 h of walking, at the end of
the test, and every 15 min during recovery, for 1 h after the
test) using an automated blood pressure monitor (Omron
m6 comfort, Omron healthcare, Japan). Fluid balance was
determined from nude body mass, measured before and after
each trial, adjusted for fluid intake and urine volume, and
used to calculate sweat loss, which was then normalized to
body surface area and presented as the hourly sweat rate
(SR). Relative perceived exertion (RPE) was assessed every
15 min during the HTT using the Borg scale (Borg, 1998),
and a scale from 1 to 13 (unbearably cold to unbearably
hot sensation, respectively), was used to rate the subjective
sensation of thermal comfort (Thermal comfort rate, TCR).
Safety thresholds for test cessation were set at Trec = 39◦C or
Frontiers in Sports and Active Living | www.frontiersin.org
4
September 2019 | Volume 1 | Article 17
Fleischmann et al.
Astaxanthin Exercise Recovery and Heat Tolerance
HR = 180 bpm, at the study physician’s discretion or at the
participant’s request.
Maximal oxygen uptake (VO2 Max) was determined by cardio
pulmonary exercise testing (CPET), using a modified Bruce
protocol composed of 5 min seated rest, followed by 5 min of
walking on a treadmill at 5 kph, and 2% incline, followed by
running at 9 kph, with an incrementally increasing incline (2%
every 2 min), until reaching VO2 Max, which was determined by
3 of 4 criteria during the test: (1) leveling off of the VO2 curve
to a plateau, (2) reaching >90% of the participant’s predicted
maximal heart rate (210–0.65 × Age), (3) reaching a respiratory
exchange ratio (RER) ≥1.1 or 4) at the participant’s request,
after reaching a subjective state of extreme physical tiredness.
Additional supportive indications after test completion were
reaching >8 mmol/l of blood lactate, or RPE > 17 (Edvardsen
et al., 2014; Debeaumont et al., 2016). The ventilatory anaerobic
threshold (AT) was determined visually by two trained examiners
according to the American heart association guidelines (Balady
et al., 2010). Continuous monitoring lasted throughout recovery,
which consisted of 3 min at 5 kph and 2% incline, followed by
3 kph at 0% incline, and finally, 1 min, seated. The test was
performed on a CPET machine (ZAN 600, Nspire Health, USA)
connected to a treadmill ergometer (Model 770 S, RAM medical
and industrial instruments, Germany). Reaching onset of blood
lactate accumulation (OBLA) was confirmed by examining blood
lactate level before and after the test (lactate scout+ analyzer,
Sports Resource Group Inc., USA). Heart rate was continuously
monitored by a heart rate monitor (model: RS800CX, POLAR,
Finland). Assessment of RPE took place before and after the VO2
Max test.
Blood was drawn on physical testing days before the VO2 Max
test and on HTT days before, immediately after and at 60 min
after the end of the HTT. Blood was collected in yellow gel
chemistry collection tubes (Becton, Dickinson and Co., NJ, USA),
allowed to clot for 30 min and centrifuged. Serum was separated
immediately and stored at −80◦C pending analysis. Serum lipid
and triglyceride (TG) profile, CRP, and CPK were analyzed by the
central laboratories at the Sheba medical center. A commercially
available ELISA kit for High-Sensitivity HSP72 detection was
used to measure serum HSP72 levels in optical density (OD),
which was used to calculate the HSP72 concentration in ng/ml,
according to the manufacturer’s instructions (AMP’D R⃝ HSP70
high sensitivity ELISA kit, ENZ-KIT-101, Enzo life sciences,
NY, USA).
STATISTICS
Anthropometric, physiological and biochemical parameters were
statistically analyzed using the SPSS software (version 23, IBM,
USA). Treatments and time point were taken as the independent
variable and participants were considered a random sample of the
general population. Normality of distribution was assessed by the
Kolomogorov-Smirnov test and comparison between treatment
groups and between pre- and post-supplementation time points
was made with 1-way ANOVA, with Tukey post hoc analysis
for normally distributing variables, or Mann-Whitney U-test for
non-normally distributing variables. Analysis of the difference
in the change in parameters due to supplementation between
treatment groups was conducted by calculating the delta between
the pre- and post-supplementation states (pre-supplemented
state subtracted from the post-supplemented state). Normality
of distribution was assessed by the Kolomogorov-Smirnov test
and comparison between treatment groups was made by T-test
for normally distributing variables and Mann-Whitney U-test
for non-normally distributing variables. Leven’s test was used
to evaluate the equality of variance between treatment groups,
followed by the appropriate Student’s t-test (2-tailed) to assess
significance. In order to assess the significance of difference
between repetitive blood tests, ANOVA for repeated measures
followed by Bonferroni post-hoc analysis or Friedman’s omnibus
test followed by Wilcoxon’s signed-rank test with Bonferroni
adjustment were used, for normally or non-normally distributing
variables, respectively. A significant p-value was set at 0.05.
RESULTS
Table 1 lists key parameters of aerobic capacity, as recorded by
the VO2 Max test. In both groups, anaerobic threshold (AT)
was achieved at approximately 72% of the VO2 Max value, VO2
Max was similar and did not improve post-supplementation.
Aerobic characteristics did not differ between the ATX and PLA
groups both before and after supplementation, as seen in the
unchanged AT, maximal oxygen uptake, reduction in heart rate
during recovery, and in substrate utilization demonstrated by the
scatter plot of respiratory exchange ratio (RER) vs. oxygen uptake
(VO2) (Supplemental Figure 1).
However, a significant difference was observed between
the two groups post supplementation in the blood lactate
concentration measured after the VO2 Max test. Additionally,
a significant reduction was observed in oxygen uptake at the
end of recovery between the pre-supplementation and post-
supplementation time points in the ATX group compared to the
PLA group (Table 1). Supplemental Figure 2 depicts the VO2
values during the test by group, before and after supplementation.
Table 2 lists the results from the HTT. The physiological
parameters monitored continuously during the test, including
HR, Trec, and Tsk displayed no significant difference between
the ATX and PLA groups. During the first, un-supplemented
HTT, and the second, supplemented HTT, Basal Trec in both
groups was below 37◦C, and increased by about 1◦C. Heart rate
began at nearly 80 bpm and increased to just under 120 bpm
in both groups.
Pre-supplementation sweat rate in the PLA group was
significantly higher than the ATX group (which disappeared
post-supplementation), and post-supplementation in the PLA
group (p < 0.001). The subjective scales representing sensations
of relative perceived exertion (RPE) and thermal comfort (TCR),
which were monitored every 15 min during the test and for 1 h
after its completion, also displayed no difference between groups
or exposures. Participants perceived a mild to moderate effort in
reporting their subjective sensations in the Borg scale (RPE) and
moderate heat in the TCR scale.
Frontiers in Sports and Active Living | www.frontiersin.org
5
September 2019 | Volume 1 | Article 17
Fleischmann et al.
Astaxanthin Exercise Recovery and Heat Tolerance
TABLE 1 | Main VO2 Max findings: This table lists the main findings from the maximal oxygen uptake tests performed before (pre) and after (post) supplementation in the two study groups.
Test parameter
ATX pre
ATX post
PLA pre
PLA post
ANOVA/Mann-Whitney U
Delta ATX
Delta PLA
T-test/
Mann-Whitney U
Mean ± St. Error
Mean ± St. Error
Mean ± St. Error
Mean ± St. Error
p-value
Post hoc
Tukey p-value
Mean ± St. Error
Mean ± St. Error
p-value
RPE Before
6 ± 0
7 ± 0
7 ± 1
7 ± 0
N.S
0 ± 1
−2 ± 1
N.S
RPE After
18 ± 1
17 ± 0
16 ± 1
17 ± 0
N.S
−2 ± 2
−3 ± 3
N.S
1 RPE
12 ± 1
11 ± 0
9 ± 1
11 ± 1
N.S
−2 ± 1
−1 ± 2
N.S
BLA before (mmole.l−1)
2.3 ± 0.31
2.57 ± 0.21
2.04 ± 0.24
2.08 ± 0.2
N.S
0.27 ± 0.37
−0.17 ± 0.38
N.S
BLA after (mmole.l−1)
13.46 ± 0.94
11.92 ± 0.85
12.65 ± 0.68
15.09 ± 1.02
N.S
−1.54 ± 0.85
0.93 ± 1.84
N.S
1 BLA (mmole.l−1)
11.16 ± 0.78
9.35 ± 0.89
10.61 ± 0.73
13.01 ± 1.05
0.044
0.027*
−1.81 ± 0.89
1.1 ± 1.67
N.S
Sys. BP before (mmHg)
119 ± 2.77
120.64 ± 3.96
118.5 ± 2.87
121.29 ± 2.9
N.S
−8.42 ± 10.34
−33.6 ± 19.81
N.S
Dias. BP before (mmHg)
78.58 ± 3.14
77.64 ± 3.65
72.7 ± 3.09
79.57 ± 5.07
N.S
−7.42 ± 8.94
−17 ± 14.49
N.S
Sys. BP after (mmHg)
118.9 ± 4.42
127.2 ± 3.84
122.9 ± 5.34
117.71 ± 5.51
N.S
6.92 ± 21.73
−40.5 ± 17.89
N.S
Dias. BP after (mmHg)
75.1 ± 2.93
79.4 ± 3.11
76 ± 3.24
73.29 ± 4.36
N.S
3.58 ± 14.96
−24.7 ± 11.12
N.S
AT VO2 (ml.kg−1.min−1)
37.51 ± 3.35
40.1 ± 1.97
37.27 ± 3.51
38.53 ± 3.98
N.S
2.59 ± 3.68
1.26 ± 2.09
N.S
AT (% of VO2 Max)
72.32 ± 6.08
78.27 ± 3.17
71.3 ± 3.77
73.23 ± 4.51
N.S
5.95 ± 6.86
1.94 ± 3.79
N.S
Max load (Watt)
284.67 ± 11.46
288.5 ± 12.66
276.7 ± 19.6
281.1 ± 19.06
N.S
3.83 ± 5.92
4.4 ± 2.61
N.S
VO2 Max (ml.kg−1.min−1)
52.55 ± 1.83
51.24 ± 1.6
51.49 ± 2.32
51.54 ± 2.66
N.S
−1.31 ± 0.7
0.05 ± 0.71
N.S
HR Max (bpm)
190.67 ± 2.85
191.17 ± 2.33
191.2 ± 2.62
190.1 ± 2.85
N.S
0.5 ± 1.58
−1.1 ± 1.42
N.S
RER Max
1.14 ± 0.02
1.13 ± 0.02
1.15 ± 0.02
1.15 ± 0.03
N.S
0 ± 0.02
0 ± 0.02
N.S
#End recov. VO2 (ml*kg−1*min−1)
12.08 ± 0.32
10.76 ± 0.45
11.43 ± 0.43
11.85 ± 0.51
N.S
−2.22 ± 0.99
0.42 ± 0.48
##0.006
#End recov. VO2 (% of VO2 Max)
23.52 ± 0.64
21.31 ± 0.68
22.44 ± 0.98
23.27 ± 1.04
N.S
−2.02 ± 0.64
0.83 ± 0.79
0.01
1#end recov. HR (bpm)
47.25 ± 2
43.08 ± 2.24
40.3 ± 4
40.3 ± 4.7
N.S
−4.17 ± 2.15
0 ± 5.19
N.S
RPE, relative perceived exertion; BLA, blood lactate; Sys., systolic; Dias., diastolic; BP, blood pressure; AT, Anaerobic threshold; Max, maximal; HR, heart rate; bpm, beats per minute; recov., recovery.
#End recovery is the average value of the last 30 s recorded while seated at the end of the test.
*Between ATX post and PLA post.
##Value calculated with Mann-Whitney U-test, for non-normally distributing variables.
Frontiers in Sports and Active Living | www.frontiersin.org
6
September 2019 | Volume 1 | Article 17
Fleischmann et al.
Astaxanthin Exercise Recovery and Heat Tolerance
TABLE 2 | Main HTT findings: This table lists the main findings from the HTT performed before (pre) and after (post) supplementation in both treatment groups.
Test parameter
ATX pre
ATX post
PLA pre
PLA post
ANOVA/Mann-Whitney U
Delta ATX
Delta PLA
T-test/
Mann-Whitney U
Mean ± St. Error
Mean ± St. Error
Mean ± St. Error
Mean ± St. Error
p-value
post hoc Tukey p-value
Mean ± St. Error
Mean ± St. Error
p-value
Basal Trec (◦C)
36.93 ± 0.08
36.87 ± 0.08
36.95 ± 0.07
36.82 ± 0.12
N.S
−3.13 ± 3.04
−0.12 ± 0.12
N.S
Max Trec (◦C)
37.79 ± 0.1
37.9 ± 0.1
37.79 ± 0.1
37.91 ± 0.17
N.S
−3.05 ± 3.12
0.12 ± 0.16
N.S
1 Trec (◦C)
0.83 ± 0.09
0.93 ± 0.12
0.9 ± 0.1
1.1 ± 0.18
N.S
0.1 ± 0.09
0.2 ± 0.2
N.S
End + 1 h Trec (◦C)
37.29 ± 0.06
37.26 ± 0.13
37.26 ± 0.07
37.19 ± 0.14
N.S
3.07 ± 3.13
7.38 ± 4.97
N.S
Basal Tskin (◦C)
35.41 ± 0.21
35.12 ± 0.24
35.68 ± 0.24
35.49 ± 0.24
N.S
−0.26 ± 0.2
−7.29 ± 4.76
N.S
Max Tskin (◦C)
36.4 ± 0.14
36.43 ± 0.23
36.12 ± 0.17
36.28 ± 0.29
N.S
−3 ± 3.01
−7.1 ± 4.83
N.S
1 Tskin (◦C)
0.99 ± 0.22
−2 ± 3.41
0.4 ± 0.27
0.67 ± 0.37
N.S
−2.74 ± 3.07
0.2 ± 0.36
N.S
Sweat rate*BSA−1
−342.81 ± 23.25
−334.17 ± 22.53
−472.89 ± 29.72
−333 ± 19.71
<0.001
0.002*
8.64 ± 36.11
139.89 ± 39.16
0.023
HR Start (bpm)
79.17 ± 3.29
79.67 ± 3.17
80.4 ± 4.37
79.1 ± 3.84
N.S
0.5 ± 2.91
−1.3 ± 2.78
N.S
HR end (bpm)
118 ± 3.84
116.08 ± 4.71
117.7 ± 6.38
118.9 ± 5.47
N.S
−1.92 ± 4.15
1.2 ± 3.57
N.S
1 HR (bpm)
38.83 ± 2.29
36.42 ± 5.09
37.3 ± 5.11
39.8 ± 3.68
N.S
−2.42 ± 5.32
2.5 ± 3.42
N.S
RPE start
8 ± 1
7 ± 0
7 ± 0
7 ± 0
N.S
0 ± 1
0 ± 1
N.S
RPE end
11 ± 1
8 ± 0
9 ± 1
10 ± 2
N.S
−1 ±
0 ± 2
N.S
TCR start
9 ± 1
7 ± 1
8 ± 0
8 ± 1
N.S
−1 ± 1
1 ± 1
N.S
TCR end
11 ± 1
11 ± 1
9 ± 1
9 ± 0
N.S
0 ±
0 ± 1
N.S
MAP Start (mmHg)
83.78 ± 2.72
89.39 ± 5.03
80.57 ± 9.23
83.73 ± 2.36
N.S
5.61 ± 5.19
3.17 ± 8.96
N.S
MAP end (mmHg)
81.33 ± 2.36
80.36 ± 2.96
83.8 ± 3.69
80.97 ± 2.01
N.S
1.27 ± 2.92
−2.83 ± 4.7
N.S
1 MAP (mmHg)
−9.22 ± 6.94
−9.03 ± 5.13
3.23 ± 11.59
−2.77 ± 2.4
N.S
0.19 ± 7.92
−6 ± 10.89
N.S
MAP HTT + 1 h (mmHg)
83.48 ± 2.47
90.42 ± 1.23
87.6 ± 4.01
85.17 ± 2.24
N.S
6.67 ± 2.41
−2.43 ± 4.46
N.S
Max, maximal; Trec, rectal temp.; Tskin, skin temp.; BSA, body surface area; HR, heart rate; bpm, beats per minute; RPE, relative perceived exertion; TCR, thermal comfort rate; MAP, Mean arterial pressure. *Between PLA pre and ATX
pre and between PLA pre and PLA post.
Frontiers in Sports and Active Living | www.frontiersin.org
7
September 2019 | Volume 1 | Article 17
Fleischmann et al.
Astaxanthin Exercise Recovery and Heat Tolerance
Biochemical analyses:
Table 3 depicts measured
serum
concentrations of CRP, CPK, HSP72, and the lipid profile,
including,
high
density
lipoproteins
(HDL),
low
density
lipoproteins
LDL
total
cholesterol
and
Triglycerides.
No
significant differences were observed between the ATX and
PLA groups in the serum levels of HSP72 protein, in the lipid
and triglyceride profile, in CRP or in CPK concentrations, both
before and after the effort. However, during all HTT testing days,
CPK levels obtained before the test were significantly lower than
those obtained immediately after the test, in both groups, both
before and after supplementation.
DISCUSSION
We examined the influence of 1 month of 12 mg daily
Astaxanthin supplementation on heat tolerance and aerobic
capacity. Astaxanthin improved exercise recovery but had no
influence on performance in the heat.
Human exercise models, in contrast to animal studies have
shown conflicting results regarding the effects of Astaxanthin on
performance. For example: the beneficial effects of Astaxanthin in
competitive cyclists shown while consuming 4 mg/day (Earnest
et al., 2011), vs. no significant difference in performance of
well-trained cyclists while consuming 20 mg/day (Res et al.,
2013). Neither metabolic markers nor blood biochemistry of
human cohorts revealed dose or time dependent metabolic
changes attributable to Astaxanthin supplementation (Karppi
et al., 2007; Earnest et al., 2011; Res et al., 2013). The variance of
substrate oxidation profiles during exercise existing in the general
population and the steady state nature of the measurement
may have masked a metabolic supplementation effect. The
graded VO2 Max test used in our study, designed to answer
the questions raised regarding the influence of Astaxanthin on
substrate utilization in exercising humans over a range of exercise
intensities (Brown et al., 2018), showed no effect on aerobic
capacity or its components: energy substrate use during the VO2
Max test displayed no supplementation effect to influence fat
utilization over carbohydrates in either group, as demonstrated
in Supplemental Figure 1.
However, the change in blood lactate concentration after the
VO2 Max test (Table 1), along with the significant reduction
in oxygen uptake at the end of recovery in the ATX group
compared to the PLA group, may suggest less oxidative stress
and faster recovery in comparison with the control, which
is a possible advantage for Astaxanthin supplementation. In
Supplemental Figure 2, a more rapid return to lower VO2 values
during recovery is seen in ATX after supplementation compared
to before supplementation.
Though evidence from animal models suggests that post
exercise recovery may improve with Astaxanthin administration,
particularly, by diminishing exercise induced tissue damage
markers such as creatine kinase (CK) and myeloperoxidase
(MPO), through anti-oxidative and anti-inflammatory pathways
(Aoi et al., 2003; Guo et al., 2018), human studies are ambiguous:
muscle soreness, exercise force production and plasma CK
displayed no significant difference in highly trained individuals
who received 3 weeks of 4 mg/day Astaxanthin (Bloomer
et al., 2005). However, longer supplementation (90 days) in
young soccer players was associated with improved indirect
damage markers like reduced lactate dehydrogenase (LDH),
and non-significant improvements in CK and inflammatory
markers including CRP and leukocyte and neutrophil counts
(Djordjevic et al., 2012). Validated information on the effects
of Astaxanthin supplementation on exercise performance and
recovery, particularly in diverse populations, is lacking.
In the present experiment, though exercise recovery of
oxygen uptake was improved in the Astaxanthin group post-
supplementation, contrastingly, serum inflammation (CRP),
muscle damage (CPK) and lipid profile remained unaffected by
supplementation in both groups.
The physiological strain induced by the HTT, was mild for
both groups (Trec < 38◦C and HR < 120 bpm), as supported
by the lack of change in HSP72 post-exercise, pointing to
an insufficient perturbation of the thermoregulatory system
and the absence of an HSR. Notably, experimental conditions,
particularly the physiological safety thresholds, were limited
by ethical constraints, and could not induce a higher thermal
threshold. Under the experimental conditions employed in
this study, no participant reached the safety threshold during
heat exposure.
An Additional component contributing to the observed
physiological response may have been the fitness level of
participants and the relatively mild effort undertaken by them
during the HTT. An average VO2 Max of 51–52 ml × kg−1 ×
min−1 was typical of the study participants. The average HR
elevation during the HTT was ∼40 bpm, reflecting a 21% change
relative to the measured maximal HR in the VO2 Max test
(Table 1), indicating a state of mild stress experienced during the
HTT across treatment groups and exposures.
Nevertheless, the significant difference in CPK levels from the
beginning to the end of the HTT, in both groups indicates some
muscle damage resulting from the HTT, which was unaffected by
supplementation (Table 3).
The significantly higher sweat rate in the pre-supplemented
PLA group compared to PLA post-supplementation and to ATX
pre- and post-supplementation cannot be explained by an effect
of supplementation, and can only be attributed to a difference
between participant groups. This was, however, insignificant
when the change in sweat rate from pre- to post-supplementation
was compared between treatment groups (Table 2).
The daily dose of Astaxanthin used in this work (12 mg) was
reflective of the highest recommended dose for humans at the
time, which has been substantially increased since then to 24 mg
daily (Visioli and Artaria, 2017). Consumption of a larger dose
may have evoked greater effects in aerobic function and cellular
protective aspects important to coping with the damages of heat
stress exposure.
CONCLUSION
Preemptive nutritional supplementation is a promising avenue
for exercise science research as a way of improving physiological
Frontiers in Sports and Active Living | www.frontiersin.org
8
September 2019 | Volume 1 | Article 17
Fleischmann et al.
Astaxanthin Exercise Recovery and Heat Tolerance
TABLE 3 | Serum levels of CPK (mg/Liter), CRP (mg/Liter), HSP72 (ng/mL), and lipid profile: HDl, LDL, total cholesterol and triglycerides (mg/dL).
Supp.
Time point
ATX
PLA
T-test\Mann-Whitney U
Mean ± St. Error
Mean ± St. Error
CRP pre-supp.
Before HTT
1.09 ± 0.32
0.6 ± 0.14
N.S
After HTT
1.08 ± 0.33
0.63 ± 0.14
After HTT + 1 h
1.06 ± 0.3
0.62 ± 0.13
#1 HTT
−0.01 ± 0.03
0.03 ± 0.01
##1 HTT + 1 h
−0.03 ± 0.03
0.02 ± 0.02
CRP post-supp.
Before HTT
1.54 ± 0.46
1.31 ± 0.79
After HTT
1.53 ± 0.45
1.22 ± 0.72
After HTT + 1 h
1.55 ± 0.46
1.25 ± 0.76
#1 HTT
−0.01 ± 0.04
−0.08 ± 0.07
##1 HTT + 1 h
0.01 ± 0.07
−0.05 ± 0.03
CPK pre-supp.
Before HTT
221.42 ± 61.68
188.7 ± 36.69
After HTT
260.09 ± 69.54
208 ± 39.68
After HTT + 1 h
234.42 ± 57.17
198.8 ± 33.96
#1 test
27.64 ± 9.4
19.3 ± 5.06
##1 recovery
−15 ± 8.21
−9.2 ± 4.64
CPK post-supp.
Before HTT
191.5 ± 42.54
134.5 ± 25.83
After HTT
207.42 ± 43.94
151.3 ± 26.3
After HTT + 1 h
204.75 ± 43.39
146.1 ± 25.9
#1 test
15.08 ± 4.7
16.8 ± 3.11
##1 recovery
−2.17 ± 3.35
−5.2 ± 2
HSP72 pre-supp.
Before HTT
1.9 ± 0.75
4.06 ± 0.62
0.021
After HTT
2.68 ± 0.88
3.77 ± 0.67
N.S
**1 HTT
0.77 ± 0.54
−0.3 ± 0.53
HSP72 post-supp.
Before HTT
2.25 ± 0.79
3.75 ± 0.77
After HTT
2.31 ± 0.76
3.77 ± 0.77
**1 HTT
0.06 ± 0.09
0.03 ± 0.09
HSP72 *1Supp.
1 before HTT
0.35 ± 0.12
−0.32 ± 0.34
1 after HTT
−0.37 ± 0.59
0.01 ± 0.78
Total cholesterol
Pre-supp.
139 ± 9.82
156.6 ± 5.98
Post-supp.
144.5 ± 7.12
160.5 ± 7.12
*1
5.5 ± 8.14
3.9 ± 5.12
Triglycerides
Pre-supp.
88.25 ± 16.18
98 ± 11.62
Post-supp.
94.33 ± 15.78
95.2 ± 11.62
*1
6.08 ± 8.25
−2.8 ± 9.88
HDL
Pre-supp.
46 ± 2.7
47.2 ± 2.92
Post-supp.
46.42 ± 2.04
47.8 ± 2.12
*1
0.42 ± 1.66
0.6 ± 1.42
LDL
Pre-supp.
96.17 ± 5.29
105.7 ± 6.29
Post-supp.
94.42 ± 5.68
108.8 ± 6.69
*1
−1.75 ± 3.85
3.1 ± 3.74
n = 12 in the ATX group, and 10 in the PLA group.
HSP72 protein levels were measured in eight participants of each group using ELISA before the HTT and 1 h after completion of the test. Analysis was made in triplicate and averaged
for each measurement.
#Calculated by subtracting before HTT from after HTT values.
##Calculated by subtracting before HTT from after HTT + 1 h values.
*Supplementation delta was calculated by subtracting pre supplementation values from the post supplementation values for the appropriate time point.
**Delta (1) at each HTT was calculated by subtracting Before HTT results from the After HTT results.
Supp., supplementation; HTT, heat tolerance test; 1, delta, HDL, high density lipoproteins; LDL, low density lipoproteins.
For CRP, Friedman’s omnibus test between time points within treatments revealed no statistically significant difference.
For CPK, 1-way ANOVA revealed no significant difference for any of the time points between the ATX and PLA groups. Friedman’s omnibus test, followed by Wilcoxon signed-ranks
test revealed significant differences between CPK values before and after the HTT (p = 0.005 and p < 0.01 in the ATX and PLA groups, respectively) and in the ATX group, between
after the HTT and 1 h after its completion (p = 0.006).
Analysis of HSP72 with Wilcoxon signed ranks test revealed no significant difference between measurements within treatment groups.
Analysis of TG’s by Wilcoxon signed-ranks test revealed no significant difference between time points within treatment groups.
Analysis of Cholesterol and lipids by paired samples t-test revealed no significant difference between measurements within treatment groups.
Frontiers in Sports and Active Living | www.frontiersin.org
9
September 2019 | Volume 1 | Article 17
Fleischmann et al.
Astaxanthin Exercise Recovery and Heat Tolerance
resilience in preparation for an anticipated exposure to
adverse
conditions
and
to
strenuous
efforts.
Long
term
supplementation of 12 mg\daily Astaxanthin contributed to
improved aerobic recovery, but was not beneficially manifested
under the examined heat load conditions. It remains to
be seen if administration of larger doses of Astaxanthin or
exposure to greater environmental and physiological stress
that elicit a heat shock response might bring additional
protective
mechanisms
of
Astaxanthin
supplementation
into light.
DATA AVAILABILITY
The raw data supporting the conclusions of this manuscript will
be made available by the authors, without undue reservation, to
any qualified researcher.
AUTHOR CONTRIBUTIONS
CF contributed to the conception and design of the study,
conducted the experiments, analyzed the data, and wrote
the manuscript. MH contributed to data analysis and to
manuscript design and reviewed the manuscript. RY contributed
to conducting the experiments, to data analysis, and reviewed
the manuscript. HR participated as the study nutritionist and
contributed to conducting the experiments, to data analysis,
and reviewed the manuscript. YH contributed to the conception
and design of the study, to conducting the experiments, and
reviewed the manuscript.
FUNDING
This work was funded by the IDF medical corps research fund,
grant number: 44405899192.
ACKNOWLEDGMENTS
The authors would like to acknowledge the participants who took
part in this study.
SUPPLEMENTARY MATERIAL
The Supplementary Material for this article can be found
online
at:
https://www.frontiersin.org/articles/10.3389/fspor.
2019.00017/full#supplementary-material
Supplemental Figure S1 | Scatter plot of RER vs. VO2 during the VO2 Max
tests, for the ATX and PLA supplementation groups. The ATX group before and
after supplementation is represented by the blue and gray dots, respectively. The
PLA group before and after supplementation is represented by the orange and
yellow dots, respectively.
Supplemental Figure S2 | VO2 Max test. Depicts the VO2 Max test graphs by
group and stage: (A) upper left, green lines: ATX group, before supplementation;
(B) lower left, brown lines: ATX group, after supplementation; (C) upper right, blue
lines: PLA group, before supplementation; (D) lower right, black lines: PLA group,
after supplementation.
REFERENCES
Aoi, W., Naito, Y., Sakuma, K., Kuchide, M., Tokuda, H., Maoka, T., et al. (2003).
Astaxanthin limits exercise-induced skeletal and cardiac muscle damage in
mice. Antioxid. Redox Signal. 5, 139–144. doi: 10.1089/152308603321223630
Aoi, W., Naito, Y., Takanami, Y., Ishii, T., Kawai, Y., Akagiri, S., et al. (2008).
Astaxanthin improves muscle lipid metabolism in exercise via inhibitory
effect of oxidative CPT I modification. Biochem. Biophys. Res. Commun. 366,
892–897. doi: 10.1016/j.bbrc.2007.12.019
Balady, G. J., Arena, R., Sietsema, K., Myers, J., Coke, L., Fletcher, G. F., et al. (2010).
Clinician’s guide to cardiopulmonary exercise testing in adults: a scientific
statement from the American Heart Association. Circulation 122, 191–225.
doi: 10.1161/CIR.0b013e3181e52e69
Bloomer, R. J., Fry, A., Schilling, B., Chiu, L., Hori, N., and Weiss, L. (2005).
Astaxanthin supplementation does not attenuate muscle injury following
eccentric exercise in resistance-trained men. Int. J. Sport Nutr. Exerc. Metab.
15, 401–412. doi: 10.1123/ijsnem.15.4.401
Borg, G. (1998). Borg’s Perceived Exertion and Pain Scales. Leeds: Human Kinetics
Europe Ltd.
Brown, D. R., Gough, L. A., Deb, S. K., Sparks, S. A., and Mcnaughton, L. R. (2018).
Astaxanthin in exercise metabolism, performance and recovery: a review. Front.
Nutr. 4:76. doi: 10.3389/fnut.2017.00076
Burton, A. C. (1935). Human calorimetry: II. The average temperature of the
tissues of the body: three figures. J. Nutr. 9, 261–280. doi: 10.1093/jn/9.3.261
Carter, R. III., Cheuvront, S. N., Williams, J. O., Kolka, M. A., Stephenson,
L. A., Sawka, M. N., et al. (2005). Epidemiology of hospitalizations and
deaths from heat illness in soldiers. Med. Sci. Sports Exerc. 37, 1338–1344.
doi: 10.1249/01.mss.0000174895.19639.ed
Casa, D. J., Armstrong, L. E., Ganio, M. S., and Yeargin, S. W. (2005). Exertional
heat stroke in competitive athletes. Curr. Sports Med. Rep. 4, 309–317.
doi: 10.1097/01.CSMR.0000306292.64954.da
Cheng, C. H., Guo, Z. X., Ye, C. X., and Wang, A. L. (2018). Effect of dietary
astaxanthin on the growth performance, non-specific immunity, and
antioxidant capacity of pufferfish (Takifugu obscurus) under high temperature
stress.
Fish
Physiol.
Biochem.
44,
209–218.
doi:
10.1007/s10695-017-
0425-5
Choi, H. D., Kim, J. H., Chang, M. J., Kyu-Youn, Y., and Shin, W. G. (2011). Effects
of astaxanthin on oxidative stress in overweight and obese adults. Phytother.
Res. 25, 1813–1818. doi: 10.1002/ptr.3494
Debeaumont, D., Tardif, C., Folope, V., Castres, I., Lemaitre, F., Tourny, C., et al.
(2016). A specific prediction equation is necessary to estimate peak oxygen
uptake in obese patients with metabolic syndrome. J. Endocrinol. Invest. 39,
635–642. doi: 10.1007/s40618-015-0411-7
Djordjevic, B., Baralic, I., Kotur-Stevuljevic, J., Stefanovic, A., Ivanisevic, J.,
Radivojevic, N., et al. (2012). Effect of astaxanthin supplementation on muscle
damage and oxidative stress markers in elite young soccer players. J. Sports Med.
Phys. Fitness 52, 382–392.
Do, L. T., Luu, V. V., Morita, Y., Taniguchi, M., Nii, M., Peter, A. T., et al. (2015).
Astaxanthin present in the maturation medium reduces negative effects of heat
shock on the developmental competence of porcine oocytes. Reprod. Biol. 15,
86–93. doi: 10.1016/j.repbio.2015.01.002
Durnin, J. V., and Womersley, J. (1974). Body fat assessed from total body
density and its estimation from skinfold thickness: measurements on 481
men and women aged from 16 to 72 years. Br. J. Nutr. 32, 77–97.
doi: 10.1079/BJN19740060
Earnest, C. P., Lupo, M., White, K. M., and Church, T. S. (2011). Effect of
astaxanthin on cycling time trial performance. Int. J. Sports Med. 32, 882–888.
doi: 10.1055/s-0031-1280779
Edvardsen, E., Hem, E., and Anderssen, S. A. (2014). End criteria for reaching
maximal oxygen uptake must be strict and adjusted to sex and age:
a cross-sectional study. PLoS ONE 9:e85276. doi: 10.1371/journal.pone.00
85276
Frontiers in Sports and Active Living | www.frontiersin.org
10
September 2019 | Volume 1 | Article 17
Fleischmann et al.
Astaxanthin Exercise Recovery and Heat Tolerance
Epstein, Y., Shani, Y., Moran, D. S., and Shapiro, Y. (2000). Exertional heat stroke–
the prevention of a medical emergency. J. Basic Clin. Physiol. Pharmacol. 11,
395–401. doi: 10.1515/JBCPP.2000.11.4.395
Fassett, R. G., and Coombes, J. S. (2012). Astaxanthin in cardiovascular health and
disease. Molecules 17, 2030–2048. doi: 10.3390/molecules17022030
Guo, X., Cao, J., Wang, Y., Zhou, H., Zhang, J., Niu, Y., et al. (2018). “PL-011
astaxanthin reduces high intensity training induced myocardial cell apoptosis
via activating Nrf2 in rats,” in Proceedings of IBEC, Vol. 1 (Beijing), PL-001–PL-
041. doi: 10.14428/ebr.v1i1.8143
Horowitz,
M.
(1998).
Do
cellular
heat
acclimation
responses
modulate
central
thermoregulatory
activity?
Physiology
13,
218–225.
doi: 10.1152/physiologyonline.1998.13.5.218
Ikeuchi, M., Koyama, T., Takahashi, J., and Yazawa, K. (2006). Effects of
astaxanthin supplementation on exercise-induced fatigue in mice. Biol. Pharm.
Bull. 29, 2106–2110. doi: 10.1248/bpb.29.2106
Karppi, J., Rissanen, T. H., Nyyssonen, K., Kaikkonen, J., Olsson, A. G.,
Voutilainen, S., et al. (2007). Effects of astaxanthin supplementation on lipid
peroxidation. Int. J. Vitam. Nutr. Res. 77, 3–11. doi: 10.1024/0300-9831.77.1.3
Kidd, P. (2011). Astaxanthin, cell membrane nutrient with diverse clinical benefits
and anti-aging potential. Altern. Med. Rev. 16, 355–364.
Kuennen, M., Gillum, T., Dokladny, K., Bedrick, E., Schneider, S., and Moseley,
P. (2011). Thermotolerance and heat acclimation may share a common
mechanism in humans. Am. J. Physiol. Regul. Integr. Comp. Physiol. 301,
R524–R533. doi: 10.1152/ajpregu.00039.2011
Kupcinskas, L., Lafolie, P., Lignell, A., Kiudelis, G., Jonaitis, L., Adamonis, K.,
et al. (2008). Efficacy of the natural antioxidant astaxanthin in the treatment
of functional dyspepsia in patients with or without Helicobacter pylori
infection: a prospective, randomized, double blind, and placebo-controlled
study. Phytomedicine 15, 391–399. doi: 10.1016/j.phymed.2008.04.004
Liu, F., Shi, H.-Z., Guo, Q.-S., Yu, Y.-B., Wang, A.-M., Lv, F., et al. (2016). Effects of
astaxanthin and emodin on the growth, stress resistance and disease resistance
of yellow catfish (Pelteobagrus fulvidraco). Fish Shellfish Immunol. 51, 125–135.
doi: 10.1016/j.fsi.2016.02.020
Mclellan, T. M., Cheung, S. S., Selkirk, G. A., and Wright, H. E. (2012). Influence
of aerobic fitness on thermoregulation during exercise in the heat. Exerc. Sport
Sci. Rev. 40, 218–219. doi: 10.1097/JES.0b013e3182625a83
Miyazawa, T., Nakagawa, K., Kimura, F., and Satoh, A. (2011). Plasma
carotenoid concentrations before and after supplementation with astaxanthin
in middle-aged and senior subjects. Biosci. Biotechnol. Biochem. 75, 1856–1858.
doi: 10.1271/bbb.110368
Moran, D., Epstein, Y., Wiener, M., and Horowitz, M. (1999). Dantrolene and
recovery from heat stroke. Aviat. Space Environ. Med. 70, 987–989.
Moran, D. S., Erlich, T., and Epstein, Y. (2007). The heat tolerance test: an efficient
screening tool for evaluating susceptibility to heat. J. Sport Rehabil. 16, 215–221.
doi: 10.1123/jsr.16.3.215
Nakagawa, K., Kiko, T., Miyazawa, T., Carpentero Burdeos, G., Kimura, F.,
and Satoh, A. (2011). Antioxidant effect of astaxanthin on phospholipid
peroxidation
in
human
erythrocytes.
Br.
J.
Nutr.
105,
1563–1571.
doi: 10.1017/S0007114510005398
Park, J. S., Mathison, B. D., Hayek, M. G., Massimino, S., Reinhart, G. A.,
and Chew, B. P. (2011). Astaxanthin stimulates cell-mediated and humoral
immune responses in cats. Vet. Immunol. Immunopathol. 144, 455–461.
doi: 10.1016/j.vetimm.2011.08.019
Preuss, H. G., Echard, B., Bagchi, D., Perricone, N. V., and Yamashita, E.
(2009). Astaxanthin lowers blood pressure and lessens the activity of the
renin-angiotensin system in Zucker Fatty Rats. J. Funct. Foods 1, 13–22.
doi: 10.1016/j.jff.2008.09.001
Res, P. T., Cermak, N. M., Stinkens, R., Tollakson, T., Haenen, G. R., Bast,
A., et al. (2013). Astaxanthin supplementation does not augment fat use
or improve endurance performance. Med. Sci. Sports Exerc. 45, 1158–1165.
doi: 10.1249/MSS.0b013e31827fddc4
Spiller, G. A., and Dewell, A. (2003). Safety of an astaxanthin-rich Haematococcus
pluvialis algal extract: a randomized clinical trial. J. Med. Food 6, 51–56.
doi: 10.1089/109662003765184741
Terazawa, S., Nakajima, H., Shingo, M., Niwano, T., and Imokawa, G. (2012).
Astaxanthin attenuates the UVB-induced secretion of prostaglandin E(2) and
interleukin-8 in human keratinocytes by interrupting MSK1 phosphorylation
in a ROS depletion-independent manner. Exp. Dermatol. 21(Suppl. 1), 11–17.
doi: 10.1111/j.1600-0625.2012.01496.x
Visioli, F., and Artaria, C. (2017). Astaxanthin in cardiovascular health and disease:
mechanisms of action, therapeutic merits, and knowledge gaps. Food Funct. 8,
39–63. doi: 10.1039/C6FO01721E
Yazaki, K., Yoshikoshi, C., Oshiro, S., and Yanase, S. (2011). Supplemental
cellular
protection
by
a
carotenoid
extends
lifespan
via
Ins/IGF-1
signaling in Caenorhabditis elegans. Oxid. Med. Cell. Longev. 2011:596240.
doi: 10.1155/2011/596240
Yoshida, H., Yanai, H., Ito, K., Tomono, Y., Koikeda, T., Tsukahara, H.,
et al. (2010). Administration of natural astaxanthin increases serum
HDL-cholesterol and adiponectin in subjects with mild hyperlipidemia.
Atherosclerosis
209,
520–523.
doi:
10.1016/j.atherosclerosis.2009.
10.012
Yoshihara, T., Sugiura, T., Miyaji, N., Yamamoto, Y., Shibaguchi, T., Kakigi,
R., et al. (2018). Effect of a combination of astaxanthin supplementation,
heat stress, and intermittent reloading on satellite cells during disuse
muscle atrophy. J. Zhejiang Univ. Sci. B 19, 844–852. doi: 10.1631/jzus.B18
00076
Conflict of Interest Statement: The authors declare that the research was
conducted in the absence of any commercial or financial relationships that could
be construed as a potential conflict of interest.
Copyright © 2019 Fleischmann, Horowitz, Yanovich, Raz and Heled. This is an
open-access article distributed under the terms of the Creative Commons Attribution
License (CC BY). The use, distribution or reproduction in other forums is permitted,
provided the original author(s) and the copyright owner(s) are credited and that the
original publication in this journal is cited, in accordance with accepted academic
practice. No use, distribution or reproduction is permitted which does not comply
with these terms.
Frontiers in Sports and Active Living | www.frontiersin.org
11
September 2019 | Volume 1 | Article 17
| Asthaxanthin Improves Aerobic Exercise Recovery Without Affecting Heat Tolerance in Humans. | 09-04-2019 | Fleischmann, Chen,Horowitz, Michal,Yanovich, Ran,Raz, Hany,Heled, Yuval | eng |
PMC7181553 | Vol.:(0123456789)
1 3
European Journal of Applied Physiology (2020) 120:961–968
https://doi.org/10.1007/s00421-020-04312-w
ORIGINAL ARTICLE
Gross and delta efficiencies during uphill running and cycling
among elite triathletes
Magnus Carlsson1,2 · Viktor Wahrenberg1 · Marie S. Carlsson1 · Rasmus Andersson1 · Tomas Carlsson1,2
Received: 18 October 2019 / Accepted: 31 January 2020 / Published online: 1 April 2020
© The Author(s) 2020
Abstract
Purpose To investigate the gross efficiency (GE) and delta efficiency (DE) during cycling and running in elite triathletes.
Methods Five male and five female elite triathletes completed two incremental treadmill tests with an inclination of 2.5°
to determine their GE and DE during cycling and running. The speed increments between the 5-min stages were 2.4 and
0.6 km h−1 during the cycling and running tests, respectively. For each test, GE was calculated as the ratio between the
mechanical work rate (MWR) and the metabolic rate (MR) at an intensity corresponding to a net increase in blood-lactate
concentration of 1 mmol l−1. DE was calculated by dividing the delta increase in MWR by the delta increase in MR for each
test. Pearson correlations and paired-sample t tests were used to investigate the relationships and differences, respectively.
Results There was a correlation between GEcycle and GErun (r = 0.66; P = 0.038; R2 = 0.44), but the correlation between DEcycle
and DErun was not statistically significant (r = − 0.045; P = 0.90; R2 = 0.0020). There were differences between GEcycle and
GErun (t = 80.8; P < 0.001) as well as between DEcycle and DErun (t = 27.8; P < 0.001).
Conclusions Elite triathletes with high GE during running also have high GE during cycling, when exercising at a treadmill
inclination of 2.5°. For a moderate uphill incline, elite triathletes are more energy efficient during cycling than during run-
ning, independent of work rate.
Keywords Triathlon · Cycling economy · Running economy · Incline · Metabolic rate · Mechanical work rate
Abbreviations
α
Treadmill inclination
DE
Delta efficiency
DEcycle
Delta efficiency during uphill cycling
DErun
Delta efficiency during uphill running
ΔMR
Change in metabolic rate
ΔMWR
Change in mechanical work rate
g
Gravitational acceleration
GE
Gross efficiency
GEcycle
Gross efficiency during uphill cycling at the
lactate threshold
GErun
Gross efficiency during uphill running at the
lactate threshold
LT
Lactate threshold, i.e. the mechanical work
rate at which the blood-lactate concentration
increased 1 mmol l−1 above the lowest meas-
ured value
mtot
Total mass of participant and equipment
MR
Metabolic rate
MWR
Mechanical work rate
RERmean Mean respiratory exchange ratio
μ
Rolling-resistance coefficient of the bicycle
̇VO2max
Maximal oxygen uptake
̇VO2mean
Mean oxygen uptake
Introduction
Triathlon comprises stages of swimming, cycling and run-
ning in a sequential order. In a World Cup Olympic-distance
competition (i.e. 1.5 km swimming, 40 km cycling and
10 km running), all three stages are important for overall
race performance (Landers et al. 2000; Ofoghi et al. 2016).
An analysis of the International Triathlon Union’s champi-
onship results from 2008 to 2012 revealed that the winners’
Communicated by Jean-René Lacour.
* Tomas Carlsson
[email protected]
1
School of Education, Health and Social Studies, Dalarna
University, Högskolegatan 2, 791 88 Falun, Sweden
2
Swedish Unit for Metrology in Sports, Dalarna University,
Högskolegatan 2, 791 88 Falun, Sweden
962
European Journal of Applied Physiology (2020) 120:961–968
1 3
mean race times were 1 h 46 min (men) and 1 h 58 min
(women) (Ofoghi et al. 2016). Moreover, it was found that
elite triathletes’ mean heart rate during an Olympic-distance
competition was 92% of their maximal heart rate, which
indicates the competition’s high-intensity character (Le
Meur et al. 2009).
From a physiological perspective, endurance perfor-
mance is determined by the sum of the aerobic and anaero-
bic energy contribution multiplied by gross efficiency (GE)
(Joyner and Coyle 2008). In triathlon, performance is mainly
determined by maximal oxygen uptake ( ̇VO2max ), lactate/
ventilatory threshold and oxygen uptake kinetics, which
together reflect the aerobic energy contribution, and exer-
cise economy (i.e. GE during the specific exercise mode)
(Jones and Carter 2000). In line with these findings, lactate-
threshold variables and peak oxygen uptake in cycling and
running were found to be predictors of Olympic-distance
triathlon performance (Miura et al. 1997; Schabort et al.
2000). Hence, the ability to exercise at a lower percentage
of ̇VO2max for a given submaximal workload (i.e. better econ-
omy) has been suggested to be of great importance for suc-
cess in triathlon (Dengel et al. 1989; Sleivert and Rowlands
1996). Accordingly, cycling economy and running economy
have been reported to be correlated with performance in
triathlon (Miura et al. 1997), and economy of movement has
been suggested to be an important determinant of triathlon
performance (Dengel et al. 1989; Tucker and Tucker 2013).
The economy of movement is reflected by the functioning
of the cardiorespiratory, metabolic, neuromuscular and bio-
mechanical systems (Barnes and Kilding 2015; Ettema and
Lorås 2009). In line with this concept, it has been suggested
that running economy is related to factors such as muscle
morphology, elastic elements and joint mechanics (Barnes
and Kilding 2015; Joyner and Coyle 2008; Lacour and Bour-
din 2015). In cycling, mechanisms such as muscle-fibre-type
transformation, changes in muscle-fibre-shortening veloci-
ties, changes within the mitochondria and biomechanical
factors have been proposed to be related to improved cycling
economy (Coyle et al. 1991; Hopker et al. 2009).
There are several ways to express cycling efficiency. Two
of these measures of efficiency are based on the relationship
between the work performed and the energy expenditure;
GE is the ratio between the mechanical work rate (MWR)
and the metabolic rate (MR) (i.e. GE = MWR/MR), whereas
delta efficiency (DE) is the ratio between the delta increase
in MWR and the delta increase in MR (i.e. DE = ΔMWR/
ΔMR). In cycling, GE varies between approximately 18
and 23% in different individuals (Coyle et al. 1992), and
the corresponding range in DE is approximately 18–27%
(Coyle et al. 1992; Ettema and Lorås 2009). The DE is usu-
ally somewhat higher than GE because the basal metabolic
rate and metabolic cost of zero-load exercise are excluded
from DE calculations.
Running economy is often measured as oxygen uptake at
a given submaximal running speed (e.g. 16 km h−1) while
running on a level treadmill, where a better running econ-
omy is indicated by a lower oxygen consumption (Barnes
and Kilding 2015). During level treadmill running, zero
external work is performed against gravity, frictional forces
or air resistance; hence, it is not appropriate to express run-
ning economy as GE or DE using a treadmill inclination of
0°. Previously, it has been found that GE during running
increases with steeper inclines (Minetti et al. 2002), which
emphasize the importance of taking the incline into account
when running efficiency is evaluated.
A recent study investigated the relationship between
triathletes’ energy expenditure during level running at
12 km h−1 and during ergometer cycling at a power output of
200 W, and no significant correlation was found between the
gross metabolic rates (Swinnen et al. 2018). Other studies
have compared DE during running and cycling using differ-
ent methods to apply external loads (e.g. running up different
inclines, applying impeding horizontal forces during level
treadmill running and treadmill cycling on a tricycle), but
the relationship between running and cycling DE was not
investigated in either study (Bijker et al. 2001, 2002).
To the best of our knowledge, no previous study has used
a fixed treadmill inclination to investigate elite triathletes’
running and cycling efficiencies. The purpose of this study
was to investigate gross efficiency and delta efficiency dur-
ing cycling and running in elite triathletes.
Methods
Participants
Five male (age: 24 ± 5 years, stature: 181 ± 4 cm, and body
mass: 73 ± 4 kg) and five female (age: 22 ± 6 years, stature:
169 ± 8 cm, and body mass: 64 ± 9 kg) elite triathletes volun-
teered to participate in the study and completed the GE and
DE tests. During a 5-year period, all ten triathletes had been
in the top 8 in the Swedish championships; seven of the par-
ticipants had at least one podium finish, and two participants
had previously won the Swedish championships in triathlon.
Testing procedures
The participants were instructed to only perform light train-
ing on the 2 days preceding their scheduled test days, to be
well hydrated, to refrain from alcohol (24 h) and caffeine
(12 h) and to avoid eating within 2 h prior to testing. On
the day of the tests, the participants completed a health-
status questionnaire, and thereafter, the participant’s stat-
ure (Harpenden Stadiometer, Holtain Limited, Crymych,
Great Britain) and body mass (Midrics 2, Sartorius AG,
963
European Journal of Applied Physiology (2020) 120:961–968
1 3
Goettingen, Germany) were measured. Additionally, the
mass of the equipment the participant used in the cycling
test (i.e. bicycle, cycling shoes, helmet and harness) and
running test (i.e. running shoes and harness) were weighed.
The cycling and running tests were performed on a motor-
driven treadmill (Saturn 450/300rs, h/p/cosmos sports &
medical GmbH, Nussdorf-Traunstein, Germany). Through-
out the tests, expired air was continuously analysed using
a metabolic cart in mixing-chamber mode (Jaeger Oxycon
Pro, Erich Jaeger Gmbh, Hoechberg, Germany). The meta-
bolic cart was calibrated according to the specifications of
the manufacturer before each test, and at the start of each
new 5-min stage, a ‘zeroing’ of the O2 and CO2 sensors was
performed. After the warm-ups and after each stage was
completed, capillary-blood samples were collected from a
fingertip and thereafter analysed to determine blood-lactate
concentrations (Biosen 5140, EKF-diagnostic GmbH, Bar-
leben, Germany).
Cycling test
Prior to the cycling test, the participants performed a
standardized warm-up. The 7.5-min warm-up started with
5 min at a treadmill inclination of 1° and treadmill speed
of 5.56 m s−1 (20 km h−1) for the men and 5.00 m s−1
(18 km h−1) for the women, which was followed by 2.5 min
at the initial work intensity of the cycling test (i.e. incli-
nation, 2.5°; speed, 4.56 m s−1 (16.4 km h−1) (men) or
3.22 m s−1 (11.6 km h−1) (women)). After the warm-up was
completed, a capillary-blood sample was collected, and
thereafter the rolling-resistance coefficient of the partici-
pant’s bicycle was determined using a previously described
method (Carlsson et al. 2016). In brief, the treadmill speed
was set at 5.56 m s−1 (20 km h−1), with the rider facing
downhill, and the treadmill’s negative inclination was then
adjusted until the participant sitting on the bicycle (without
pedalling) did not move in either the backward or forward
direction on the treadmill. Based on the equilibrium incli-
nation, the bicycle’s rolling-resistance coefficient (μ) was
calculated from the formula μ = mtot · g · sin α/mtot · g · cos α,
where mtot is the mass of the participant, including the mass
of the equipment (kg), g is the acceleration due to gravity
(9.82 m s−2 at the location of the sport-science laboratory)
and α is the treadmill inclination (°).
Throughout the cycling test, the treadmill inclination was
2.5°, and the participants were permitted to use a self-chosen
cadence. For each of the subsequent stages, the speed was
increased by 0.67 m s−1 (2.4 km h−1). Each stage lasted for
5 min, and the mean oxygen uptake ( ̇VO2mean ) and mean
respiratory exchange ratio (RERmean) during the last 2 min
of the stages were used for calculation of gross efficiency
(GEcycle) and delta efficiency (DEcycle) during cycling.
The stages were separated by a 1-min pause to collect a
capillary-blood sample, and the participants rated their per-
ceived exertion (RPE) on a scale of 6–20 (Borg 1970). The
pre-determined criteria for permitting the participants to
commence another stage were as follows: the participant’s
RPE had to be lower than 17 (“Very hard”), and the previous
stage’s RER had to be lower than 1.0.
Running test
To minimize the influence of the cycling test on the sub-
sequent running test, the running test was initiated 60 min
after the completion of the cycling test. It has previously
been reported that approximately 30 min of passive recovery
is sufficient to reduce the blood-lactate concentration from
3.9 ± 0.3 mmol l−1 to baseline values (1.0 ± 0.1 mmol l−1) in
moderately trained adults (Menzies et al. 2010); hence, the
time for the elite triathletes to recover after the submaximal
cycling test was considered to be sufficient. The partici-
pants were permitted to use a self-chosen stride frequency
throughout the test. Prior to the start of the running test, the
participants performed a 5-min warm-up at a treadmill incli-
nation of 2.5°, and the treadmill speeds were 10.0 km h−1
and 8.2 km h−1 for the men and women, respectively. After
the warm-up was completed, a capillary-blood sample was
collected. Thereafter, the running test was initiated with the
same intensities as in the warm-up; it consisted of 5-min
stages with a 1-min pause between stages to collect a cap-
illary-blood sample, and to have the participants rate their
perceived exertion. Throughout the test, the treadmill incli-
nation was fixed at 2.5°, and the treadmill speed increment
was 0.6 km h−1 between stages. The criteria for permitting
the participants to commence another stage were the same as
those for the cycling test. The ̇VO2mean and RERmean during
the last 2 min of the stages were used for calculation of the
gross efficiency (GErun) and delta efficiency (DErun) during
running.
Calculation of delta efficiency
For each completed stage in both tests, i.e. when
RER was < 1.0 and blood-lactate concentration
was < 4.0 mmol l−1, the mechanical work rate (MWR) and
metabolic rate (MR) were calculated. The MWR (W) dur-
ing the cycling test was the sum of the work against gravity
and the work related to overcoming the rolling resistance of
the bicycle: MWR = (mtot · g · sin α · v + mtot · g · cos α · μ
· v), where v is the treadmill speed (m s−1). In the running
test, only the component related to work performed against
gravity was included in the MWR calculation. The MR (W)
was based on the participant’s ̇VO2mean (l s−1) and RERmean:
MR = k1 · ̇VO2mean · k2, where k1 is 3.815 + 1.232 · RERmean
(Lusk 1928) and k2 is 4186 and converts kcal to J. Linear
regression was used to determine the relationship between
964
European Journal of Applied Physiology (2020) 120:961–968
1 3
MWR and MR for each participant. Based on the relation-
ship, DEcycle and DErun were calculated by dividing the delta
increase in MWR by the delta increase in MR for each test.
Calculation of gross efficiency
For both tests, the GE was calculated as the ratio between the
MWR and MR during the last 2 min of each stage. To make
an adequate comparison between GEcycle and GErun, the
MWR when the blood-lactate concentration had increased
1 mmol l−1 above the lowest measured value (LT) was used.
To establish the work rate at the LT, a third-order polyno-
mial equation was fitted to each of the participant’s obtained
MWR/blood-lactate concentration combinations, i.e. it was
calculated even for those stages that resulted in lactate val-
ues exceeding 4 mmol l−1. The polynomial equation was
then used to calculate the MWR at the LT. Thereafter, linear
regression was used to determine the relationship between
MWR and GE for each participant. Based on the linear equa-
tion and the participant’s MWR at the LT, the GE at the LT
was calculated.
Statistical analyses
The test results are presented as the mean and standard
deviation (SD). The agreement of test variables with a nor-
mal distribution was assessed with the Shapiro–Wilk test.
Pearson’s product–moment correlation coefficient (r) test
was used to investigate the relationship between GEcycle
and GErun as well as between DEcycle and DErun. The guide-
lines for the interpretation of the strength of the correlation
are as follows: small correlation for 0.1 ≤|r|< 0.3, moder-
ate correlation for 0.3 ≤|r|< 0.5, and large correlation for
|r|≥ 0.5 (Cohen 1988). Paired-samples t tests were used to
investigate differences between GEcycle and GErun as well
as between DEcycle and DErun. The Cohen’s effect-size cri-
teria were used to interpret the magnitude of the effect size
(η2) and to enable making more informative inferences
from the results. The substantial effects were divided into
more fine-graded magnitudes as follows: small effect for
0.01 ≤ η2 < 0.06, moderate effect for 0.06 ≤ η2 < 0.14, and
large effect for η2 ≥ 0.14 (Cohen 1988). All statistical analy-
ses were assumed to be significant at an alpha level of 0.05.
The statistical analyses were conducted using the IBM SPSS
Statistics software, Version 25 (IBM Corporation, Armonk,
NY, USA).
Results
The test results of the cycling and running test are pre-
sented in Tables 1 and 2, respectively. The mass of the
equipment was 9.6 ± 0.5 kg during the cycling test and
0.9 ± 0.2 kg during the running test. The bicycles’ μ was
determined to 0.0042 ± 0.0006 N N−1. The intercepts for
the relationship between MWR and oxygen uptake were
0.44 ± 0.12 l min−1 and 0.28 ± 0.20 l min−1 for the cycling
and running test, respectively. The blood-lactate concentra-
tions at LT were 1.9 ± 0.2 mmol l−1 for the cycling test and
2.2 ± 0.3 mmol l−1 for the running test. The maximum blood-
lactate concentration after each test was 4.3 ± 1.1 mmol l−1
and 4.0 ± 1.8 mmol l−1 for the cycling and running test,
respectively.
The test results in the DE tests were DEcycle = 23.5 ± 1.6%
and DErun = 8.3 ± 0.5%. All test variables were normally
distributed (all P > 0.05). There was a correlation between
GEcycle and GErun (r = 0.66; P = 0.038; R2 = 0.44) (Fig. 1),
and the participants’ sex was not a contributing factor
Table 1 Test results from the cycling test
All values are presented as mean ± standard deviation
̇VO2mean mean oxygen uptake (l min−1), RER respiratory exchange
ratio (l l−1), MR metabolic rate (W), MWR mechanical work rate (W),
GE gross efficiency (%), LT the MWR at which the blood-lactate con-
centration increased 1 mmol l−1 above the lowest measured value. All
ten participants completed stage 1–5. Stages 6 and 7 were completed
by eight and zero participants, respectively, N/A not applicable
Stage
̇VO2mean
RER
MR
MWR
GE
1
2.16 ± 0.50
0.80 ± 0.03
726 ± 170
144 ± 37
19.7 ± 0.8
2
2.42 ± 0.51
0.84 ± 0.03
820 ± 175
168 ± 39
20.4 ± 0.7
3
2.72 ± 0.55
0.85 ± 0.03
921 ± 187
193 ± 41
20.9 ± 0.7
4
3.03 ± 0.60
0.86 ± 0.03
1032 ± 206
217 ± 44
21.0 ± 0.6
5
3.32 ± 0.63
0.90 ± 0.02
1140 ± 216
242 ± 46
21.2 ± 0.7
6
3.61 ± 0.67
0.93 ± 0.02
1250 ± 232
269 ± 47
21.5 ± 0.7
7
N/A
N/A
N/A
N/A
N/A
LT
3.14 ± 0.62
0.87 ± 0.04
1069 ± 209
226 ± 45
21.1 ± 0.7
Table 2 Test results from the running test
All values are presented as mean ± standard deviation
̇VO2mean mean oxygen uptake (l min−1), RER respiratory exchange
ratio (l l−1), MR metabolic rate (W), MWR mechanical work rate (W),
GE gross efficiency (%), LT the MWR at which the blood-lactate con-
centration increased 1 mmol l−1 above the lowest measured value. All
ten participants completed stage 1–5. Stage 6 and 7 was completed by
9 and 5 participants, respectively.
Stage
̇VO2mean
RER
MR
MWR
GE
1
2.73 ± 0.47
0.83 ± 0.03
923 ± 157
76 ± 15
8.2 ± 0.5
2
2.89 ± 0.49
0.84 ± 0.03
977 ± 164
81 ± 15
8.2 ± 0.6
3
3.03 ± 0.51
0.85 ± 0.03
1027 ± 172
86 ± 16
8.3 ± 0.6
4
3.23 ± 0.54
0.86 ± 0.04
1097 ± 180
90 ± 17
8.2 ± 0.6
5
3.38 ± 0.53
0.87 ± 0.04
1152 ± 177
95 ± 17
8.3 ± 0.5
6
3.57 ± 0.63
0.89 ± 0.05
1222 ± 210
101 ± 19
8.2 ± 0.5
7
3.79 ± 0.52
0.89 ± 0.05
1298 ± 171
112 ± 15
8.6 ± 0.2
LT
3.41 ± 0.69
0.88 ± 0.04
1165 ± 227
96 ± 22
8.2 ± 0.5
965
European Journal of Applied Physiology (2020) 120:961–968
1 3
(P = 0.79). There was a significant relationship between
oxygen uptake values during cycling and running at LT
(r = 0.95; P < 0.001; R2 = 0.90). No significant relationship
was found between DEcycle and DErun (r = − 0.045; P = 0.90;
R2 = 0.0020) (Fig. 2), and the participants’ sex was not a
contributing factor (P = 0.38).
There were differences between GEcycle and GErun
(t = 80.8; P < 0.001; η2 = 0.99) as well as between DEcycle
and DErun (t = 27.8; P < 0.001; η2 = 0.98) (Fig. 3). Moreo-
ver, there was a difference between GEcycle and DEcycle
(t = − 5.85; P < 0.001; η2 = 0.79); however, no difference
was found between GErun and DErun (t = − 0.40; P = 0.70;
η2 = 0.018).
Discussion
The results of this study demonstrate that there is a large
correlation between elite triathletes’ GE during running and
cycling on a moderate uphill incline. However, no corre-
lation was found between DE during running and cycling.
The results reveal that GE and DE differ between cycling
and running, with large effect sizes, where cycling is more
energy efficient than running on a moderate uphill incline.
The finding that GErun and GEcycle are strongly correlated
(Fig. 1) is consistent with results of previous studies that
reported a significant positive correlation between cyclists’
running economy and cycling economy when the economy
of movement was measured during level running and ergom-
eter cycling, respectively (Lundby et al. 2017; Swinnen et al.
2018). Moreover, in the current study, there was a signifi-
cant correlation between gross metabolic rates (i.e. oxygen
uptake) during cycling and running. This result contradicts
the result from recent study that found no significant rela-
tionship between the gross metabolic rates during running
and cycling for nine sub-elite triathletes (Swinnen et al.
10
9
8
7
26
25
24
23
22
21
20
19
GEcycle (%)
GErun (%)
Fig. 1 Significant relationship between gross efficiency during run-
ning (GErun) and gross efficiency during cycling (GEcycle) (P < 0.05)
10
9
8
7
26
25
24
23
22
21
20
19
DEcycle (%)
DErun (%)
Fig. 2 No significant relationship between delta efficiency dur-
ing running (DErun) and delta efficiency during cycling (DEcycle)
(P > 0.05)
Fig. 3 Significant differences between gross efficiency during run-
ning (GErun) and cycling (GEcycle) is reported as †P < 0.001, and
between delta efficiency during running (DErun) and cycling (DEcycle)
is reported as ‡P < 0.001. Squares and circles represent mean values,
and error bars represent ± 1 standard deviation
966
European Journal of Applied Physiology (2020) 120:961–968
1 3
2018); however, the relationship was close to significance
(P = 0.053).
The large correlation between exercise-mode efficiencies
found in the current study indicates that an elite triathlete
with a high GErun also has a high GEcycle. This result con-
tradicts previous findings that the efficiency in one exercise
mode does not predict the efficiency in other exercise modes
(Daniels et al. 1984). However, it should be noted that in the
study by Daniels et al. (1984), trained runners were tested in
exercise modes outside their specific sport (i.e. bench step-
ping, arm cranking, graded walking and cycling) in addi-
tion to running. In the case of the participants in the current
study, one can assume that to become a national level elite
triathlete it is important to be efficient at all three disciplines
in triathlon, which partly could explain the interrelationship
of GErun and GEcycle.
The exercise efficiency is determined by the cardiores-
piratory, metabolic, neuromuscular and biomechanical effi-
ciencies (Barnes and Kilding 2015). The cardiorespiratory
and metabolic efficiencies reflect the delivery of oxygen
to the force-producing muscles and the adenosine triphos-
phate re-synthesis therein (Barnes and Kilding 2015; Saun-
ders et al. 2004). The neuromuscular and biomechanical
efficiencies reflect the interactions between the neural and
musculoskeletal systems as well as the efficiency of convert-
ing produced power to forward propulsion (Anderson 1996;
Barnes and Kilding 2015). The energy expenditure during
cycling and running is related to the increase in potential
energy during the pedal cycle/stride cycle (i.e. the raising
of centre of mass vertically during the pedal cycle/stride
cycle), the translational kinetic energy (i.e. the braking and
propelling of the body mass in the forward direction paral-
lel to the surface) and the rotational kinetic energy (i.e. the
swinging of the legs and arms) as well as the maintenance of
balance and energy cost of supporting body weight (Bergh
1987; Hoogkamer et al. 2014). Hence, a triathlete’s GE is
determined by these four underlying efficiencies (i.e. cardi-
orespiratory, metabolic, neuromuscular and biomechanical
efficiency) and at least one of these underlying efficiencies is
significantly higher for a ‘more efficient’ triathlete compared
to their ‘less efficient’ counterpart.
In the current study, it was found that GEcycle was sig-
nificantly lower than DEcycle (Fig. 3), and the difference
was associated with a large effect size. This difference is to
a large extent explained by the influence of baseline energy
expenditure in the GE calculations, which previously has
suggested being an artefact (Gaesser and Brooks 1975).
The relative contribution of the baseline energy expendi-
ture in the GE calculations decreases gradually with
increasing work intensity; hence, it should be expected that
the GE during cycling is related to work intensity. This
is in line with a previous study that reported a positive
relationship between cyclists’ GE and crank inertial load
(Bertucci et al. 2012). In the running test, there was no dif-
ference between GErun and DErun, which means that triath-
letes’ GE during moderate uphill running does not change
significantly with increasing work rate. Calculations of GE
during uphill running, based on reported values for run-
ning speed and treadmill inclination as well as the mean
values of body mass and oxygen uptake (Hoogkamer et al.
2014), showed that GE was independent of running speed
(2–3 m s−1); GE was calculated to be approximately 7%
at a 2° incline and 10% at a 3° incline, which are in good
agreement with the results in the current study.
When comparing GErun with GEcycle at the same exter-
nal work rate and treadmill incline, cycling was shown
to be more energy efficient than running (21.1% versus
8.2%, respectively) (Fig. 3), despite the limitation that the
equation for calculating MWR during cycling does not
account for the work done to overcome the friction of the
drivetrain of the bicycle. Hence, the GEcycle is therefore
somewhat underestimated. At an equivalent metabolic
energy expenditure rate, the mechanical power output
during cycling was approximately 2.5 times higher than
during running. Hence, based on the previously presented
equation that endurance performance is equal to the sum
of aerobic and anaerobic energy contributions multiplied
by GE (Joyner and Coyle 2008), it can be concluded that
the cycling speed is much higher than the running speed
for a moderate uphill incline at the same work intensity as
a consequence of the higher GE during cycling. The GE
difference between cycling and running is to a large extent
explained by differences in the factors related to force gen-
eration to support body weight, an increase in potential
energy during the pedal cycle/stride cycle and to trans-
lational kinetic energy. Running entails a considerably
higher vertical raising of the centre of mass (~ 8–10 cm)
(Cavagna et al. 2005; Tartaruga et al. 2012) compared to
cycling, where the raising of the centre of mass is mini-
mal during pedalling (Connolly 2016). The importance of
having a relatively low vertical centre-of-mass displace-
ment during running is indicated by a lower energy cost
and thus better running economy (Folland et al. 2017).
Moreover, the stride cycle during running implies a decel-
eration at the foot plant, followed by an acceleration of
body mass at push-off (Hamner et al. 2010). Based on fun-
damental physics, these deceleration/acceleration phases
are associated with a significant energy cost; however, the
relative energy-cost contribution related to translational
kinetic energy decreases during uphill running, because
on inclines the braking forces at the foot plant decrease
(Gottschall and Kram 2005). In cycling, the fluctuation in
speed during the pedal cycle is lower than in running due
to the continuous supply of power (Fintelman et al. 2016;
van Ingen Schenau et al. 1990). The described energy-
expenditure differences between the exercise modes result
967
European Journal of Applied Physiology (2020) 120:961–968
1 3
in a reduced biomechanical efficiency during running com-
pared to cycling, which is the major factor explaining the
lower GE during running.
The corresponding reasoning could be applied to under-
stand the difference, and large effect size, between DErun
and DEcycle (Fig. 3), because DE reflects how much a tri-
athlete needs to increase his/her MR for an increment in
MWR. Hence, the enhanced biomechanical efficiency dur-
ing cycling compared to running is reflected in a higher
DEcycle than DErun. This finding contradicts results from
previous studies that reported higher DE during running
(~ 44%) compared to cycling (~ 25%), where DE was
derived from tests using a constant running speed with an
incremental increase in treadmill inclination and station-
ary ergometer cycling (Bijker et al. 2001, 2002). Previ-
ously it was reported that the metabolic cost of running
parallel to the running surface decreases with incline,
whereas the efficiency of producing mechanical power
to lift the centre of mass vertically is constant and inde-
pendent of incline and running speed (Hoogkamer et al.
2014). Hence, the methodological differences (i.e. constant
speed and incremental increase in incline vs. incremental
increase in speed and constant incline) could to a large
extent explain the contradicting DE values during running.
The correlation analysis showed that participants’ sex was
not a contributing factor to the relationship between GEcycle
and GErun. This result is in line with previously reported
results where GE during cycling did not differ between male
and female competitive cyclists when the sexes were com-
pared at the same relative intensity (Hopker et al. 2010).
Moreover, in a recently published review investigating fac-
tors affecting the energy cost of running, it was concluded
that men and women with the same body mass have similar
running economies (Lacour and Bourdin 2015). However,
because of the low number of participants in the current study
further research is warranted to investigate potential sex dif-
ferences in GE for elite triathletes during cycling and running.
Conclusions
The results show that elite triathletes with high GE dur-
ing running also have a high GE during cycling, when
exercising at a treadmill inclination of 2.5°. However, a
triathlete’s relative efficiency in attaining an increased
power output in terms of DE is not transferable between
the two exercise modes. In a moderate uphill incline, elite
triathletes are more energy efficient during cycling than
running, independent of work rate.
Acknowledgements Open access funding provided by Dalarna
University.
Author contributions All authors contributed to the study concep-
tion and design, data collection, and analysis. The first draft of the
manuscript was written by MC and TC and all authors commented on
previous versions of the manuscript. All authors read and approved
the final manuscript.
Compliance with ethical standards
Conflict of interest The authors declare that they have no conflicts of
interest.
Ethical approval The study was approved by the Research Ethics Com-
mittee at Dalarna University, Falun, Sweden, and the test procedures
were performed in accordance with the World Medical Association’s
Declaration of Helsinki—Ethical Principles for Medical Research
Involving Human Subjects 2008.
Informed consent Written informed consent was obtained from all
individual participants included in the study.
Open Access This article is licensed under a Creative Commons Attri-
bution 4.0 International License, which permits use, sharing, adapta-
tion, distribution and reproduction in any medium or format, as long
as you give appropriate credit to the original author(s) and the source,
provide a link to the Creative Commons licence, and indicate if changes
were made. The images or other third party material in this article are
included in the article’s Creative Commons licence, unless indicated
otherwise in a credit line to the material. If material is not included in
the article’s Creative Commons licence and your intended use is not
permitted by statutory regulation or exceeds the permitted use, you will
need to obtain permission directly from the copyright holder. To view a
copy of this licence, visit http://creat iveco mmons .org/licen ses/by/4.0/.
References
Anderson T (1996) Biomechanics and running economy. Sports Med
22:76–89. https ://doi.org/10.2165/00007 256-19962 2020-00003
Barnes KR, Kilding AE (2015) Running economy: measurement,
norms, and determining factors. Sports Med Open 1:8. https ://
doi.org/10.1186/s4079 8-015-0007-y
Bergh U (1987) The influence of body mass in cross-country skiing.
Med Sci Sports Exerc 19:324–331
Bertucci WM, Betik AC, Duc S, Grappe F (2012) Gross efficiency and
cycling economy are higher in the field as compared with on an
axiom stationary ergometer. J Appl Biomech 28:636–644. https
://doi.org/10.1123/jab.28.6.636
Bijker KE, De Groot G, Hollander AP (2001) Delta efficiencies of
running and cycling. Med Sci Sports Exerc 33:1546–1551. https
://doi.org/10.1097/00005 768-20010 9000-00019
Bijker KE, de Groot G, Hollander AP (2002) Differences in leg muscle
activity during running and cycling in humans. Eur J Appl Physiol
87:556–561. https ://doi.org/10.1007/s0042 1-002-0663-8
Borg G (1970) Perceived exertion as an indicator of somatic stress.
Scand J Rehabil Med 2:92–98
Carlsson M, Carlsson T, Wedholm L, Nilsson M, Malm C, Tonkonogi
M (2016) Physiological demands of competitive sprint and dis-
tance performance in elite female cross-country skiing. J Strength
Cond Res 30:2138–2144. https ://doi.org/10.1519/jsc.00000 00000
00132 7
Cavagna GA, Heglund NC, Willems PA (2005) Effect of an increase in
gravity on the power output and the rebound of the body in human
968
European Journal of Applied Physiology (2020) 120:961–968
1 3
running. J Exp Biol 208:2333–2346. https ://doi.org/10.1242/
jeb.01661
Cohen JW (1988) Statistical power analysis for the behavioral sciences,
2nd edn. Lawrence Erlbaum Associates, Hillsdale
Connolly JW (2016) Understanding the magic of the bicycle. Morgan
& Claypool, Bristol
Coyle EF, Feltner ME, Kautz SA, Hamilton MT, Montain SJ, Bay-
lor AM, Abraham LD, Petrek GW (1991) Physiological and
biomechanical factors associated with elite endurance cycling
performance. Med Sci Sports Exerc 23:93–107. https ://doi.
org/10.1249/00005 768-19910 1000-00015
Coyle EF, Sidossis LS, Horowitz JF, Beltz JD (1992) Cycling effi-
ciency is related to the percentage of type I muscle fibers. Med
Sci Sports Exerc 24:782–788. https ://doi.org/10.1249/00005 768-
19920 7000-00008
Daniels JT, Scardina N, Foley P (1984) VO2max during five modes of
exercise. In: Bachl N, Prokop L, Suckert R (eds) Proceedings of
the world congress on sports medicine. Urban & Schwartzenberg,
Vienna, pp 604–615
Dengel DR, Flynn MG, Costill DL, Kirwan JP (1989) Determinants of
success during triathlon competition. Res Q Exerc Sport 60:234–
238. https ://doi.org/10.1080/02701 367.1989.10607 445
Ettema G, Lorås HW (2009) Efficiency in cycling: a review. Eur J Appl
Physiol 106:1–14. https ://doi.org/10.1007/s0042 1-009-1008-7
Fintelman DM, Sterling M, Hemida H, Li FX (2016) Effect of different
aerodynamic time trial cycling positions on muscle activation and
crank torque. Scand J Med Sci Sports 26:528–534. https ://doi.
org/10.1111/sms.12479
Folland JP, Allen SJ, Black MI, Handsaker JC, Forrester SE (2017)
Running technique is an important component of running econ-
omy and performance. Med Sci Sports Exerc 49:1412–1423. https
://doi.org/10.1249/mss.00000 00000 00124 5
Gaesser GA, Brooks GA (1975) Muscular efficiency during steady-
rate exercise: effects of speed and work rate. J Appl Physiol
38:1132–1141
Gottschall JS, Kram R (2005) Ground reaction forces during down-
hill and uphill running. J Biomech 38:445–452. https ://doi.
org/10.1016/j.jbiom ech.2004.04.023
Hamner SR, Seth A, Delp SL (2010) Muscle contributions to propul-
sion and support during running. J Biomech 43:2709–2716. https
://doi.org/10.1016/j.jbiom ech.2010.06.025
Hoogkamer W, Taboga P, Kram R (2014) Applying the cost of gen-
erating force hypothesis to uphill running. Peerj. https ://doi.
org/10.7717/peerj .482
Hopker J, Passfield L, Coleman D, Jobson S, Edwards L, Carter
H (2009) The effects of training on gross efficiency in
cycling: a review. Int J Sports Med 30:845–850. https ://doi.
org/10.1055/s-0029-12377 12
Hopker J, Jobson S, Carter H, Passfield L (2010) Cycling efficiency in
trained male and female competitive cyclists. J Sports Sci Med
9:332–337
Jones AM, Carter H (2000) The effect of endurance training on param-
eters of aerobic fitness. Sports Med 29:373–386. https ://doi.
org/10.2165/00007 256-20002 9060-00001
Joyner MJ, Coyle EF (2008) Endurance exercise performance: the
physiology of champions. J Physiol 586:35–44. https ://doi.
org/10.1113/jphys iol.2007.14383 4
Lacour JR, Bourdin M (2015) Factors affecting the energy cost of level
running at submaximal speed. Eur J Appl Physiol 115:651–673.
https ://doi.org/10.1007/s0042 1-015-3115-y
Landers GJ, Blanksby BA, Ackland TR, Smith D (2000) Morphology
and performance of world championship triathletes. Ann Hum
Biol 27:387–400. https ://doi.org/10.1080/03014 46005 00448 65
Le Meur Y, Hausswirth C, Dorel S, Bignet F, Brisswalter J, Bernard T
(2009) Influence of gender on pacing adopted by elite triathletes
during a competition. Eur J Appl Physiol 106:535–545. https ://
doi.org/10.1007/s0042 1-009-1043-4
Lundby C, Montero D, Gehrig S, Hall UA, Kaiser P, Boushel R, Lun-
dby AKM, Kirk N, Valdivieso P, Fluck M, Secher NH, Edin F,
Hein T, Madsen K (2017) Physiological, biochemical, anthro-
pometric, and biomechanical influences on exercise economy
in humans. Scand J Med Sci Sports 27:1627–1637. https ://doi.
org/10.1111/sms.12849
Lusk G (1928) The elements of the science of nutrition, 4th edn. W. B.
Saunders Co., Philadephia
Menzies P, Menzies C, McIntyre L, Paterson P, Wilson J, Kemi OJ
(2010) Blood lactate clearance during active recovery after an
intense running bout depends on the intensity of the active recov-
ery. J Sports Sci 28:975–982. https ://doi.org/10.1080/02640
414.2010.48172 1
Minetti AE, Moia C, Roi GS, Susta D, Ferretti G (2002) Energy cost
of walking and running at extreme uphill and downhill slopes. J
Appl Physiol 93:1039–1046. https ://doi.org/10.1152/jappl physi
ol.01177 .2001
Miura H, Kitagawa K, Ishiko T (1997) Economy during a simu-
lated laboratory test triathlon is highly related to Olympic
distance triathlon. Int J Sports Med 18:276–280. https ://doi.
org/10.1055/s-2007-97263 3
Ofoghi B, Zeleznikow J, Macmahon C, Rehula J, Dwyer DB (2016)
Performance analysis and prediction in triathlon. J Sports Sci
34:607–612. https ://doi.org/10.1080/02640 414.2015.10653 41
Saunders PU, Pyne DB, Telford RD, Hawley JA (2004) Factors affect-
ing running economy in trained distance runners. Sports Med
34:465–485. https ://doi.org/10.2165/00007 256-20043 4070-00005
Schabort EJ, Killian SC, Gibson AS, Hawley JA, Noakes TD (2000)
Prediction of triathlon race time from laboratory testing in
national triathletes. Med Sci Sports Exerc 32:844–849. https ://
doi.org/10.1097/00005 768-20000 4000-00018
Sleivert GG, Rowlands DS (1996) Physical and physiological factors
associated with success in the triathlon. Sports Med 22:8–18.
https ://doi.org/10.2165/00007 256-19962 2010-00002
Swinnen W, Kipp S, Kram R (2018) Comparison of running and
cycling economy in runners, cyclists, and triathletes. Eur J
Appl Physiol 118:1331–1338. https ://doi.org/10.1007/s0042
1-018-3865-4
Tartaruga MP, Brisswalter J, Peyre-Tartaruga LA, Avila AO, Alberton
CL, Coertjens M, Cadore EL, Tiggemann CL, Silva EM, Kruel
LF (2012) The relationship between running economy and bio-
mechanical variables in distance runners. Res Q Exerc Sport
83:367–375. https ://doi.org/10.1080/02701 367.2012.10599 870
Tucker R (2013) Economy. In: Friel JG, Vance JS (eds) Triathlon Sci-
ence. Human Kinetics Press, Champaign.
van Ingen Schenau GJ, Vanwoensel W, Boots PJM, Snackers RW,
Degroot G (1990) Determination and interpretation of mechanical
power in human movement: application to ergometer cycling. Eur
J Appl Physiol Occup Physiol 61:11–19. https ://doi.org/10.1007/
bf002 36687
Publisher’s Note Springer Nature remains neutral with regard to
jurisdictional claims in published maps and institutional affiliations.
| Gross and delta efficiencies during uphill running and cycling among elite triathletes. | 04-01-2020 | Carlsson, Magnus,Wahrenberg, Viktor,Carlsson, Marie S,Andersson, Rasmus,Carlsson, Tomas | eng |
PMC10358465 | Assessing the agreement
between a global navigation
satellite system and an optical-
tracking system for measuring
total, high-speed running, and
sprint distances in official
soccer matches
Piotr Makar1, Ana Filipa Silva2,3,4,
Rafael Oliveira4,5,6
, Marcin Janusiak7,
Przemysław Parus8, Małgorzata Smoter9
and Filipe Manuel Clemente2,10
1Gdańsk University of Physical Education and Sport, Gdańsk, Poland
2Escola Superior Desporto e Lazer, Instituto Politécnico de Viana do
Castelo, Viana do Castelo, Portugal
3Research Center in Sports Performance, Recreation, Innovation and
Technology (SPRINT), Melgaço, Portugal
4The Research Centre in Sports Sciences, Health Sciences and Human
Development (CIDESD), Vila Real, Portugal
5Sports Science School of Rio Maior–Polytechnic Institute of Santarém,
Rio Maior, Portugal
6Life Quality Research Centre, Rio Maior, Portugal
7Śląsk Wrocław Basketball, Physiology Department, Wrocław, Poland
8FC WKS Śląsk Wrocław, Physical Performance Department, Wrocław,
Poland
9Department of Basics of Physiotherapy, Gdansk University of Physical
Education and Sport, Gdańsk, Poland
10Instituto de Telecomunicações, Delegação da Covilhã, Lisboa, Portugal
Abstract
This study aimed to compare the agreement of total distance (TD), high-speed running (HSR) dis-
tance, and sprint distance during 16 official soccer matches between a global navigation satellite
Corresponding author:
Filipe Manuel Clemente, Escola Superior Desporto e Lazer, Instituto Politécnico de Viana do Castelo, Rua
Escola Industrial e Comercial de Nun’Álvares, 4900-347 Viana do Castelo, Portugal.
Email: fi[email protected]
Creative Commons CC BY: This article is distributed under the terms of the Creative Commons
Attribution 4.0 License (https://creativecommons.org/licenses/by/4.0/) which permits any use, reproduction
and distribution of the work without further permission provided the original work is attributed as specified on the SAGE and
Open Access page (https://us.sagepub.com/en-us/nam/open-access-at-sage).
Sports Sciences, Exercise, and Health – Original
Manuscript
SCIENCE PROGRESS
Science Progress
2023, Vol. 106(3) 1–14
© The Author(s) 2023
Article reuse guidelines:
sagepub.com/journals-permissions
DOI: 10.1177/00368504231187501
journals.sagepub.com/home/sci
system (GNSS) and an optical-tracking system. A total of 24 male soccer players, who are actively
participating in the Polish Ekstraklasa professional league, were included in the analysis conducted
during official competitions. The players were systematically monitored using Catapult GNSS (10-
Hz, S7) and Tracab optical-tracking system (25-Hz, ChyronHego). TD, HSR distance, sprint dis-
tance, HSR count (HSRC), and sprint count (SC) were collected. The data were extracted in 5-
min epochs. A statistical approach was employed to visually examine the relationship between
the systems based on the same measure. Additionally, R2 was utilized as a metric to quantify
the proportion of variance accounted for by a variable. To assess agreement, Bland–Altman
plots were visually inspected. The data from both systems were compared using the estimates
derived from the intraclass correlation (ICC) test and Pearson product–moment correlation.
Finally, a paired t-test was employed to compare the measurements obtained from both systems.
The interaction between Catapult and Tracab systems revealed an R2 of 0.717 for TD, 0.512 for
HSR distance, 0.647 for sprint distance, 0.349 for HSRC, and 0.261 for SC. The ICC values for
absolute agreement between the systems were excellent for TD (ICC = 0.974) and good for
HSR distance (ICC = 0.766), sprint distance (ICC = 0.822). The ICC values were not good for
HSRCs (ICC = 0.659) and SCs (ICC = 0.640). t-test revealed significant differences between
Catapult and Tracab for TD (p < 0.001; d = −0.084), HSR distance (p < 0.001; d = −0.481), sprint
distance (p < 0.001; d = −0.513), HSRC (p < 0.001; d = −0.558), and SC (p < 0.001; d = −0.334).
Although both systems present acceptable agreement in TD, they may not be perfectly inter-
changeable, which sports scientists and coaches must consider when using them.
Keywords
: Football, athletic performance, player tracking systems, training load monitoring, locomotor
demands
Introduction
Monitoring locomotor demands through technological devices has become a widespread
and recurring practice in soccer training. A survey conducted among 94 coaches and 88
practitioners belonging to elite English soccer found that tracking systems (e.g., global
navigation satellite system [GNSS]) were used more than other different training load
monitoring methods (e.g., blood lactates, ratings of perceived exertion, heart rates).1 In
another survey conducted among 82 high-level soccer teams competing in the top
leagues of countries such as the United Kingdom, Spain, France, Germany, and Italy,
the findings revealed that approximately 40% of the teams utilized time motion analysis
and accelerometers as the primary tools for quantifying training load.2 It is noteworthy to
acknowledge that the term “training load” has been a topic of discussion regarding its
accuracy. This is primarily due to the conventional association of the term “load” with
a mechanical variable measured in newtons within the International System of Units.3
However, it is crucial to emphasize that in the specific context of training load, it
serves as a scientific construct rather than a direct “measurement” per se.4 Therefore,
its usage does not contravene any scientific principles.4
To enhance the efficacy of collecting sports science training data, it is crucial to
acknowledge the inherent value of coaches and performance staff. This recognition
plays a pivotal role in facilitating the widespread adoption and seamless integration of
meticulously engineered tracking systems that are purposefully designed to augment
2
Science Progress 106(3)
training load practices. Supporting this notion, a comprehensive survey was conducted
among 176 soccer coaches and performance coaches, revealing that sport science data
holds substantial importance in guiding their practice, being perceived as both somewhat
important and very important.5
In light of the proliferation of tracking systems, it is evident that a multitude of options
are now available on the market. However, it becomes increasingly challenging to
conduct fair comparisons and establish the interchangeability of data among these
diverse alternatives.6 Among the various technological options, GNSS, ultra-wideband
technology, and optical video tracking systems are prominent examples. The usability
of these options is contingent upon specific contexts and objectives. GNSS, while cost-
effective, is primarily applicable in outdoor facilities. Conversely, ultra-wideband tech-
nology, despite its higher cost, offers versatility by functioning effectively in both
outdoor and indoor settings.7
In the context of outdoor facilities, GNSS remains widely utilized, potentially owing
to its user-friendly nature compared to installation-dependent optical video tracking
systems, as well as its relatively lower cost in comparison to ultra-wideband technology.
It is important to note, however, that optical video tracking systems can present an alter-
native and intriguing solution for sports scientists and players in outdoor facilities. These
systems ensure high-quality data collection and provide the opportunity to combine time-
motion analysis with notational analysis, all without necessitating any devices to be
placed on the players. Having in mind the multiplicity of the options and technical
aspects of the devices, it is particularly essential to focus on the agreement between
such systems,7 considering that data collected can vary from system to system with a
remarkable impact on data interpretation.
Taking that into consideration, different studies have focused on testing absolute agree-
ment between different tracking systems.6,8,9 For example, a 10-Hz multi-GNSS GNSS
device (vector, Catapult) and two optical tracking systems (25-Hz Tracab and Second
Spectrum) were compared. The results indicated that in comparison to GNSS, Tracab
revealed significantly higher values for most locomotor measures followed by the other
optical system (Second Spectrum).6 Another study comparing two 10-Hz GNSS systems
(Viper, StatSports; and Apex, StatSports) with Tracab optical tracking system demonstrated
significant differences between GNSS and the optical tracking systems for locomotor mea-
sures such as total distance (TD), high-speed running (HSR) distance, and sprint distance.8
However, the different systems presented very large correlations.8 Another comparison of
Tracab and a 10-Hz GNSS (Wimu) revealed that the optical tracking system slightly overes-
timated most locomotor measures compared to GNSS.9
These aforementioned studies6,8,9 suggest notable differences in locomotor measures
when comparing different technologies, specifically GNSS and optical video tracking
systems. However, they also demonstrate a strong correlation between the two, indicating
the potential for interchangeability. Despite these findings, the current research has not
focused on analyzing peak demands within 5-min epoch periods. Peak demands, or
worst-case scenarios, have recently garnered attention,10 raising concerns about the
accuracy, precision, and interchangeability of different systems.
Although Ellens’ study6 investigated the interchangeability between a 10-Hz GNSS
Vector and Tracab (an optical tracking system with 25-Hz), the analysis did not
Makar et al.
3
specifically examine the interchangeability of epoch periods. Therefore, further research
is needed to determine whether interchangeability can also be achieved within 5-min
epochs. Such research would provide another independent assessment of absolute agree-
ment between different systems and models, particularly comparing new GNSS systems
like the Catapult S7 with Tracab (an optical tracking system with 25-Hz).
Testing for interchangeability holds significant importance for several reasons. Firstly, it
allows for better control over the comparisons made between scientific articles and bench-
marks conducted on players. By establishing the interchangeability of devices and technolo-
gies, it becomes possible to ensure the validity and reliability of such comparisons.
Secondly, considering that clubs often change their devices and technologies, having
access to interchangeability values becomes crucial. It provides clubs with the necessary
information to determine whether fair and accurate comparisons can be made or if caution
should be exercised due to potential discrepancies between different measurement
systems. This knowledge empowers clubs to make informed decisions regarding the
compatibility and comparability of data collected from different sources.
Therefore, the objective of this study is to assess the absolute agreement between the 10-Hz
Catapult S7 GNSS and the 25-Hz Tracab optical video tracking system in terms of TD, HSR
distance, sprint distance, as well as the number of HSR and sprints. These measures were spe-
cifically chosen due to their significance in quantifying training load within the given context.
TD serves as a comprehensive measure of locomotor demand, which is closely asso-
ciated with internal load responses.11 It provides valuable insights into the magnitude of
demands imposed on players. HSR and sprint distances were selected as they represent
the most demanding locomotor demands observed during matches. Moreover, these mea-
sures are known to have traditionally lower values of precision,12 necessitating further
examination to ensure optimal accuracy and precision of the collected data.
Methods
Study design
This study employed a longitudinal design, focusing on a group of twenty-one soccer
players from a single professional team. Over a period of 16 official soccer matches,
which took place outdoors in stadium facilities, the players were consecutively observed.
The observation period spanned from July 15, 2022, to November 13, 2022, correspond-
ing to the competitive phase of the season. For the analysis, only data from official
matches in the domestic competition, including league matches and cup matches, were
considered. The players were monitored using two tracking systems: (a) a GNSS and
(b) an optical-tracking system. The study aimed to test the agreement between both
systems for monitoring locomotor demands of the players during the match.
Participants
We used nonprobabilistic convenience sampling. A group of 21 male professional foot-
ball players (231 observations) from the first team of one of the Polish Ekstraklasa clubs
(age: 25 ± 3 years, body height: 179.6 ± 5.5 cm, body mass: 76.1 ± 5.0 kg) participated in
4
Science Progress 106(3)
the research. The data were collected over 16 matches played in the autumn round of the
2022/2023 season and were recorded simultaneously during each of the observed games.
Goalkeepers were not included in this study due to the unique nature of their position.
Considering the specific movements and actions performed by goalkeepers during
matches, the use of GNSS instruments may potentially cause damage or interfere with
their typical movements. Hence, to ensure the integrity of the study and avoid any poten-
tial harm to the goalkeepers, their data was not collected or analyzed as part of this
research. All data were created as a condition of employment where players were rou-
tinely monitored throughout league play.
In order to uphold ethical standards, all players involved in the study were provided
with detailed information about the study design, the associated risks, and the potential
benefits of participation. Only after obtaining their informed consent was the study con-
ducted. The informed consent process ensured that the players were fully aware of the
study’s objectives, procedures, and potential implications before agreeing to participate.
This study adhered to the ethical guidelines outlined in the Declaration of Helsinki for
research involving human participants. Confidentiality was strictly maintained through-
out the study by anonymizing all data prior to analysis.
Methodological procedures
Two tracking systems were used simultaneously: (a) a GNSS unit (Vector S7, Catapult
Innovations, Melbourne, Australia; 81mm×43mm×16mm), operating at a frequency of
10-Hz and (b) an optical tracking system (TRACAB, ChyronHego, New York, USA)
using two multicamera units (each containing three HD-SDI cameras with a resolution
of 1920 × 1080 pixels) with a sampling frequency of 25-Hz. On average, the number
of satellites connected during data collection was 15, and the average horizontal dilution
of precision (HDOP) was 0.7. Vector S7 was preliminarily tested for its ability to assess a
force-velocity profile.13 Furthermore, the Tracab system underwent a validation process
to assess its accuracy and precision in measuring locomotor demands across various
running speed thresholds.14
The players always used the same GNSS unit to reduce inter-unit variability.15 The
GNSS units were placed between the players’ shoulder blades and were activated accord-
ing to a manufacturer’s guidelines before kickoff. To avoid potential unit differences, the
players wore the same GNSS unit for each match.8 The data recorded by the units were
downloaded after each match for further analysis using Catapult OpenField Cloud
Analytics (OpenField 3.9.0 Catapult Sports, Melbourne, Australia). The following vari-
ables were selected for analysis during this study: field time, defined as the time spent on
the field (FT; min), TD (m), distance in HSR, defined as a running speed between 19.81
and 25.2 km/h (HSR; m), sprint, defined as velocity greater than 25.2 km/h (SPR; m),
High-speed running count (HSRC) and sprint count (SC). The HSR speed threshold of
19.81 km/h was determined based on the research conducted by Abt and Lovell,16
who identified this value as the reference for the second ventilatory threshold. This spe-
cific threshold has gained broad acceptance and is widely used as a prevalent measure to
define arbitrary speed thresholds in soccer players. The selection of the 25.2 km/h speed
threshold for sprints aligns with established conventions based on previous research
Makar et al.
5
conducted on sprinting in soccer players.17 The velocity thresholds chosen are those
defined by both tracking system providers. All data from the Tracab system were pro-
vided by ChyronHego as a match report. Data from both the Catapult and Tracab
systems were extracted in 5-min epochs, which involved dividing the official match
time into consecutive 5-min time periods. This approach ensured that all 5-min epochs
within the match time were considered and included in the analysis.
Statistical procedures
Descriptive statistics are presented in the form of mean and standard deviation. Plotting
data was performed to visually inspect the relationship between the systems for the same
measure. At the same time, R2 was used as a measure to represent the proportion of the
variance for a variable. Measuring agreement was visually inspected by Bland–Altman
plots with a 95% confidence interval using the mean difference between measures.
The estimates of the intraclass correlation (ICC) test and their 95% confidence intervals
were calculated by means of SPSS statistical package (28.0.0.0, IBM, Chicago, IL) based
on a mean-rating (k = 2), absolute-agreement, a two-way mixed effects model. The clas-
sification of the agreement18 was considered good between ICC = 0.75 and ICC = 0.90,
while the values above ICC = 0.90 were considered excellent. The Pearson–product cor-
relation test was executed on SPSS (version 28.0.0., IBM, Chicago, USA) for a p-value
less than 0.05 to analyze the strength of the relationship between the systems. The cor-
relation coefficients19 between r = 0.50 and r = 0.7 were considered large, between r =
0.7 and r = 0.9 very large, and above r = 0.9 nearly perfect. The paired t-test was used
to compare the measures obtained for both systems, followed by the standardized
effect size (Cohen’s d) that was interpreted as20: 0.0–0.2, trivial; 0.2–0.5, medium;
0.5–0.8, large; and >0.8, very large. The statistical procedures were executed in SPSS
(version 28.0.0., IBM, Chicago, USA) for a p < 0.05.
Results
Figure 1 displays a scatter plot comparing the Catapult and Tracab systems for the
various running-based measures extracted during the matches. It is important to note
that all data presented in the results correspond to the values obtained for the 5-min
epochs. The interaction between Catapult and Tracab systems revealed an R2 of 0.717
for TD, 0.512 for HSR distance, 0.647 for sprint distance, 0.349 for HSRC, and 0.261
for SC.
Table 1 presents the ICC and the Pearson–product correlation tests comparing both
systems for the different running-based measures. The ICC values for absolute agreement
between the systems were excellent for TD (ICC = 0.974) and good for HSR distance
(ICC = 0.766), and sprint distance (ICC = 0.822). The ICC values were not good for
HSRCs (ICC = 0.659) and SCs (ICC = 0.640).
The measurement agreement was performed by visual inspection of the data. Figure 2
presents the Bland–Altman plots for the different running-based measurements. The
mean difference for TD (for 5-min epochs) was −11.5 [95% CI: −165; 142], while for
HSR distance was −11 [95% CI: −44; 22] and for sprinting −8.0 [95% CI: −26; 11].
6
Science Progress 106(3)
Figure 1. Scatter plot between Catapult and Tracab for TD, HSR distance, sprint distance, HSRC,
and SC. The data displayed in the figure represents the aggregation of values over 5-min epochs.
Makar et al.
7
Regarding HSR and SCs, the mean difference was −1 [95% CI: −3; 2] and 0 [95% CI:
−1; 1].
Table 2 presents descriptive statistics of the running-based measures collected in both
the Catapult and Tracab systems. t-test revealed significant differences between Catapult
and Tracab for TD (p < 0.001; d = −0.084), HSR distance (p < 0.001; d = −0.481), sprint
distance (p < 0.001; d = −0.513), HSRC (p < 0.001; d = −0.558), and SC (p < 0.001; d =
−0.334).
Discussion
The objective of this study was to evaluate the absolute agreement of TD, HSR distance,
and sprint distance between the 10-Hz Catapult S7 GNSS and the 25-Hz Tracab optical
video tracking system. This evaluation was conducted using data collected in 5-min
epochs. The main findings showed excellent absolute agreement between the systems
for TD and good absolute agreement for HSR distance, sprint distance, HSRC, and
SCs. Nonetheless, it was noted that Catapult underestimated the values in comparison
to Tracab, particularly for HSR and sprint distances/counts. This corresponds with the
findings of previous studies that revealed higher values for tracking systems when com-
pared with GNSS6
Regarding TD, the present study showed higher values for Tracab in comparison to
Vector S7, which was also previously confirmed.6 Moreover, the findings concerning
speed values, higher for the tracking system when compared to GNSS, are similar to
the findings of previous research.7 The observed discrepancies may arise from disparities
in data filtering methodologies employed by device software. Notably, the implementa-
tion of filters such as moving averages has been shown to yield more refined speed data.
However, it is crucial to elucidate that these filtering techniques do not affect the funda-
mental principle of peak velocity. Peak velocity signifies the utmost speed attained within
a specified timeframe, irrespective of any averaged HSR or sprinting encompassed during
the said interval. It is noteworthy that the present study adopted 5-min epochs as the tem-
poral units for data analysis.
This study also found that the above differences tend to increase significantly with
higher speed distances and counts, which was found in the previous study that analyzed
Tracab and GNSS (GPEXE®, Exelio, Udine, Italia) with different epochs (15, 30, and 45
min).21 Hence, it can be inferred that as the velocity achieved during high-speed distances
Table 1. ICC and Pearson-product correlation tests comparing both systems for the different
running-based measures.
Intraclass correlation
r Pearson
Total distance (m)
0.906 [95% CI: 0.897; 0.914]
0.851 [0.841; 0.860] | p < 0.001
High-speed running (m)
0.766 [95% CI: 0.522; 0.864]
0.716 [0.697; 0.734] | p < 0.001
Sprint (m)
0.822 [95% CI: 0.445; 0.917]
0.804 [0.780; 0.826] | p < 0.001
High-speed running (n)
0.659 [95% CI: 0.399; 0.784]
0.590 [0.564; 0.615] | p < 0.001
Sprint (n)
0.640 [95% CI: 0.542; 0.712]
0.523 [0.468; 0.574] | p < 0.001
8
Science Progress 106(3)
Table 2. Descriptive statistics (mean ± standard deviation) for the different running-based measures collected by 5-min epoch and inferential
comparisons.
Catapult
Tracab
Mean difference (Catapult − Tracab)
p-value
d
Total distance (m)
for a 5-min epoch
491.0 ± 116.6
502.7 ± 144.8
−11.8 [95% CI: −14.4; −9.2]
<0.001
−0.084 [95% CI: −0.103; −0.065]
High-speed running
(m) for a 5-min
epoch
27.5 ± 19.1
38.2 ± 24.0
−10.8 [95% CI: −11.4; −10.1]
<0.001
−0.481 [95% CI: −0.512; −0.450]
Sprint (m) for a
5-min epoch
18.0 ± 13.5
25.7 ± 15.9
−7.7 [95% CI: −8.4; −7.1]
<0.001
−0.513 [95% CI: −0.560; −0.465]
High-speed running
(n) for a 5-min
epoch
2.1 ± 1.2
2.9 ± 1.5
−0.8 [95% CI: −0.8; −0.7]
<0.001
−0.558 [95% CI: −0.597; −0.519]
Sprint (n) for a
5-min epoch
1.2 ± 0.4
1.4 ± 0.6
−0.2 [95% CI: −0.2; −0.1]
<0.001
−0.334 [95% CI: −0.408; −0.261]
Makar et al.
9
Figure 2. Difference against mean for running-based measures.
10
Science Progress 106(3)
increases, the likelihood of encountering greater differences between the systems also
amplifies.
In addition, the number of HSR and sprint efforts detected was the greatest for
TRACAB, which is also in line with a previous study.6 In this regard, it is important
to acknowledge that the count detection of speed distances requires a minimum duration
above a fixed velocity. For instance, Varley et al.22 showed moderate to large differences
when different minimum effort durations were applied in the number of HSR detected
with 10-Hz GNSS during a soccer match (∼150 efforts detected for 0.1 s duration com-
pared to ∼90 efforts detected for 1 s duration). Thus, when comparing different systems,
it is relevant to consider filtering technology differences to understand the pros and cons
of each system23 as well as other factors, such as the sampling rate, the number of satel-
lites, HDOP, and the software analysis of different systems.7
A limitation of the study is a small sample size of participants and, consequently,
speculation that a larger sample size could potentially provide a different result when
comparing both systems. However, in the context of professional soccer matches, collect-
ing data by means of two different systems is complex and limits the opportunity to gather
larger sample sizes. Nonetheless, this had also been pointed as a limitation due to
non-ecological environments by the simulation of circuits or matches.24 Besides, such
devices are expensive and not all teams have access to them. The lack of analysis of
accelerometer-based
variables
is
another
limitation
that
may
provide
further
state-of-the-art knowledge, considering that accelerating or decelerating with or
without changing direction has been reported as very important in soccer.25 Therefore,
the investigation with a similar design but with a larger sample size and analysis of
accelerometer-based variables is recommended for future studies.
Conclusions
Although both systems present a strong relationship and acceptable agreement for TD,
the interchangeability should be considered cautiously, mainly regarding significant dif-
ferences in the collected measures. Coaches and sports scientists should be mindful of the
differences between both systems when comparing data and avoid using them
interchangeably.
Acknowledgments
The authors appreciate the study participants. All authors have read and agreed to the published
version of the manuscript.
Author contributions
Conceptualization by P.M. and F.M.C.; methodology by P.M. and F.M.C.; formal analysis by
F.M.C.; investigation by P.M., M.J., P.P., F.M.C.; data curation by F.M.C.; writing—original
draft preparation by P.M., A.F.S., R.O., M.J., P.P., M.S., F.M.C.; writing—review and editing
by P.M., A.F.S., R.O., M.J., P.P., M.S., F.M.C.; project administration by F.M.C.
Makar et al.
11
Availability of data and materials
All data generated or analyzed during this study are available at the request of the corresponding author.
Consent for publication
Not applicable.
Declaration of conflicting interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/
or publication of this article.
Ethics approval
Medical Ethic Committee in University of Gdańsk (Decision number 62/2022).
Funding
The author(s) disclosed receipt of the following financial support for the research, authorship, and/
or publication of this article: Filipe Manuel Clemente and this work are funded by the Fundação
para a Ciência e Tecnologia/Ministério da Ciência, Tecnologia e Ensino Superior through national
funds, and when applicable, co-funded by EU funds under the project UIDB/50008/2020.
Informed consent
All participants were informed about the study and signed free consent.
ORCID iDs
Rafael Oliveira
https://orcid.org/0000-0001-6671-6229
Filipe Manuel Clemente
https://orcid.org/0000-0001-9813-2842
References
1. Weston M. Training load monitoring in elite English soccer: a comparison of practices and per-
ceptions between coaches and practitioners. Sci Med Footb 2018; 2: 216–224.
2. Akenhead R and Nassis GP. Training load and player monitoring in high-level football: current
practice and perceptions. Int J Sports Physiol Perform 2016; 11: 587–593.
3. Staunton CA, Abt G, Weaving D, et al. Misuse of the term ‘load’ in sport and exercise science.
J Sci Med Sport 2022; 25: 439–444.
4. Impellizzeri FM, Jeffries AC, Weisman A, et al. The ‘training load’ construct: why it is appro-
priate and scientific. J Sci Med Sport 2022; 25: 445–448.
5. Nosek P, Brownlee TE, Drust B, et al. Feedback of GPS training data within professional
English soccer: a comparison of decision making and perceptions between coaches, players
and performance staff. Sci Med Footb 2021; 5: 35–47.
6. Ellens S, Hodges D, McCullagh S, et al. Interchangeability of player movement variables from
different athlete tracking systems in professional soccer. Sci Med Footb 2022; 6: 6.
7. Buchheit M, Allen A, Poon TK, et al. Integrating different tracking systems in football: mul-
tiple camera semi-automatic system, local position measurement and GPS technologies. J
Sports Sci 2014; 32: 1844–1857.
8. Taberner M, O’Keefe J, Flower D, et al. Interchangeability of position tracking technologies;
can we merge the data? Sci Med Footb 2020; 4: 76–81.
12
Science Progress 106(3)
9. Pons E, García-Calvo T, Resta R, et al. A comparison of a GPS device and a multi-camera
video technology during official soccer matches: agreement between systems. PLoS One
2019; 14: e0220729.
10. Novak AR, Impellizzeri FM, Trivedi A, et al. Analysis of the worst-case scenarios in an elite
football team: towards a better understanding and application. J Sports Sci 2021; 39: 1850–1859.
11. McLaren SJ, Macpherson TW, Coutts AJ, et al. The relationships between internal and external
measures of training load and intensity in team sports: a meta-analysis. Sports Med 2018; 48:
641–658.
12. Scott MTU, Scott TJ and Kelly VG. The validity and reliability of global positioning systems
in team sport. J Strength Cond Res 2016; 30: 1470–1490.
13. Clavel P, Leduc C, Morin JB, et al. Concurrent validity and reliability of sprinting force–vel-
ocity profile assessed with GPS devices in elite athletes. Int J Sports Physiol Perform 2022; 17:
1527–1531.
14. Linke D, Link D and Lames M. Football-specific validity of TRACAB’s optical video tracking
systems. PLoS One 2020; 15: e0230179.
15. Beato M and de Keijzer K. The inter-unit and inter-model reliability of GNSS STATSports
Apex and Viper units in measuring peak speed over 5, 10, 15, 20 and 30 meters. Biol Sport
2019; 36: 317–321.
16. Abt G and Lovell RIC. The use of individualized speed and intensity thresholds for determin-
ing the distance run at high-intensity in professional soccer. J Sports Sci 2009; 27: 893–898.
17. Bush M, Barnes C, Archer DT, et al. Evolution of match performance parameters for various
playing positions in the English Premier League. Hum Mov Sci 2015; 39: 1–11.
18. Koo TK and Li MY. A guideline of selecting and reporting intraclass correlation coefficients
for reliability research. J Chiropr Med 2016; 15: 155–163.
19. Hopkins WG, Marshall SW, Batterham AM, et al. Progressive statistics for studies in sports
medicine and exercise science. Med Sci Sports Exerc 2009; 41: 3–13.
20. Cohen J. Statistical Power Analysis for the Behavioral Sciences. 2nd ed. Vol 2. Hillsdale, NJ,
USA: Lawrence Erlbaum Associates, 1988.
21. Castellano J, Casamichana D, Campos-Vázquez MA, et al. Interchangeability of two tracking
systems to register physical demands in football: multiple camera video versus GPS technol-
ogy. Arch de Med del Deporte 2019; 36: 268–269.
22. Varley MC, Jaspers A, Helsen WF, et al. Methodological considerations when quantifying
high-intensity efforts in team sport using global positioning system technology. Int J Sports
Physiol Perform 2017; 12: 1059–1068.
23. Delves RIM, Aughey RJ, Ball K, et al. The quantification of acceleration events in elite team
sport: a systematic review. Sports Med Open 2021; 7: 45.
24. Cust EE, Sweeting AJ, Ball K, et al. Machine and deep learning for sport-specific movement
recognition: a systematic review of model development and performance. J Sports Sci 2019;
37: 568–600.
25. Delaney JA, Cummins CJ, Thornton HR, et al. Importance, reliability, and usefulness of accel-
eration measures in team sports. J Strength Cond Res 2018; 32: 3485–3493.
Author biographies
Piotr Makar is Head of Department at Gdansk University of Physical Education and Sport (Poland).
Ana Filipa Silva is an assistant professor at Instituto Politécnico de Viana do Castelo (Portugal).
Rafael Oliveira is an assistant professor at Instituto Politénico de Santarém.
Makar et al.
13
Marcin Janusiak is professor at Ś ląsk Wrocław Basketball, Physiology Department, Wrocław,
Poland.
Przemysław Parus is sport scientist at FC WKS Ś ląsk Wrocław, Physical Performance Department,
Wrocław, Poland.
Małgorzata Smoter is professor at Department of Basics of Physiotherapy, Gdansk University of
Physical Education and Sport, Gdańsk, Poland.
Filipe Manuel Clemente is an assistant professor at Instituto Politécnico de Viana do Castelo,
Portugal.
14
Science Progress 106(3)
| Assessing the agreement between a global navigation satellite system and an optical-tracking system for measuring total, high-speed running, and sprint distances in official soccer matches. | [] | Makar, Piotr,Silva, Ana Filipa,Oliveira, Rafael,Janusiak, Marcin,Parus, Przemysław,Smoter, Małgorzata,Clemente, Filipe Manuel | eng |
PMC10708873 | Citation: Bascuas, P.J.; Gutiérrez, H.;
Piedrafita, E.; Rabal-Pelay, J.; Berzosa,
C.; Bataller-Cervero, A.V. Running
Economy in the Vertical Kilometer.
Sensors 2023, 23, 9349. https://
doi.org/10.3390/s23239349
Academic Editor: Nicola Francesco
Lopomo
Received: 16 October 2023
Revised: 16 November 2023
Accepted: 17 November 2023
Published: 23 November 2023
Copyright:
© 2023 by the authors.
Licensee MDPI, Basel, Switzerland.
This article is an open access article
distributed
under
the
terms
and
conditions of the Creative Commons
Attribution (CC BY) license (https://
creativecommons.org/licenses/by/
4.0/).
sensors
Article
Running Economy in the Vertical Kilometer
Pablo Jesus Bascuas
, Héctor Gutiérrez
, Eduardo Piedrafita
, Juan Rabal-Pelay
, César Berzosa *
and Ana Vanessa Bataller-Cervero
Facultad de Ciencias de la Salud, Universidad San Jorge, Autov. A-23 Zaragoza-Huesca,
50830 Villanueva de Gallego, Spain; [email protected] (P.J.B.); [email protected] (H.G.); epiedrafi[email protected] (E.P.);
[email protected] (J.R.-P.); [email protected] (A.V.B.-C.)
* Correspondence: [email protected]
Abstract: New and promising variables are being developed to analyze performance and fatigue in
trail running, such as mechanical power, metabolic power, metabolic cost of transport and mechanical
efficiency. The aim of this study was to analyze the behavior of these variables during a real vertical
kilometer field test. Fifteen trained trail runners, eleven men (from 22 to 38 years old) and four women
(from 19 to 35 years old) performed a vertical kilometer with a length of 4.64 km and 835 m positive
slope. During the entire race, the runners were equipped with portable gas analyzers (Cosmed K5)
to assess their cardiorespiratory and metabolic responses breath by breath. Significant differences
were found between top-level runners versus low-level runners in the mean values of the variables of
mechanical power, metabolic power and velocity. A repeated-measures ANOVA showed significant
differences between the sections, the incline and the interactions between all the analyzed variables,
in addition to differences depending on the level of the runner. The variable of mechanical power
can be statistically significantly predicted from metabolic power and vertical net metabolic COT. An
algebraic expression was obtained to calculate the value of metabolic power. Integrating the variables
of mechanical power, vertical velocity and metabolic power into phone apps and smartwatches is a
new opportunity to improve performance monitoring in trail running.
Keywords: performance monitoring; energy expenditure; human movement; trail running
1. Introduction
Over the past decade, there has been a significant increase in interest in sport field
applications, driven by both users and technological companies. This interest has been
propelled by advancements in the development of wearable sensors based on micro-
electromechanical systems (MEMSs) [1]. These sensors find application during training
sessions and sports competitions, serving the purpose of monitoring the internal training
load [2], scheduling workouts and tracking the athlete’s fitness level progression. To
achieve this objective, it is essential to develop automated assessment methods that analyze
highly accurate variables capable of reflecting the physiological, metabolic, biomechanical
and neuromuscular state of the athlete. Additionally, these methods should be easily
implemented in low-cost sensors, such as inertial measurement units, linear transducers,
potentiometers and global navigation satellite systems, among others [3].
Trail running races have increasingly gained the interest of amateur and professional
runners around the world due to their great accessibility and low economic cost. Specifically,
the vertical kilometer is a trend in trail running. In this modality, the athletes must complete
a course of an approximately 1000 m vertical climb in a maximum of 5000 m total race
length, although these parameters could change between different races, according to the
rules of the International Skyrunning Federation [4].
Research on key performance parameters, both in road and trail running, has been
a growing target of analysis by numerous health and sport science researchers. The aim
of these studies is to understand in more depth those factors correlated with running
Sensors 2023, 23, 9349. https://doi.org/10.3390/s23239349
https://www.mdpi.com/journal/sensors
Sensors 2023, 23, 9349
2 of 19
performance to later be able to apply this knowledge in the creation of personalized
trackers that can be implemented in phone apps and smartwatches. With technological
advances, many scientists have developed new promising concepts whose assessment
seems to be sensitive to physiological and biomechanical modifications during running
and which may be suitable real feedback measures of performance and training monitoring
in trail running and vertical kilometers. These concepts are the running economy, the
net metabolic power, the mechanical vertical center of mass power, the net mechanical
efficiency, the net metabolic cost of transport and the vertical net metabolic cost of transport.
Running economy is defined as the oxygen uptake (VO2) required to run a given
distance or run at a given submaximal velocity [5]. This parameter can also be defined and
calculated in energy terms as the amount of energy liberated per liter of oxygen, denomi-
nated in this case as net metabolic rate or power (Cmetab) (kcal·min−1·kg−1·or W·kg−1). It
is calculated by measuring the steady-state consumption of oxygen (VO2) and the respira-
tory exchange ratio [6] and is considered a physiological determinant of endurance running.
This variable is multifactorial, depending on metabolic, cardiorespiratory, biomechanical
and neuromuscular factors [7], such as heart rate, minute ventilation, substrate utilization,
muscle fiber type and core temperature, among many other variables, and is a new concept
that reflects the physiological and neuromuscular state of the athlete [8]. It is currently
considered more sensitive than VO2 itself when used to observe performance differences
between runners [7,9].
The mechanical vertical power of the center of mass (Cmec) is defined as the external
mechanical work performed to lift the body mass at each running stride, calculated by
multiplying the vertical running velocity by the weight of the subject. Recent studies related
to running power have found a linear relationship between running power and aerobic
power (VO2 consumption) [10,11]. In addition, lower limb power is related to running
spatiotemporal improvements (increased contact time), reduction in the energy cost of
running [12] and reduction in the increase in energy cost of running due to fatigue in trail
running [13]. Specifically, in vertical kilometers, runners must overcome extreme uphill
running slopes, lifting the center of body mass in each step more than in level running
by increasing the net mechanical work. This mechanism entails an increase in energy
expenditure and a poorer mechanical advantage for producing force against the ground by
the hip extensors [14].
Finally, from the previous concepts, the parameters of net mechanical efficiency, net
metabolic cost of transport and vertical net metabolic cost of transport have emerged. The
first authors to evaluate these parameters were Margaria et al., (1963) [15] and Minetti et al.,
(2002) [16]. They calculated the net metabolic cost of transport (both walking and running)
(cost of walking (Cw) and cost of running (Cr)) by dividing the metabolic power or rate
by running or walking velocity (vertical velocity for the vertical net metabolic cost of
transport (VCw and VCr)). This parameter is a key factor in road running [4] and describes
the amount of energy needed to transport a kilogram of body mass per unit of distance
covered (kcal·kg−1·km−1 or J·kg−1·m−1). In their studies, Margaria et al., (1963) [15] and
Minetti et al., (2002) [16] observed that the metabolic cost of running (Cr) was dependent
on gradient and independent of speed, except for the steepest positive slopes (above 15%
or 8.5◦).
Based on these data, subsequent studies have found a great increase in Cr between
slopes among runners, whose cause is still unknown, since uphill Cr correlates with neither
level Cr nor with biomechanical parameters, such as stride frequency, stride length and
body mass index [17]. Likewise, there is no correlation between either the initial Cr values
or the changes in Cr values before and after the trail running race with performance
time, in contrast to the observed correlation in road running [18]. The increase in Cr with
a positive incline is due to an increase in power output and greater muscular activity
at all joints, especially in the hip [19]. Unlike level running, where the center of mass
behavior oscillates cyclically and both potential and kinetic energy fluctuation are in-phase
during the stride [20], in uphill running above 15% (8.5◦), positive work predominately
Sensors 2023, 23, 9349
3 of 19
lifts the center of mass and decreases the use of elastic energy (the stretch–shortening cycle
mechanism disappears) and bouncing mechanisms [21,22]. Consequently, the metabolic
demand increases, coinciding with an increase in blood lactate values and cardiorespiratory
values [17,19,23].
In connection with the concepts of mechanical and metabolic power, Margaria et al.,
(1963) [15] and Minetti et al., (2002) [16] also introduced the concept of net mechanical
efficiency (Eff) by explaining the ratio of these two variables. In their analysis, they observed
that trained athletes were only 5–7% more efficient than non-athletes [15]. They predicted
that mechanical efficiency was approximately 22–24% with positive slopes above 15%
(8.5◦) and 25% above 20% (11.3◦), corresponding to concentric muscle contraction [15,16].
Peyré-Tartaruga et al., (2018) [24] proposed that overall efficiency in locomotion (walking
and running) is determined by muscular efficiency, defined as the fraction of metabolic
energy transformed into muscular mechanical work, and transmission efficiency, defined
as the fraction of muscular mechanical work utilized as total work. However, for practical
purposes, the concept net mechanical efficiency (Eff) is considered the fraction of metabolic
power transformed into mechanical power or total work. These authors also contended
that if the efficiency value was close to 25% (indicating pure concentric muscle efficiency),
it would suggest good efficiency transmission. If the value exceeded 25%, it would indicate
that passive elastic elements in series within muscles (fascial tissues) and tendons provided
either the same or significant negative work.
Based on the studies analyzed to date, most research has been conducted on a treadmill
in trail running, and any study of the vertical kilometer was executed through a field test.
For these reasons, the present study aims to determine the correlation with performance in
the previously mentioned concepts (Cmec, Cmetab, Cw, Cr, Vcw, VCr and Eff), as well as
to observe the effect of fatigue on these concepts during the progress of a vertical kilometer
field test.
2. Materials and Methods
2.1. Participants
Fifteen trained trail runners participated in the study (eleven males, four females).
Demographic, anthropometric and training level data are presented in Table 1. All runners
had been training regularly for more than 3 years, and none of them had a history of
musculoskeletal injuries in the last year. Before the experiment, all subjects were informed
about the objectives, benefits and risks of the investigation, and they signed an informed
consent form. The experimental protocol received approval from the University Ethics
Committee (Ref 005-19/20), and all procedures adhered to the principle of the Declaration
of Helsinki.
Table 1. Demographic, anthropometric and training level data.
Men
Women
Age (years)
22–38 *
19–35 *
28.4 ± 5.11
27.7 ± 6.70
Height (cm)
174 ± 4.54
163 ± 2.36
Body mass (kg)
69.8 ± 5.56
54 ± 4.08
BMI (kg/m2)
22.8 ± 1.63
20.2 ± 1.01
Running training duration per session (min)
52 ± 7.58
60 ± 21.6
Running training frequency per week (days/week)
4.40 ± 1.14
4.75 ± 1.26
Pre-test heart rate (bpm)
73.8 ± 10.7
79.5 ± 3.31
HR change (%)
16.1 ± 4.99
61.2 ± 56.6
VO2 peak (mL/kg/min)
65.8 ± 7.00
57.9 ± 6.61
Values: Mean ± SD. BMI: body mass index. HR change: percentage change in heart rate during the vertical
kilometer test. VO2 peak achieved in the vertical kilometer test. *: age range of participants.
Sensors 2023, 23, 9349
4 of 19
2.2. Procedure
Each participant completed a vertical kilometer (VK) route spanning 4.64 km with a
positive slope of 835 m. The vertical kilometer entails a continuous uphill test, comprising
natural segments with varying positive inclinations ranging from 0◦ to 20◦ on this specific
route. To facilitate analysis, the route was divided into three equal parts, each measuring
1.58 km, as illustrated in Figure 1. Within each of these segments, five sections with a
constant slope were chosen (0◦, 5◦, 10◦, 15◦, and 20◦ positive slope). Each section had to
last a minimum of 30 s to extract stable physiological data. Furthermore, to ensure data
stability, only the central 20 s of each section were analyzed, excluding the initial and final
portions of the positive slope.
Values: Mean ± SD. BMI: body mass index. HR change: percentage change in heart r
vertical kilometer test. VO2 peak achieved in the vertical kilometer test. *: age range o
2.2. Procedure
Each participant completed a vertical kilometer (VK) route spanning 4.6
positive slope of 835 m. The vertical kilometer entails a continuous uphill test
natural segments with varying positive inclinations ranging from 0° to 20° on
route. To facilitate analysis, the route was divided into three equal parts, eac
1.58 km, as illustrated in Figure 1. Within each of these segments, five section
stant slope were chosen (0°, 5°, 10°, 15°, and 20° positive slope). Each section
minimum of 30 s to extract stable physiological data. Furthermore, to ensure d
only the central 20 s of each section were analyzed, excluding the initial and fi
of the positive slope.
Figure 1. Vertical kilometer track. Race course divided into 3 sections of 1.58 km.
2.3. Measurements
2.3.1. Metabolic Data
Throughout the entire course, the runners were equipped with a porta
lyzer (Cosmed K5 (Rome, Italy)) to assess cardiorespiratory and metabolic re
breath-by-breath basis. This measurement was facilitated by a turbine flowm
to a properly fitted face mask. The gas analyzer was secured to the runner’s
harness, and the entire system weighted 900 g. To ensure time alignment, t
parameters from the gas analyzer (including GPS data) were synchronized a
the data logger. Calibration of the Cosmed system was performed before ea
ment, using a calibration syringe (3L) for the turbine. The oxygen (O2) and ca
(CO2) sensors of the gas analyzer were also calibrated to ambient air condit
O2 and 0.03% CO2), along with delay calibration. Each experimental day comm
determining the metabolic rate during a 10-min standing trial. Subsequently
ygen consumption (VO2) and carbon dioxide production (VCO2) were measur
Cosmed K5 analyzer. For statistical analysis, the data for each slope and sect
d
th
l
t d 20
i t
l
1600 m
1400 m
1200 m
1000 m
800 m
Figure 1. Vertical kilometer track. Race course divided into 3 sections of 1.58 km.
2.3. Measurements
2.3.1. Metabolic Data
Throughout the entire course, the runners were equipped with a portable gas analyzer
(Cosmed K5 (Rome, Italy)) to assess cardiorespiratory and metabolic responses on a breath-
by-breath basis. This measurement was facilitated by a turbine flowmeter attached to
a properly fitted face mask. The gas analyzer was secured to the runner’s back using a
harness, and the entire system weighted 900 g. To ensure time alignment, the analyzed
parameters from the gas analyzer (including GPS data) were synchronized and stored in the
data logger. Calibration of the Cosmed system was performed before each measurement,
using a calibration syringe (3L) for the turbine. The oxygen (O2) and carbon dioxide
(CO2) sensors of the gas analyzer were also calibrated to ambient air conditions (20.93%
O2 and 0.03% CO2), along with delay calibration. Each experimental day commenced
with determining the metabolic rate during a 10-min standing trial. Subsequently, rates of
oxygen consumption (VO2) and carbon dioxide production (VCO2) were measured using
the Cosmed K5 analyzer. For statistical analysis, the data for each slope and section were
averaged over the selected 20-s intervals.
Sensors 2023, 23, 9349
5 of 19
2.3.2. Calculations
The calculation of mechanical vertical center of mass (COM) power (Watts/kg) utilized
GPS velocity and incline, as expressed in (Equation (1)):
Mechanical vertical COM power = g × v × sin (θ)
(1)
where θ represents the incline in degrees, and v is the instantaneous velocity in m/s.
Net metabolic power (Watts/kg) was calculated from running respiratory measure-
ments using the Peronnet and Massicot equation [6], adjusted by subtracting the standing
metabolic rate measured 10 min before the test. The calculation is outlined in (Equation (2)):
Net Metabolic power = ((16.89 × VO2 + 4.84 × VCO2)/kg) − standing metabolic rate
(2)
The net mechanical efficiency was derived by dividing the mechanical vertical COM
power by the net metabolic power, as illustrated in (Equation (3)) [25]:
Net mechanical efficiency = Mechanical vertical COM power/Net metabolic power (3)
The net metabolic cost of transport (J/kg/m) was computed by dividing the net
metabolic power by the running velocity, representing the mean net metabolic cost per unit
distance traveled parallel to the running surface. (Equation (4)) summarizes this calculation:
Net Metabolic COT = Net metabolic power/v
(4)
The vertical net metabolic cost of transport (J/kg/m) was determined by dividing the
net metabolic power by vertical velocity, factored by the mean net metabolic cost to ascend
a vertical meter. (Equation (5)) outlines this computation:
Vertical Net Metabolic COT = Net metabolic power/v × sin (θ)
(5)
2.4. Statistical Analysis
The following statistical analysis of the data was conducted:
•
Normality testing: the Shapiro–Wilk test was used to assess the normality of the variables.
•
Gender and performance level comparison: A T-student parametric test was employed
to compare gender and performance level differences. The sample was divided into
quartiles based on the final test time, and values from the first quartile were compared
to the remaining quartiles.
•
Comparison of assessed variables: A two-factor repeated-measures ANOVA was
utilized to compare means across multiple analyzed variables. The analysis compared
three sections and five positive slopes in each section. Before applying ANOVA, the
Mauchly’s sphericity test was performed. If sphericity was rejected, the univariated
F-statistic was used, adjusted with the Greenhouse–Geisser correction index. Bonfer-
roni’s post hoc analysis was performed when significant differences were found for
pairwise comparison.
•
Statistical power and effect size determination: The statistical power (SP) and effect
size (partial eta squared, ηp2) were determined. The effect size was categorized as
trivial (ηp2 ≤ 0.01), small (0.01 ≤ ηp2 < 0.06), moderate (0.06 ≤ ηp2 < 0.14) or large
(ηp2 ≥ 0.14) [26].
•
Relationship analysis with final uphill time: Multiple regression and correlation
models were calculated using an “intro” method. Mechanical vertical COM power
was considered the dependent variable, and net metabolic power and vertical net
metabolic cost of transport were the independent variables in the three VK sections.
The entry and exit criteria were set at F probabilities greater than 0.05 and 0.10, re-
spectively. The residual linearity and independence assumptions were checked with
the Durbin–Watson test. The homoscedasticity was studied in a partial standardized
Sensors 2023, 23, 9349
6 of 19
residual-standardized prediction plot. The method of Bland and Altman was used to
determine systematic bias and random error in the prediction model, as well as the
lower and upper limits of agreement (1.96 × SD). The multicollinearity was estimated
using a variance inflation factor (VIF), with values greater than 10 considered exces-
sive. Influential cases (Cook’s distance > 1) and atypical cases (residual > 3 standard
deviations) were removed from the analysis.
•
A significance level of p < 0.05 was established. All statistical tests were conducted
using the statistical package SPSS version 25.0 (SPSS, Chicago, IL, USA).
3. Results
Mean values for the three sections and five slope conditions are presented in Table 2.
Regarding gender, no statistical differences were observed. Furthermore, when analyzing
the aforementioned variables based on runner performance level (vertical kilometer final
time) (Table 3), significant differences emerged between the first quartile and the remaining
quartiles in the variables mechanical vertical COM power, net metabolic power, velocity
and vertical velocity. On the other hand, no significant differences were identified in
the variables net mechanical efficiency, net metabolic cost of transport and vertical net
metabolic cost of transport.
A repeated-measures ANOVA revealed significant differences between sections, in-
cline and the interaction of section x incline in all the variables presented in Table 4. These
findings indicate a “Large” effect size of fatigue on all variables as the VK progresses.
Additional distinctions are detailed in Table 5 through percentages.
Conducting a two-way ANOVA with performance level as a factor (first quartile versus
remaining quartiles), significant differences were only identified in mechanical vertical
COM power (incline p < 0.001, SP = 0.967, ηp2 = 0.320) and vertical velocity (incline p < 0.001,
SP = 0.972, ηp2 = 0.307). No significant differences were observed in net metabolic power,
net mechanical efficiency, net metabolic COT, vertical net metabolic COT, and velocity. The
percentage of change with corresponding p-values is depicted in Figures 2 and 3.
Sensors 2023, 23, 9349
7 of 19
Table 2. Descriptive data of values in the three sections and five slope conditions.
Section 1
Section 2
Section 3
0◦
5◦
10◦
15◦
20◦
0◦
5◦
10◦
15◦
20◦
0◦
5◦
10◦
15◦
20◦
Velocity (m/s)
3.42 ± 0.39
1.98 ± 0.34
1.52 ± 0.19
1.35 ± 0.21
1.00 ± 0.13
2.26 ± 0.38
1.95 ± 0.30
1.40 ± 0.24
1.03 ± 0.14
0.88 ± 0.14
2.39 ± 0.60
1.67 ± 0.27
1.33 ± 0.19
1.00 ± 0.22
0.73 ± 0.16
Vertical velocity
(m/s)
0 ± 0
0.17 ± 0.03
0.26 ± 0.03
0.35 ± 0.05
0.34 ± 0.05
0 ± 0
0.17 ± 0.03
0.24 ± 0.04
0.27 ± 0.04
0.30 ± 0.05
0 ± 0
0.14 ± 0.02
0.23 ± 0.03
0.26 ± 0.06
0.25 ± 0.06
RER
0.94 ± 0.08
0.82 ± 0.08
0.88 ± 0.10
0.90 ± 0.10
0.81 ± 0.08
0.82 ± 0.08
0.81 ± 0.09
0.81 ± 0.08
0.80 ± 0.08
0.80 ± 0.08
0.79 ± 0.08
0.80 ± 0.08
0.78 ± 0.08
0.79 ± 0.07
0.80 ± 0.08
Mechanical
vertical COM
power (W/kg)
0 ± 0
1.69 ± 0.29
2.58 ± 0.32
3.42 ± 0.54
3.38 ± 0.47
0 ± 0
1.67 ± 0.26
2.37 ± 0.40
2.62 ± 0.36
2.94 ± 0.48
0 ± 0
1.42 ± 0.23
2.25 ± 0.33
2.54 ± 0.55
2.46 ± 0.55
Net metabolic
power (W/kg)
17 ± 2.41
17.4 ± 3.10
18.8 ± 2.97
18.7 ± 2.80
17.5 ± 2.69
16.3 ± 2.89
16.4 ± 2.84
16.3 ± 2.93
16.5 ± 2.82
16.3 ± 2.97
15.1 ± 3.33
16 ± 2.92
16.3 ± 2.65
16.7 ± 2.77
16.8 ± 2.91
Net mechanical
efficiency
0 ± 0
9.88 ± 1.54
13.9 ± 2.15
18.6 ± 3.10
19.5 ± 2.83
0 ± 0
10.3 ± 1.66
14.8 ± 2.91
16 ± 1.92
18.1 ± 2.05
0 ± 0
9.03 ± 1.37
14.0 ± 1.75
15.6 ± 4.48
14.7 ± 2.23
Net metabolic cost
of transport
(J/kg/m)
5.01 ± 0.72
8.84 ± 1.37
12.4 ± 1.81
14.0 ± 2.36
17.4 ± 2.16
7.26 ± 0.97
8.47 ± 1.19
11.8 ± 2.15
16.0 ± 1.96
18.7 ± 2.03
6.77 ± 2.47
9.66 ± 1.39
12.3 ± 1.54
17.2 ± 4.19
23.4 ± 3.72
Vertical net
metabolic cost of
transport (J/kg/m)
0 ± 0
101.6 ± 15.7
71.9 ± 10.5
54.2 ± 9.15
51.0 ± 6.32
0 ± 0
97.3 ± 13.6
68.5 ± 12.4
62.1 ± 7.59
54.8 ± 5.93
0 ± 0
111.1 ± 15.9
71.3 ± 8.92
66.8 ± 16.2
68.5 ± 10.9
Values: mean ± standard deviation. RER: respiratory exchange rate; COM: center of mass.
Table 3. Differences in values in the three sections and five slope conditions between first quartile and remaining quartiles.
Section 1
Section 2
Section 3
0◦
5◦
10◦
15◦
20◦
0◦
5◦
10◦
15◦
20◦
0◦
5◦
10◦
15◦
20◦
Vertical
velocity (m/s)
1st quartile (n = 5)
0 ± 0
0.20 ± 0.02 *
0.29 ± 0.02 *
0.39 ± 0.06 *
0.37 ± 0.02 *
0 ± 0
0.19 ± 0.01 *
0.26 ± 0.04
0.30 ± 0.02 *
0.35 ± 0.03 **
0 ± 0
0.16 ± 0.02 *
0.26 ± 0.03 *
0.28 ± 0.02 *
0.30 ± 0.01 **
Remaining
quartiles (n = 9)
0 ± 0
0.16 ± 0.02
0.26 ± 0.03
0.33 ± 0.04
3.33 ± 0.05
0 ± 0
0.16 ± 0.02
0.23 ± 0.04
0.25 ± 0.03
0.28 ± 0.03
0 ± 0
0.14 ± 0.02
0.21 ± 0.02
0.24 ± 0.06
0.22 ± 0.04
Velocity (m/s)
1st quartile
3.74 ± 0.23 *
2.26 ± 0.27 *
1.67 ± 0.15 *
1.53 ± 0.22 *
1.10 ± 0.07 *
2.64 ± 0.14 *
2.25 ± 0.15 *
1.52 ± 0.24
1.17 ± 0.09 **
1.02 ± 0.09 *
2.79 ± 0.24 *
1.90 ± 0.24 *
1.50 ± 0.20 *
1.10 ± 0.68 *
0.89 ± 0.03 **
Remaining
quartiles
3.33 ± 0.41
1.86 ± 0.28
1.49 ± 0.22
1.27 ± 0.15
0.96 ± 0.14
2.10 ± 0.32
1.83 ± 0.27
1.35 ± 0.22
0.98 ± 0.12
0.81 ± 0.09
2.22 ± 0.61
1.58 ± 0.23
1.23 ± 1.10
0.94 ± 0.25
0.65 ± 0.13
Mechanical
vertical COM
power (W/kg)
1st quartile
0 ± 0
1.92 ± 0.23 *
2.82 ± 0.25 *
3.87 ± 0.56 *
3.62 ± 0.23 *
0 ± 0
1.92 ± 0.13 *
2.58 ± 0.41 *
2.97 ± 0.24 *
3.43 ± 0.32 **
0 ± 0
1.62 ± 0.20 *
2.54 ± 0.35 *
2.77 ± 0.17
2.98 ± 0.12 **
Remaining
quartiles
0 ± 0
1.56 ± 0.24
2.44 ± 0.28
3.19 ± 0.37
3.25 ± 0.53
0 ± 0
1.53 ± 0.21
2.25 ± 0.37
2.43 ± 0.26
2.66 ± 0.30
0 ± 0
1.32 ± 0.18
2.09 ± 0.19
2.42 ± 0.65
2.17 ± 0.48
Net metabolic
power (W/kg)
1st quartile
19 ± 0.64 *#
20.1 ± 2.62 *#
22 ± 1.10 **#
21.6 ± 1.66 **#
20.1 ± 1.32 *#
19.5 ± 0.93 **#
19.6 ± 1.61 **#
19.7 ± 1.35 **#
19.7 ± 1.24 **#
19.8 ± 1.46 **#
18.6 ± 2.16 **#
19.4 ± 1.42 **#
19.3 ± 1.24 **#
19.9 ± 1.11 **#
20.2 ± 1.46 **#
Remaining
quartiles
15.9 ± 2.31
15.8 ± 2.21
17.1 ± 2.06
17.1 ± 1.81
16 ± 2.08
14.5 ± 1.75
14.6 ± 1.31
14.4 ± 1.28
14.8 ± 1.52
14.4 ± 1.19
13.1 ± 1.88
14.1 ± 1.13
14.6 ± 1.25
14.9 ± 1.42
14.9 ± 1.25
Values: mean ± standard deviation. COM: center of mass. * p-value < 0.05. ** p-value < 0.001. #: strong effect size (g Hedges > 0.8).
Sensors 2023, 23, 9349
8 of 19
Table 4. Repeated-measures ANOVA results.
p-Value
Power (SP)
Effect Size (ηp2)
Vertical velocity
(m/s)
Section
<0.001
1
0.779
Large
Slope
<0.001
1
0.973
Large
Interaction
<0.001
1
0.463
Large
Velocity (m/s)
Section
<0.001
1
0.872
Large
Slope
<0.001
1
0.949
Large
Interaction
<0.001
1
0.654
Large
Mechanical
vertical COM
power (W/kg)
Section
<0.001
1
0.776
Large
Slope
<0.001
1
0.972
Large
Interaction
<0.001
1
0.452
Large
Net metabolic
power (W/kg)
Section
<0.001
0.993
0.600
Large
Slope
<0.001
1
0.489
Large
Interaction
<0.001
0.991
0.243
Large
Net mechanical
efficiency
Section
<0.001
1
0.626
Large
Slope
<0.001
1
0.969
Large
Interaction
<0.001
0.994
0.379
Large
Net metabolic
cost of transport
(J/kg/m)
Section
<0.001
1
0.706
Large
Slope
<0.001
1
0.952
Large
Interaction
<0.001
0.997
0.406
Large
Vertical net
metabolic cost of
transport (J/kg/m)
Section
<0.001
1
0.648
Large
Slope
<0.001
1
0.972
Large
Interaction
<0.001
0.964
0.304
Large
Table 5. Differences in values between sections and inclines.
Sections 1 vs. 2
Sections 1 vs. 3
Sections 2 vs. 3
Vertical velocity (m/s)
5◦
=0%
↓21.4% *
↓21.4% *
10◦
↓8.33%
↓13% *
↓4.35%
15◦
↓29.6% **
↓35.6% **
↓3.85%
20◦
↓13.3% *
↓36% **
↓20% *
Velocity (m/s)
0◦
↓51.3% **
↓43.1% **
↑5.75%
5◦
↓1.53%
↓18.6% *
↓16.8% *
10◦
↓8.6%
↓14.3% *
↓5.3%
15◦
↓31% **
↓35% **
↓3%
20◦
↓13.6% *
↓37% **
↓20.5% *
Mechanical vertical
COM power (W/kg)
5◦
↓1.19%
↓19% *
↓17% *
10◦
↓8.86%
↓14.7% *
↓5.33%
15◦
↓30.5% **
↓34.6% **
↓3.15%
20◦
↓15% *
↓37.4% **
↓19.5% *
Net metabolic power
(W/kg)
0◦
↓4.3%
↓12.6%
↓7.95%
5◦
↓6.10%
↓8.75%
↓2.50%
10◦
↓15.3% **
↓15.3% **
=0%
15◦
↓13.3% **
↓12% **
↑1.21%
20◦
↓7.36%
↓4.17%
↑3.07%
Net mechanical
efficiency
5◦
↑4.25%
↓9.41% *
↓14.1% *
10◦
↑6.47%
↓0.72%
↓5.71%
15◦
↓16.2% *
↓19.2% *
↓2.56%
20◦
↓7.73%
↓32.6% **
↓23.1% *
Sensors 2023, 23, 9349
9 of 19
Table 5. Cont.
Sections 1 vs. 2
Sections 1 vs. 3
Sections 2 vs. 3
Net metabolic cost of
transport (J/kg/m)
0◦
↑44.9% **
↑35.1% *
↓7.24%
5◦
↓4.37%
↑9.28% *
↑14% *
10◦
↓5.08%
↓0.81%
↑4.24%
15◦
↑14.3% *
↑22.8% *
↑7.5%
20◦
↑7.47%
↑34.5% **
↑25.1% *
Vertical net metabolic
cost of transport
(J/kg/m)
5◦
↓4.42%
↑9.35% *
↑14.2% *
10◦
↓4.96%
↓0.84%
↑4.10%
15◦
↑14.6% *
↑23.2% *
↑7.57%
20◦
↑7.45%
↑34.3% **
↑25% *
% of change in mean values with p-values of Bonferroni post hoc. Up and down arrows correspond to increases
and decreases respectively; equal symbols indicate no change. * p-value < 0.05. ** p-value < 0.001.
Sensors 2023, 23, x FOR PEER REVIEW
9 of 19
velocity. The percentage of change with corresponding p-values is depicted in Figures 2
and 3.
(a)
(b)
Figure 2. Percentage of change in mechanical vertical center of mass power between slopes. * p-value
< 0.05. ** p-value < 0.001. (a) Percentage of change in the runners of the first quartile. The figures are
arranged according to VK section (a1: first section; a2: second section; a3: third section). (b) Percent-
age of change in the runners of the remaining quartiles (b1: First section; b2: second section; b3: third
section).
1.62
2.54
2.77
2.98
0,00
0,50
1,00
1,50
2,00
2,50
3,00
3,50
5º
10º
15º
20º
Section 3
1st Quartile
3
1.93
2.83
3.87
3.63
0,00
0,50
1,00
1,50
2,00
2,50
3,00
3,50
4,00
4,50
5º
10º
15º
20º
Section 1
1st Quartile
1
88%**
28%*
1.92
2.58
2.97
3.43
0,00
0,50
1,00
1,50
2,00
2,50
3,00
3,50
4,00
5º
10º
15º
20º
Section 2
1st Quartile
2
46%**
100%**
37%*
34%*
55%**
79%**
33%*
57%**
71%**
84%**
9%
17%
7%
− 7%
15%
15%
1.65
2.45
3.19
3.25
0,00
0,50
1,00
1,50
2,00
2,50
3,00
3,50
5º
10º
15º
20º
Section 1
Remaining Quartiles
1
1.53
2.25
2.43
2.67
0,00
0,50
1,00
1,50
2,00
2,50
3,00
5º
10º
15º
20º
Section 2
Remaining Quartiles
2
1.32
2.09
2.42
2.18
0,00
0,50
1,00
1,50
2,00
2,50
3,00
5º
10º
15º
20º
Section 3
Remaining Quartiles
3
48%**
93%**
97%**
30%*
33%*
8%
47%**
59%**
74%**
2%
19%
10%
− 11%
4%
16%
65%**
83%**
58%**
5°
10°
15°
20°
5°
10°
15°
20°
5°
10°
15°
20°
5°
10°
15°
20°
5°
10°
15°
20°
5°
10°
15°
20°
4.50
4.00
3.50
3.00
2.50
2.00
1.50
1.00
0.50
0.00
3.50
0.00
3.00
2.50
0.50
2.00
1.50
1.00
3.50
4.00
3.00
2.50
2.00
1.50
1.00
0.50
0.00
3.00
2.50
2.00
1.50
1.00
0.50
0.00
3.00
3.50
2.50
2.00
1.50
1.00
0.50
0.00
3.00
2.50
2.00
1.50
1.00
0.50
0.00
Figure 2.
Percentage of change in mechanical vertical center of mass power between slopes.
* p-value < 0.05. ** p-value < 0.001. (a) Percentage of change in the runners of the first quartile.
The figures are arranged according to VK section (a1: first section; a2: second section; a3: third sec-
tion). (b) Percentage of change in the runners of the remaining quartiles (b1: First section; b2: second
section; b3: third section).
Sensors 2023, 23, 9349
10 of 19
Sensors 2023, 23, x FOR PEER REVIEW
10 of 19
(a)
(b)
Figure 3. Percentage of change in vertical velocity between slopes. * p-value < 0.05. ** p-value < 0.001.
(a) Percentage of change in the runners of the first quartile. The figures are arranged according to
VK section (a1: first section; a2: second section; a3: third section). (b) Percentage of change in the
runners of the remaining quartiles (b1: first section; b2: second section; b3: third section).
A multiple regression analysis was conducted to predict mechanical vertical COM
power from the remaining variables. The analysis revealed that the variable mechanical
vertical COM power can be statistically significantly predicted using net metabolic power
and vertical net metabolic cost of transport. This relationship held true across all sections
and slopes.
The resulting model can be expressed algebraically as follows (Equation (6)):
Mechanical Vertical COM power = 𝛼𝐶𝑚𝑒𝑡𝑎𝑏 + 𝛽𝑉𝐶𝑟 + 𝛾
(6)
where Cmetab represents net metabolic power (Equation (2)) and VCr stands for vertical
net metabolic cost of transport (Equation (5)).
0.16
0.26
0.28
0.30
0,00
0,05
0,10
0,15
0,20
0,25
0,30
0,35
5º
10º
15º
20º
Section 3
1st Quartile
3
0.20
0.29
0.39
0.37
0,00
0,05
0,10
0,15
0,20
0,25
0,30
0,35
0,40
0,45
5º
10º
15º
20º
Section 1
1st Quartile
1
0.20
0.26
0.30
0.35
0,00
0,05
0,10
0,15
0,20
0,25
0,30
0,35
0,40
5º
10º
15º
20º
Section 2
1st Quartile
2
45%**
− 5%
95%**
85%**
30%*
34%*
27%*
50%**
75%**
15%
35%*
17%
62%**
75%**
7%
87%**
8%
15%
0.16
0.26
0.33
0.33
0,00
0,05
0,10
0,15
0,20
0,25
0,30
0,35
5º
10º
15º
20º
Section 1
Remaining Quartiles
1
0.16
0.23
0.25
0.28
0,00
0,05
0,10
0,15
0,20
0,25
0,30
5º
10º
15º
20º
Section 2
Remaining Quartiles
2
0.14
0.21
0.24
0.22
0,00
0,05
0,10
0,15
0,20
0,25
0,30
5º
10º
15º
20º
Section 3
Remaining Quartiles
3
106%**
106%**
62%**
27%*
27%**
0%
44%**
56%**
75%**
9%
22%*
12%
50%**
71%**
57%**
14%
5%
− 9%
5°
10°
15°
20°
5°
10°
15°
20°
5°
10°
15°
20°
5°
10°
15°
20°
5°
10°
15°
20°
5°
10°
15°
20°
0.00
0.45
0.40
0.35
0.30
0.25
0.20
0.15
0.10
0.05
0.00
0.00
0.00
0.00
0.00
0.35
0.30
0.25
0.20
0.15
0.10
0.05
0.40
0.35
0.30
0.25
0.20
0.15
0.10
0.05
0.30
0.25
0.20
0.15
0.10
0.05
0.35
0.30
0.25
0.20
0.10
0.15
0.05
0.30
0.20
0.25
0.15
0.10
0.05
Figure 3. Percentage of change in vertical velocity between slopes. * p-value < 0.05. ** p-value < 0.001.
(a) Percentage of change in the runners of the first quartile. The figures are arranged according to VK
section (a1: first section; a2: second section; a3: third section). (b) Percentage of change in the runners
of the remaining quartiles (b1: first section; b2: second section; b3: third section).
A multiple regression analysis was conducted to predict mechanical vertical COM
power from the remaining variables. The analysis revealed that the variable mechanical
vertical COM power can be statistically significantly predicted using net metabolic power
and vertical net metabolic cost of transport. This relationship held true across all sections
and slopes.
The resulting model can be expressed algebraically as follows (Equation (6)):
Mechanical Vertical COM power = αCmetab + βVCr + γ
(6)
Sensors 2023, 23, 9349
11 of 19
where Cmetab represents net metabolic power (Equation (2)) and VCr stands for vertical
net metabolic cost of transport (Equation (5)).
The adjusted R2 of the multiple linear regression indicates that 94% of the variation in
mechanical vertical COM power is explained by the proposed model (R2adjusted = 0.942).
The scatter plot for this model is illustrated in Figure 4. The model reached a significance
level of p < 0.001. All variables included in the model exhibited a significance level below
0.001, suggesting their retention in the considered model. The Durbin–Watson test fell
within the critical interval (1 < D–W < 3), allowing the acceptance of residual linearity
and independence assumptions. However, Bland and Altman plots (Figure 5) revealed
randomly distributed residuals concerning the average net mechanical vertical COM power
predicted values. Only one value outside ±1.96 × SD was observed, and the residuals
exhibited normal distribution based on the Shapiro–Wilk test (SW = 0.941; p = 0.434).
All values presented a variance inflation factor (VIF) lower than 10 units. Therefore, the
multicollinearity assumption is satisfied.
Sensors 2023, 23, x FOR PEER REVIEW
11 of 19
The adjusted R2 of the multiple linear regression indicates that 94% of the variation
in mechanical vertical COM power is explained by the proposed model (R2adjusted = 0.942).
The scatter plot for this model is illustrated in Figure 4. The model reached a significance
level of p < 0.001. All variables included in the model exhibited a significance level below
0.001, suggesting their retention in the considered model. The Durbin–Watson test fell
within the critical interval (1 < D–W < 3), allowing the acceptance of residual linearity and
independence assumptions. However, Bland and Altman plots (Figure 5) revealed ran-
domly distributed residuals concerning the average net mechanical vertical COM power
predicted values. Only one value outside ±1.96 × SD was observed, and the residuals ex-
hibited normal distribution based on the Shapiro–Wilk test (SW = 0.941; p = 0.434). All
values presented a variance inflation factor (VIF) lower than 10 units. Therefore, the mul-
ticollinearity assumption is satisfied.
Figure 4. Scatter plot of the multiple linear regression model. Each data point represents the value
of a subject in the study.
Figure 4. Scatter plot of the multiple linear regression model. Each data point represents the value of
a subject in the study.
Sensors 2023, 23, x FOR PEER REVIEW
12 of 19
Figure 5. Bland–Altman plot of the multiple linear regression model. Each data point represents
the value of a subject in the study.
The results of the Bland–Altman analysis indicate the absence of systematic biases
and random errors in our regression model, attributed to the randomness of the scatter-
plot dispersion and the absence of outliers.
Based on these results the following prediction equations are derived (Equations (7)
-0,15
-0,1
-0,05
0
0,05
0,1
0,15
U nstandarized residuals
A verage net m echanical vertical C O M pow erpredicted values
1.96×SD=0.12
2
2.1
2.2
2.3
2.4
2.6
2.5
2.7
2.8
2.9
3
0.1
0.05
0
−0.05
−0.15
−0.1
0.15
Figure 5. Bland–Altman plot of the multiple linear regression model. Each data point represents the
value of a subject in the study.
Sensors 2023, 23, 9349
12 of 19
The results of the Bland–Altman analysis indicate the absence of systematic biases
and random errors in our regression model, attributed to the randomness of the scatterplot
dispersion and the absence of outliers.
Based on these results, the following prediction equations are derived (Equations (7)
and (8)):
Mechanical Vertical COM power = 0.133 × Cmetab − 0.030 × VCr + 2.376
(7)
Net metabolic power =
V × g × sinθ − 2.376
0.133 − 0.030 × (V × g × sinθ)−1
(8)
Through mathematical calculation, the obtained algebraic expression allows us to
calculate the value of net metabolic power solely from the subject’s vertical velocity (parallel
velocity × sin θ (positive slope)).
4. Discussion
Metabolic efforts in trail running have recently become a significant focus of research,
with studies conducted in both ultra-distance events and short trail running. In the ma-
jority of these studies, simulations of race slopes have been conducted using treadmill
tests [27–29]. However, the metabolic demand appears to differ when the test is conducted
outdoors, potentially making it a more suitable method [30]. To our knowledge, there are
no metabolic studies during a vertical kilometer field test simulating a real race.
For these reasons, the primary objective of this study was to evaluate new concepts
such as mechanical vertical COM power, net metabolic power, net metabolic cost of trans-
port, vertical net metabolic cost of transport, and net mechanical efficiency during a real
outdoor vertical kilometer field test, examining their changes with fatigue and the perfor-
mance level of the athletes. The secondary goal was to analyze their relationships with the
final time of the test.
4.1. Vertical Kilometer Performance Analysis
The T-test results showed no significant differences between genders, while revealing
distinctions based on the subjects’ performance level, as despicted in Table 3.
Concerning the mean value differences between the first quartile and the remaining
quartiles, better mean values were observed in top-level runners across all sections and
inclines, achieving higher values in mechanical vertical COM power, net metabolic power,
velocity and vertical velocity. The results suggest that better runners can apply more force
and achieve greater vertical velocities as the slope increases. These disparities in power and
vertical velocity persist throughout the entire duration of the VK. These outcomes align
with expectations, as several researchers have observed that uphill running requires an
increase in net mechanical work to increase the potential energy of the body, with concurrent
increases in parallel propulsive force peaks and impulses with positive grades [31], since
the bouncing mechanism gradually disappears as speed and slope increase [22]. The hip
and knee joints are identified as the primary contributors to the augmented mechanical
power [14,32]. Additionally, in short trail running, it has been observed that local endurance
of knee extensors, assessed through repeated maximal concentric contractions, is a key
performance factor in uphill running sections [33].
Net metabolic power reflects the instantaneous energy requirement for running, and it
has been observed to increase linearly with speed in VK runners [34], attributable to the rise
in O2 consumption and CO2 production. The higher metabolic power values among first
quartile athletes are primarily explained by their greater velocity, stemming from either
enhanced cardiorespiratory development or greater strength and power values. Moreover,
the same study suggest that running is more efficient than walking above 0.8 m/s [34].
This reference value is crucial, as first-quartile runners could maintain speeds greater
than 0.8 m/s with 20◦ positive grade in the section 3 of our VK test, while the remaining
Sensors 2023, 23, 9349
13 of 19
quartiles’ runners could not. This decision to walk instead of running may partly account
for the observed difference in test performance.
Regarding the remaining variables, no significant differences were found based on
performance level. Our results align with other studies where no differences were identified
in the cost of running [35], and only a 5–7% difference in efficiency values [15] was observed
among trail runners of different levels. The minimal variation in the net metabolic cost
of transport in a real VK race could indicate that, despite first-quartile runners exhibiting
higher metabolic power, their ability to attain higher speeds resulted in comparable cost
of transport. This observation implies that net mechanical vertical COM power and net
metabolic power may serve as more informative indicators of trail running performance
compared to net metabolic cost of transport, as suggested by the existing literature [35].
These variables could prove more suitable for real-time tracking outdoors, utilizing po-
tentiometers [36] or mobile applications, or for analyzing average values in both men and
women to observe changes with training.
4.2. The Impact of Fatigue on the Vertical Kilometer
Analyzing the impact of fatigue throughout the progression of the VK (Table 4), we
observed a deterioration in mean values across all monitored variables, occurring with
all slopes, particularly notable between the first section and subsequent sections, and to a
lesser extent between the second and third sections. The changes were more pronounced
with steeper inclines (20◦). There was a reduction in velocity and vertical velocity, possibly
associated with the diminished ability to apply force (indicated by lower mechanical
vertical COM power values). This reduction was more significant between the first and
third sections, especially with 15◦ and 20◦ inclines, which are the most demanding due to
lower use of elastic energy [21,37] and biomechanical changes during the transition from
running to walking [38].
This power loss could stem from central fatigue (decreased amplitude and frequency
of motor unit recruitment) or peripheral fatigue (alterations in potential transmission
along the sarcolemma, excitation–contraction coupling and actin–myosin myofilament
interaction) [39]. Both types of fatigue might be implicated based on previous findings in
ultra-trail running [39–43].
Decreases in mechanical vertical COM power values could be attributed to fatigue
in both plantar flexors and knee extensor muscles. Recent studies suggest that central
fatigue tends to affect knee extensors more, while peripheral fatigue affects the plantar
flexors [39,41,44]. However, caution is warranted in applying these conclusions to the VK,
as these data were observed after an ultra-marathon.
A potential factor contributing to the onset of fatigue, particularly of central origin as
posited by the central command theory [45], is muscle damage and inflammation. However,
Pokora et al., (2014) [46] did not observe changes in creatine kinase (a marker of muscle
damage) after 1 h of uphill running (10◦) at 60%VO2max. Therefore, investigating muscle
damage as a cause of fatigue in uphill running requires further exploration [46].
Decreases in metabolic power values were also observed, possibly caused by impair-
ments in running biomechanics (such as increased step frequency, ankle joint changes and
duty-free alterations) [39], arising from neuromuscular fatigue and behavioral changes
in runners, especially with 15◦ and 20◦inclines, choosing gaits that minimize metabolic
cost [47].
Concerning net metabolic COT and vertical net metabolic COT, both continuously
increased across all sections with steeper uphill inclines due to greater loss of velocity
than metabolic power values as the test progressed. This suggests that neuromuscular,
rather than cardiorespiratory factors, may be the primary contributors to the decline in
performance in the VK. These increases align with observations in the literature after short-
distance running races [48,49], 1 h of treadmill running [50] and the vertical kilometer [34].
Multiple reasons have been proposed for this increase in COT. Firstly, the steep inclines
of the VK, coupled with a decrease in velocity, induce changes in running biomechanics,
Sensors 2023, 23, 9349
14 of 19
such as decreased step length, increased non-optimal step frequency, mid- to fore-foot
strike patterns, and decreased leg stiffness, all associated with increased COT [9,51–53].
Prolonged running step contact times (“Groucho running” pattern concept) [54] could
impair spring-like bouncing and elevate the COT due to changes in potential-kinetic energy
savings [34,55]. These biomechanical changes may be induced by neuromuscular fatigue
(reflected in decreased mechanical power) [39,56] or serve as a protective mechanism to
reduce running impacts [57].
Regarding net mechanical efficiency changes, this variable decreased due to greater
losses in mechanical power than metabolic power. The substantial and continuous losses in
mechanical power could signify a decrease in workload due the loss of velocity, providing
a significant limitation to performance due to the inability to utilize maximum metabolic
potential. This theory is supported by data from Ettema et al., (2009) [58], who stated that
power output is the main determinant of efficiency (more power leads to more efficiency
and vice versa), owing to a greater utilization of metabolic power in running. The imbalance
between mechanical power and metabolic power, resulting in a decrease in net mechanical
efficiency, could be attributed to decreased energy transduction (due to decreased speed
and stretch-shortening cycle) coupled with an increase in respiratory cost [24].
4.3. Examining Fatigue Effects Based on Runners’ Performance Levels
When examining the impact of fatigue based on the runners’ performance levels, we
observed differential changes in only two variables, namely mechanical vertical COM
power (Figure 2) and vertical velocity (Figure 3). Notably, elite runners demonstrated a
better ability to sustain power values across all slopes, particularly evident with 10◦ and
20◦ inclines, resulting in more pronounced differences in power values between slopes.
This phenomenon suggests their enhanced capability to exert force consistently across
all slopes throughout the entire race. Similarly, top-level runners exhibited a superior
ability to maintain vertical velocity values across all inclines, likely attributable to their
heightened application of force throughout the entire VK. These findings align with prior
research indicating a significant correlation between performance in short trail running
races and neuromuscular capacity, as assessed by isometric knee extensor muscle torque,
maximal theoretical force and maximal power from the force–velocity curve [59]. This
underscores the importance of incorporating resistance training [60], uphill interval run-
ning training [61] and pulled running training [62] to enhance power and neuromuscular
function [39,62] in runners. Additionally, it emphasizes the significance of monitoring
these two variables using apps that measure speed and incline or smartwatches, which are
increasingly employed in outdoor races and training sessions.
4.4. Metabolic Power Calculation
The outcomes of the multiple regression analysis revealed that 94% of the variance
in mechanical vertical COM power during the VK test could be accounted for by net
metabolic power and the vertical metabolic COT. This substantial explanation is primarily
attributed to the fact that these two variables elucidate the vertical velocity, a key component
of mechanical vertical COM power. From the derived equation (Equation (6)), three
coefficients sensitive to the progression of the test and inclination were obtained (Table 6).
These coefficients are likely subject to variations depending on the characteristics of the
uphill test, such as slope, section lengths and their interaction. This observation is consistent
with our study’s results, where net metabolic power levels exhibited changes due to slope
and fatigue. The findings of this regression analysis suggest that, once an ascent has
been characterized, net metabolic power can be estimated based on the runner’s vertical
velocity. Consequently, a reliable equation (Equation (8)) was established from the multiple
regression to calculate the runner’s metabolic power during a VK field test. This equation
utilizes only the vertical velocity and the coefficients found in the model, eliminating the
need for expensive portable gas analyzers. The ease of analysis with common devices
like phones, smartwatches and GPS is a notable advantage [63]. These results align with
Sensors 2023, 23, 9349
15 of 19
the increasing interest among researchers to determine metabolic power during actual
competitions in various sports. This pursuit aims to enhance the understanding of the real
workload for athletes, thereby improving training methods and periodization [64–67].
Table 6. Multiple linear regression model for mechanical vertical COM power.
R
R2
adR2
SEE
p
Durbin–
Watson
B
SE
Beta
p
B
VIF
LL95%
UL95%
0.975
0.951
0.942
0.07
<0.001
1.911
α
0.133
0.009
1.243
<0.001
0.113
0.152
1.626
β
−0.030
0.003
−0.797
<0.001
−0.037
−0.023
1.626
γ
2.376
0.183
<0.001
1.973
2.779
R: correlation coefficient, R2: determination coefficient; adR2: adjusted determination coefficient; SEE: standard
error of the estimation; p: significance level; LL95%: lower limit for 95% confidence interval; UL95%: upper
limit for 95% confidence interval, B: multiple linear regression coefficients of each variable; SE: B-standard error;
Beta: standardized coefficients; VIF: variance inflation factor; α: net metabolic power coefficient; β: vertical net
metabolic COT coefficient; γ: independent coefficient of the multiple regression.
Our formula, combined with the VO2 submax at 30◦ formula developed by Giovanelli
et al. [38], can serve as a valuable tool for characterizing VK runners based on easily
measured variables in a real field test.
The study results offer novel insights into the significance of utilizing mechanical
power, metabolic power and vertical velocity variables for performance analysis in vertical
kilometer runners, regardless of gender. Furthermore, it underscores their susceptibility to
impairment due to the influence of fatigue. These findings align with the increasing interest
in acquiring high-quality information on athletes’ internal load through the progressive
improvement of technology and data analysis methods [3]. Moreover, it opens up the
possibility of conducting further research to deeper analyze these variables across various
running modalities and both cyclic and acyclic sports.
A major strength of the present study lies in the simplicity with which these variables
can be implemented in any existing wearable sensor on the market that utilizes IMUs
and GNSS to calculate real-time velocity, accelerations, anthropometric data and terrain
characteristics. These data facilitate the calculation of key parameters, eliminating the need
for athletes and coaches to undergo time-consuming and fatiguing tests and allowing data
collection during training and competition [2].
Finally, possessing a comprehensive understanding of key variables within each
sporting context is crucial. This clarity is essential for precise data collection, enabling
researchers and companies to save a significant amount of time developing software and
sensors [2].
As future lines of research, it would be interesting to validate the metabolic power
formula and continue studying runners through real field tests.
5. Limitations
The main limitation of the study is the small sample size, with only 11 male and
4 female participants. This limitation arose from the technical complexity and time cost
associated with conducting the analyses in a true vertical kilometer field test. Future
extensive analyses with a larger and more diverse sample should be conducted, particularly
for the reliability and validity assessment of the metabolic power formula identified in
this study.
Additionally, the absence of anthropometric analysis to determine the body fat per-
centage and the level of lower limb muscle mass among the runners represents another
limitation. This lack of information prevents readers from gaining insights into the sub-
jects’ fitness levels, which would provide better context for the study’s findings. Notably,
individuals with lower body fat percentages and higher levels of leg muscle mass are often
observed to perform better in trail running tests. Consequently, future studies should
Sensors 2023, 23, 9349
16 of 19
incorporate analyses of these parameters. Lastly, another limitation is the absence of a
pre-vertical kilometer maximum treadmill test to assess the physiological condition of the
runners, as well as a strength test to gauge their neuromuscular level. These factors are also
crucial for race performance and should undergo thorough examination in future studies.
6. Conclusions
The study results revealed significant differences in the mean values of variables such
as velocity, vertical velocity, mechanical vertical COM power and net metabolic power
when comparing top-level runners to low-level runners during a vertical kilometer field
test. Additionally, all analyzed variables were affected by fatigue as the test progressed,
showing significant differences in how fatigue altered mechanical vertical COM power and
vertical velocity when comparing top-level to low-level runners. A multiple regression
analysis demonstrated that 94% of the mechanical vertical COM power during the vertical
kilometer test could be explained by net metabolic power and vertical net metabolic cost
of transport. Subsequently, a reliable equation was derived from the multiple regression
to calculate each runner’s metabolic power during a vertical kilometer field test, utilizing
only the vertical velocity and the coefficients identified in the model. These findings
present an opportunity to explore new variables correlated with performance in short
trail running, particularly in vertical kilometer races. These new variables are sensitive to
performance disparities, exhibit changes with fatigue and are applicable to both male and
female athletes. Importantly, they can be easily measured through apps, smartwatches,
foot-pod potentiometers and GPS.
Author Contributions: Conceptualization, P.J.B., A.V.B.-C. and C.B.; methodology, H.G., J.R.-P. and
C.B.; formal analysis, P.J.B. and H.G.; investigation, A.V.B.-C., E.P. and J.R.-P.; data curation, P.J.B. and
H.G.; writing—original draft preparation, P.J.B. and H.G.; writing—review and editing, A.V.B.-C.,
C.B., E.P. and J.R.-P.; funding acquisition, C.B. All authors have read and agreed to the published
version of the manuscript.
Funding: This work was partially funded by Departamento de Ciencia, Universidad y Sociedad del
Conocimiento, from the Gobierno de Aragón (Spain) (Research Group ValorA, under grant S08_23R.
In addition, this research was partially supported by the Spanish Ministry of Universities (FPU grant
FPU19/00967).
Institutional Review Board Statement: This study was conducted in accordance with the Declaration
of Helsinki and approved by the Institutional Ethics Committee of the Universidad San Jorge, protocol
code Ref 005-19/20.
Informed Consent Statement: Informed consent was obtained from all subjects involved in the
study. Written informed consent was obtained from the subjects to publish this paper.
Data Availability Statement: The raw data belong to the Universidad San Jorge and can be requested
from the corresponding author with the permission of Universidad San Jorge.
Conflicts of Interest: The authors declare no conflict of interest. The funders had no role in the design
of the study; in the collection, analyses or interpretation of data; in the writing of the manuscript; or
in the decision to publish the results.
References
1.
Bunn, J.A.; Navalta, J.W.; Fountaine, C.J.; Reece, J.D. Current State of Commercial Wearable Technology in Physical Activity
Monitoring 2015–2017. Int. J. Exerc. Sci. 2018, 11, 503–515. [PubMed]
2.
Helwig, J.; Diels, J.; Röll, M.; Mahler, H.; Gollhofer, A.; Roecker, K.; Willwacher, S. Relationships between External, Wearable
Sensor-Based, and Internal Parameters: A Systematic Review. Sensors 2023, 23, 827. [CrossRef] [PubMed]
3.
Passos, J.; Lopes, S.I.; Clemente, F.M.; Moreira, P.M.; Rico-González, M.; Bezerra, P.; Rodrigues, L.P. Wearables and Internet of
Things (IoT) Technologies for Fitness Assessment: A Systematic Review. Sensors 2021, 21, 5418. [CrossRef] [PubMed]
4.
Home—The International Skyrunning Federation.
Available online:
https://www.skyrunning.com/ (accessed on
20 August 2023).
5.
Conley, D.L.; Krahenbuhl, G.S. Running Economy and Distance Running Performance of Highly Trained Athletes. Med. Sci.
Sports Exerc. 1980, 12, 357–360. [CrossRef]
Sensors 2023, 23, 9349
17 of 19
6.
Kipp, S.; Byrnes, W.C.; Kram, R. Calculating Metabolic Energy Expenditure across a Wide Range of Exercise Intensities: The
Equation Matters. Appl. Physiol. Nutr. Metab. 2018, 43, 639–642. [CrossRef]
7.
Saunders, P.U.; Pyne, D.B.; Telford, R.D.; Hawley, J.A. Factors Affecting Running Economy in Trained Distance Runners. Sports
Med. 2004, 34, 465–485. [CrossRef]
8.
Barnes, K.R.; Kilding, A.E. Running Economy: Measurement, Norms, and Determining Factors. Sport. Med. Open 2015, 1, 8.
[CrossRef] [PubMed]
9.
Fletcher, J.; Esau, S.; Macintosh, B. Economy of Running: Beyond the Measurement of Oxygen Uptake. J. Appl. Physiol. 2009, 107,
1918–1922. [CrossRef]
10.
Cerezuela-Espejo, V.; Hernández-Belmonte, A.; Courel-Ibáñez, J.; Conesa-Ros, E.; Mora-Rodríguez, R.; Pallarés, J.G. Are We
Ready to Measure Running Power? Repeatability and Concurrent Validity of Five Commercial Technologies. Eur. J. Sport Sci.
2021, 21, 341–350. [CrossRef]
11.
Imbach, F.; Candau, R.; Chailan, R.; Perrey, S. Validity of the Stryd Power Meter in Measuring Running Parameters at Submaximal
Speeds. Sports 2020, 8, 103. [CrossRef]
12.
Giovanelli, N.; Taboga, P.; Rejc, E.; Lazzer, S. Effects of Strength, Explosive and Plyometric Training on Energy Cost of Running in
Ultra-Endurance Athletes. Eur. J. Sport Sci. 2017, 17, 805–813. [CrossRef] [PubMed]
13.
Lazzer, S.; Salvadego, D.; Taboga, P.; Rejc, E.; Giovanelli, N.; di Prampero, P.E. Effects of the Etna Uphill Ultramarathon on Energy
Cost and Mechanics of Running. Int. J. Sports Physiol. Perform. 2015, 10, 238–247. [CrossRef] [PubMed]
14.
Roberts, T.J.; Belliveau, R.A. Sources of Mechanical Power for Uphill Running in Humans. J. Exp. Biol. 2005, 208, 1963–1970.
[CrossRef]
15.
Margaria, R.; Cerretelli, P.; Aghemo, P.; Sassi, G. Energy Cost of Running. J. Appl. Physiol. 1963, 18, 367–370. [CrossRef]
16.
Minetti, A.E.; Moia, C.; Roi, G.S.; Susta, D.; Ferretti, G. Energy Cost of Walking and Running at Extreme Uphill and Downhill
Slopes. J. Appl. Physiol. 2002, 93, 1039–1046. [CrossRef]
17.
Balducci, P.; Clémençon, M.; Morel, B.; Quiniou, G.; Saboul, D.; Hautier, C.A. Comparison of Level and Graded Treadmill Tests to
Evaluate Endurance Mountain Runners. J. Sports Sci. Med. 2016, 15, 239–246. [PubMed]
18.
Balducci, P.; Clémençon, M.; Trama, R.; Blache, Y.; Hautier, C. Performance Factors in a Mountain Ultramarathon. Int. J. Sports
Med. 2017, 38, 819–826. [CrossRef]
19.
Vernillo, G.; Savoldelli, A.; Zignoli, A.; Skafidas, S.; Fornasiero, A.; La Torre, A.; Bortolan, L.; Pellegrini, B.; Schena, F. Energy Cost
and Kinematics of Level, Uphill and Downhill Running: Fatigue-Induced Changes after a Mountain Ultramarathon. J. Sports Sci.
2015, 33, 1998–2005. [CrossRef]
20.
Saibene, F.; Minetti, A.E. Biomechanical and Physiological Aspects of Legged Locomotion in Humans. Eur. J. Appl. Physiol. 2003,
88, 297–316. [CrossRef]
21.
Snyder, K.L.; Kram, R.; Gottschall, J.S. The Role of Elastic Energy Storage and Recovery in Downhill and Uphill Running. J. Exp.
Biol. 2012, 215, 2283–2287. [CrossRef]
22.
Dewolf, A.H.; Peñailillo, L.E.; Willems, P.A. The Rebound of the Body during Uphill and Downhill Running at Different Speeds. J.
Exp. Biol. 2016, 219, 2276–2288. [CrossRef] [PubMed]
23.
Lemire, M.; Falbriard, M.; Aminian, K.; Millet, G.P.; Meyer, F. Level, Uphill, and Downhill Running Economy Values Are
Correlated Except on Steep Slopes. Front. Physiol. 2021, 12, 697315. [CrossRef] [PubMed]
24.
Peyré-Tartaruga, L.A.; Coertjens, M. Locomotion as a Powerful Model to Study Integrative Physiology: Efficiency, Economy, and
Power Relationship. Front. Physiol. 2018, 9, 1789. [CrossRef] [PubMed]
25.
Gaesser, G.A.; Brooks, G.A. Muscular Efficiency during Steady-Rate Exercise: Effects of Speed and Work Rate. J. Appl. Physiol.
1975, 38, 1132–1139. [CrossRef]
26.
Field, A.P. Discovering Statistics Using IBM SPSS Statistics, 4th ed.; SAGE Publications: Thousand Oaks, CA, USA, 2013;
ISBN 9781446249178.
27.
De Lucas, R.D.; Karam De Mattos, B.; Tremel, A.D.C.; Pianezzer, L.; De Souza, K.M.; Guglielmo, L.G.A.; Denadai, B.S. A Novel
Treadmill Protocol for Uphill Running Assessment: The Incline Incremental Running Test (IIRT). Res. Sports Med. 2022, 30,
554–565. [CrossRef]
28.
Doucende, G.; Chamoux, M.; Defer, T.; Rissetto, C.; Mourot, L.; Cassirame, J. Specific Incremental Test for Aerobic Fitness in Trail
Running: IncremenTrail. Sports 2022, 10, 174. [CrossRef]
29.
Cassirame, J.; Godin, A.; Chamoux, M.; Doucende, G.; Mourot, L. Physiological Implication of Slope Gradient during Incremental
Running Test. Int. J. Environ. Res. Public Health 2022, 19, 12210. [CrossRef]
30.
Schöffl, I.; Jasinski, D.; Ehrlich, B.; Dittrich, S.; Schöffl, V. Outdoor Uphill Exercise Testing for Trail Runners, a More Suitable
Method? J. Hum. Kinet. 2021, 79, 123–133. [CrossRef]
31.
Gottschall, J.S.; Kram, R. Ground Reaction Forces during Downhill and Uphill Running. J. Biomech. 2005, 38, 445–452. [CrossRef]
32.
Khassetarash, A.; Vernillo, G.; Martinez, A.; Baggaley, M.; Giandolini, M.; Horvais, N.; Millet, G.Y.; Edwards, W.B. Biomechanics
of Graded Running: Part II-Joint Kinematics and Kinetics. Scand. J. Med. Sci. Sports 2020, 30, 1642–1654. [CrossRef]
33.
Ehrström, S.; Tartaruga, M.P.; Easthope, C.S.; Brisswalter, J.; Morin, J.-B.; Vercruyssen, F. Short Trail Running Race: Beyond the
Classic Model for Endurance Running Performance. Med. Sci. Sports Exerc. 2018, 50, 580–588. [CrossRef]
34.
Ortiz, A.L.R.; Giovanelli, N.; Kram, R. The Metabolic Costs of Walking and Running up a 30-Degree Incline: Implications for
Vertical Kilometer Foot Races. Eur. J. Appl. Physiol. 2017, 117, 1869–1876. [CrossRef] [PubMed]
Sensors 2023, 23, 9349
18 of 19
35.
Zimmermann, P.; Müller, N.; Schöffl, V.; Ehrlich, B.; Moser, O.; Schöffl, I. The Energetic Costs of Uphill Locomotion in Trail
Running: Physiological Consequences Due to Uphill Locomotion Pattern—A Feasibility Study. Life 2022, 12, 2070. [CrossRef]
[PubMed]
36.
Drobniˇc, M.; Verdel, N.; Holmberg, H.-C.; Supej, M. The Validity of a Three-Dimensional Motion Capture System and the Garmin
Running Dynamics Pod in Connection with an Assessment of Ground Contact Time While Running in Place. Sensors 2023, 23,
7155. [CrossRef] [PubMed]
37.
Lichtwark, G.A.; Wilson, A.M. Interactions between the Human Gastrocnemius Muscle and the Achilles Tendon during Incline,
Level and Decline Locomotion. J. Exp. Biol. 2006, 209, 4379–4388. [CrossRef] [PubMed]
38.
Giovanelli, N.; Ortiz, A.L.R.; Henninger, K.; Kram, R. Energetics of Vertical Kilometer Foot Races; Is Steeper Cheaper? J. Appl.
Physiol. 2016, 120, 370–375. [CrossRef]
39.
Giandolini, M.; Vernillo, G.; Samozino, P.; Horvais, N.; Edwards, W.B.; Morin, J.-B.; Millet, G.Y. Fatigue Associated with Prolonged
Graded Running. Eur. J. Appl. Physiol. 2016, 116, 1859–1873. [CrossRef]
40.
Fourchet, F.; Millet, G.P.; Tomazin, K.; Guex, K.; Nosaka, K.; Edouard, P.; Degache, F.; Millet, G.Y. Effects of a 5-h Hilly Running
on Ankle Plantar and Dorsal Flexor Force and Fatigability. Eur. J. Appl. Physiol. 2012, 112, 2645–2652. [CrossRef]
41.
Millet, G.Y.; Tomazin, K.; Verges, S.; Vincent, C.; Bonnefoy, R.; Boisson, R.-C.; Gergelé, L.; Féasson, L.; Martin, V. Neuromuscular
Consequences of an Extreme Mountain Ultra-Marathon. PLoS ONE 2011, 6, e17059. [CrossRef]
42.
Temesi, J.; Arnal, P.J.; Rupp, T.; Féasson, L.; Cartier, R.; Gergelé, L.; Verges, S.; Martin, V.; Millet, G.Y. Are Females More Resistant
to Extreme Neuromuscular Fatigue? Med. Sci. Sports Exerc. 2015, 47, 1372–1382. [CrossRef]
43.
Muñoz-Pérez, I.; Varela-Sanz, A.; Lago-Fuentes, C.; Navarro-Patón, R.; Mecías-Calvo, M. Central and Peripheral Fatigue in
Recreational Trail Runners: A Pilot Study. Int. J. Environ. Res. Public Health 2022, 20, 402. [CrossRef] [PubMed]
44.
Millet, G.Y.; Martin, V.; Lattier, G.; Ballay, Y. Mechanisms Contributing to Knee Extensor Strength Loss after Prolonged Running
Exercise. J. Appl. Physiol. 2003, 94, 193–198. [CrossRef]
45.
Noakes, T.D. Fatigue Is a Brain-Derived Emotion That Regulates the Exercise Behavior to Ensure the Protection of Whole Body
Homeostasis. Front. Physiol. 2012, 3, 82. [CrossRef]
46.
Pokora, I.; Kempa, K.; Chrapusta, S.J.; Langfort, J. Effects of Downhill and Uphill Exercises of Equivalent Submaximal Intensities
on Selected Blood Cytokine Levels and Blood Creatine Kinase Activity. Biol. Sport 2014, 31, 173–178. [CrossRef] [PubMed]
47.
Mercier, J.; Le Gallais, D.; Durand, M.; Goudal, C.; Micallef, J.P.; Préfaut, C. Energy Expenditure and Cardiorespiratory Responses
at the Transition between Walking and Running. Eur. J. Appl. Physiol. Occup. Physiol. 1994, 69, 525–529. [CrossRef] [PubMed]
48.
Vercruyssen, F.; Tartaruga, M.; Horvais, N.; Brisswalter, J. Effects of Footwear and Fatigue on Running Economy and Biomechanics
in Trail Runners. Med. Sci. Sports Exerc. 2016, 48, 1976–1984. [CrossRef]
49.
Sabater Pastor, F.; Varesco, G.; Besson, T.; Koral, J.; Feasson, L.; Millet, G.Y. Degradation of Energy Cost with Fatigue Induced by
Trail Running: Effect of Distance. Eur. J. Appl. Physiol. 2021, 121, 1665–1675. [CrossRef]
50.
Hunter, I.; Smith, G.A. Preferred and Optimal Stride Frequency, Stiffness and Economy: Changes with Fatigue during a 1-h
High-Intensity Run. Eur. J. Appl. Physiol. 2007, 100, 653–661. [CrossRef]
51.
Vernillo, G.; Giandolini, M.; Edwards, W.B.; Morin, J.-B.; Samozino, P.; Horvais, N.; Millet, G.Y. Biomechanics and Physiology of
Uphill and Downhill Running. Sports Med. 2017, 47, 615–629. [CrossRef]
52.
Degache, F.; Guex, K.; Fourchet, F.; Morin, J.B.; Millet, G.P.; Tomazin, K.; Millet, G.Y. Changes in Running Mechanics and
Spring-Mass Behaviour Induced by a 5-Hour Hilly Running Bout. J. Sports Sci. 2013, 31, 299–304. [CrossRef]
53.
Morin, J.-B.; Samozino, P.; Millet, G.Y. Changes in Running Kinematics, Kinetics, and Spring-Mass Behavior over a 24-h Run. Med.
Sci. Sports Exerc. 2011, 43, 829–836. [CrossRef]
54.
McMahon, T.A.; Valiant, G.; Frederick, E.C. Groucho Running. J. Appl. Physiol. 1987, 62, 2326–2337. [CrossRef] [PubMed]
55.
Shorten, M.R. Mechanical Energy Transformations and Energy Expenditure in Running Man. Ph.D. Thesis, Loughborough
University, Loughborough, UK, 1984.
56.
Vernillo, G.; Millet, G.P.; Millet, G.Y. Does the Running Economy Really Increase after Ultra-Marathons? Front. Physiol. 2017, 8,
783. [CrossRef] [PubMed]
57.
Millet, G.Y.; Hoffman, M.D.; Morin, J.B. Sacrificing Economy to Improve Running Performance—A Reality in the Ultramarathon?
J. Appl. Physiol. 2012, 113, 507–509. [CrossRef] [PubMed]
58.
Ettema, G.; Lorås, H.W. Efficiency in Cycling: A Review. Eur. J. Appl. Physiol. 2009, 106, 1–14. [CrossRef] [PubMed]
59.
Pastor, F.S.; Besson, T.; Varesco, G.; Parent, A.; Fanget, M.; Koral, J.; Foschia, C.; Rupp, T.; Rimaud, D.; Féasson, L.; et al.
Performance Determinants in Trail-Running Races of Different Distances. Int. J. Sports Physiol. Perform. 2022, 17, 844–851.
[CrossRef] [PubMed]
60.
Blagrove, R.C.; Howatson, G.; Hayes, P.R. Effects of Strength Training on the Physiological Determinants of Middle- and
Long-Distance Running Performance: A Systematic Review. Sports Med. 2018, 48, 1117–1149. [CrossRef] [PubMed]
61.
Ferley, D.D.; Osborn, R.W.; Vukovich, M.D. The Effects of Incline and Level-Grade High-Intensity Interval Treadmill Training on
Running Economy and Muscle Power in Well-Trained Distance Runners. J. Strength Cond. Res. 2014, 28, 1298–1309. [CrossRef]
62.
Gimenez, P.; Arnal, P.J.; Samozino, P.; Millet, G.Y.; Morin, J.-B. Simulation of Uphill/Downhill Running on a Level Treadmill
Using Additional Horizontal Force. J. Biomech. 2014, 47, 2517–2521. [CrossRef]
63.
Davidson, P.; Virekunnas, H.; Sharma, D.; Piché, R.; Cronin, N. Continuous Analysis of Running Mechanics by Means of an
Integrated INS/GPS Device. Sensors 2019, 19, 1480. [CrossRef]
Sensors 2023, 23, 9349
19 of 19
64.
Osgnach, C.; Poser, S.; Bernardini, R.; Rinaldo, R.; di Prampero, P.E. Energy Cost and Metabolic Power in Elite Soccer: A New
Match Analysis Approach. Med. Sci. Sports Exerc. 2010, 42, 170–178. [CrossRef] [PubMed]
65.
Buchheit, M.; Manouvrier, C.; Cassirame, J.; Morin, J.-B. Monitoring Locomotor Load in Soccer: Is Metabolic Power, Powerful?
Int. J. Sports Med. 2015, 36, 1149–1155. [CrossRef] [PubMed]
66.
Di Prampero, P.E.; Botter, A.; Osgnach, C. The Energy Cost of Sprint Running and the Role of Metabolic Power in Setting Top
Performances. Eur. J. Appl. Physiol. 2015, 115, 451–469. [CrossRef] [PubMed]
67.
Coutts, A.J.; Kempton, T.; Sullivan, C.; Bilsborough, J.; Cordy, J.; Rampinini, E. Metabolic Power and Energetic Costs of
Professional Australian Football Match-Play. J. Sci. Med. Sport 2015, 18, 219–224. [CrossRef] [PubMed]
Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual
author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to
people or property resulting from any ideas, methods, instructions or products referred to in the content.
| Running Economy in the Vertical Kilometer. | 11-23-2023 | Bascuas, Pablo Jesus,Gutiérrez, Héctor,Piedrafita, Eduardo,Rabal-Pelay, Juan,Berzosa, César,Bataller-Cervero, Ana Vanessa | eng |
PMC10335662 | RESEARCH ARTICLE
The relationship between ambient
temperature and match running performance
of elite soccer players
Ryland MorgansID1,2*, Eduard Bezuglov2, Dave RhodesID1, Jose TeixeiraID3,4,5,
Toni Modric6, Sime Versic6, Rocco Di Michele7, Rafael OliveiraID3,8,9
1 Football Performance Hub, University of Central Lancashire, Preston, United Kingdom, 2 Department of
Sports Medicine and Medical Rehabilitation, Sechenov State Medical University, Moscow, Russia,
3 Research Centre in Sports Sciences, Health and Human Development, Vila Real, Portugal,
4 Departamento de Desporto e Educac¸ão Fı´sica, Instituto Polite´cnico de Braganc¸a, Braganc¸a, Portugal,
5 Instituto Polite´cnico da Guarda, Guarda, Portugal, 6 Faculty of Kinesiology, University of Split, Split,
Croatia, 7 Department of Biomedical and Neuromotor Sciences, University of Bologna, Bologna, Italy,
8 Sports Science School of Rio Maior–Polytechnic Institute of Santare´m, Rio Maior, Portugal, 9 Life Quality
Research Centre, Rio Maior, Portugal
* [email protected]
Abstract
The influence of environmental factors on key physical parameters of soccer players during
competitive match-play have been widely investigated in the literature, although little is
known on the effects of sub-zero ambient temperatures on the performance of adult elite
soccer players during competitive matches. The aim of this study was to assess how the
teams’ match running performance indicators are related to low ambient temperature during
competitive matches in the Russian Premier League. A total of 1142 matches played during
the 2016/2017 to 2020/2021 seasons were examined. Linear mixed models were used to
assess the relationships between changes in ambient temperature at the start of the match
and changes in selected team physical performance variables, including total, running (4.0
to 5.5 m/s), high-speed running (5.5 to 7.0 m/s) and sprint (> 7.0 m/s) distances covered.
The total, running and high-speed running distances showed no significant differences
across temperatures up to 10˚C, while these showed small to large decreases at 11 to 20˚C
and especially in the >20˚C ranges. On the contrary, sprint distance was significantly lower
at temperature of -5˚C or less compared to higher temperature ranges. At sub-zero temper-
atures, every 1˚C lower reduced team sprint distance by 19.2 m (about 1.6%). The present
findings show that a low ambient temperature is negatively related to physical match perfor-
mance behavior of elite soccer players, notably associated with a reduced total sprint
distance.
Introduction
Soccer is a worldwide sport that is played in differing environmental conditions varying from
extreme heat to severely cold temperatures. The physical demands that are required on players
PLOS ONE
PLOS ONE | https://doi.org/10.1371/journal.pone.0288494
July 11, 2023
1 / 9
a1111111111
a1111111111
a1111111111
a1111111111
a1111111111
OPEN ACCESS
Citation: Morgans R, Bezuglov E, Rhodes D,
Teixeira J, Modric T, Versic S, et al. (2023) The
relationship between ambient temperature and
match running performance of elite soccer players.
PLoS ONE 18(7): e0288494. https://doi.org/
10.1371/journal.pone.0288494
Editor: Emiliano Cè, Università degli Studi di
Milano: Universita degli Studi di Milano, ITALY
Received: February 6, 2023
Accepted: June 28, 2023
Published: July 11, 2023
Copyright: © 2023 Morgans et al. This is an open
access article distributed under the terms of the
Creative Commons Attribution License, which
permits unrestricted use, distribution, and
reproduction in any medium, provided the original
author and source are credited.
Data Availability Statement: All relevant data are
within the paper.
Funding: This research was funded by the
Portuguese Foundation for Science and
Technology, I.P., Grant/Award Number UIDP/
04748/2020. The funders had no role in study
design, data collection and analysis, decision to
publish, or preparation of the manuscript.
Competing interests: The authors have declared
that no competing interests exist.
can vary significantly as a result of the environmental conditions in which soccer matches are
played [1–3]. Over the course of a soccer season, environment conditions can change signifi-
cantly from warm summer months to cold winter months within most European soccer lea-
gues. These changes may dictate important considerations for the physical preparation of
players over the course of a soccer season in relation to the specific environmental conditions
experienced locally.
Much of the previous research that has been undertaken into the effects of environmental
conditions on soccer performance has focused on the impact of hot environments and expo-
sure to altitude [4]. While this may be of significant importance in certain areas of the world, it
would appear that it is also important to investigate the impact of extremely cold temperatures
which are more commonly seen during the winter months of many European leagues. There
is scant previous research in this area, however, some studies have assessed the effects of cold
environments on physical performance during soccer match-play [5]. For instance, this study
showed greater distances covered by midfielders at high-intensity running (>19.8 km/h),
between the 30-45-minute period, in temperatures 5˚C versus 21˚C. However, total dis-
tance covered was unaffected in colder conditions, although this finding could not be con-
firmed across all positions as defenders and strikers were not assessed [5].
Assessment of the effects of different temperatures on the physiological response to sub-
maximal exercise in soccer players shows that exercise capacity is reduced in cool environ-
ments when compared with moderate conditions [6]. This was evident by changes in players’
heart rate responses and ventilation rates in cool environments when cycling for 20 minutes at
60% maximum oxygen uptake and until reaching exhaustion in three different environmental
conditions (10, 22, 35˚C) [6]. Furthermore, Armstrong [7] stated that exercising in cold envi-
ronments has various effects on physiological processes such as an increase in muscle glycogen
utilization, a reduction in aerobic capacity, and a reduction in muscular strength and power.
More specifically, Carling et al. [5] investigated the effect of cold temperature on physical
activity profiles in matches over four seasons in the French Ligue 1. The authors reported that
generally physical outputs were unaffected by changes in environmental temperatures in the
matches analyzed. However, the lowest match temperature involved in this study were 5˚C.
This temperature accounts for a large range of temperatures which may significantly affect
physical match performance. Although, further research may be required to assess the differ-
ences in the effect on physical match performance of very cold and freezing temperatures
(<5˚C).
Therefore, the aim of this study was to investigate the relationships between ambient tem-
peratures and physical match activity of elite soccer players during official competitive matches
in the Russian Premier League, where matches are often played in a cold environment, espe-
cially during autumn and winter months. Given the different findings of the previous studies
about the lower capacity of exercising in a cold environment on physical performance [6, 7],
and at the same time that midfielders run higher distances at a speed > 19.8 km/h [5], it was
hypothesized that playing at very low temperatures would impact physical match behavior
when compared with warmer environmental conditions, specially at speeds < 19.8 km/h.
Materials and methods
Experimental design
An observational design to assess the impact of ambient temperature on soccer match physical
performance in official matches played in the Russian Premier League over a period of five
competitive seasons was employed.
PLOS ONE
Cold exposure and match running performance
PLOS ONE | https://doi.org/10.1371/journal.pone.0288494
July 11, 2023
2 / 9
Match sample
The ambient temperature at the start of the match, and a selection of the team’s physical match
performance variables, were assessed for all official matches played in the Russian Premier
League during five consecutive seasons, 2016/2017 to 2020/2021. Due to some missing temper-
ature and physical performance data, and the poor quality of some match data due to environ-
mental or other reasons, a total of 1146 out of 1200 matches were examined. For every match,
data for the home and away team were examined, resulting in a total of 2292 team perfor-
mance data points, from a total of 24 different teams. Ethical approval for the study was
granted by the Ethics Committee of Sechenov University (N06-21 dated 04/07/2021). The
study was performed in accordance with the Helsinki Declaration principles. All professional
players and the Russian Premier League agreed to the collection of match performance data
and for research purposes.
Data collection
League match data across each season was recorded and analyzed via a two-camera Optical
Tracking System (InStat, Moscow, Russia) to report physical performance data. The matches
were filmed using two full HD, static cameras positioned on the centre line of the field, not less
than 3-metres from the field and 7-metres in height. A consistent 25Hz format was provided.
Data were linearly interpolated to 50Hz, smoothed using a 5-point moving average and then
down-sampled to 10Hz. The reliability and validity of InStat have been demonstrated by
assessing velocity and position data collected during soccer-specific exercises in comparison
with a reference stereophotogrammetric system (Vicon Motion Systems Ltd., Oxford, UK).
These assessments are included in the official FIFA test protocol for Electronic and Perfor-
mance Tracking Systems document, stating that the system has passed [8]. The InStat Analysis
Software System was used to measure and analyze physical performance. InStat provided writ-
ten permission to allow all match data to be used for research purposes.
The following distances, describing the whole team physical match performance, were
assessed: total distance covered by all team players (m); running distance covered by all team
players (m; total distance covered 4.0 to 5.5 m/s); high-speed running (HSR) distance covered
by all team players (m; total distance covered 5.5 to 7.0 m/s); and sprint distance covered by all
team players (m; total distance covered >7.0 m/s). These variables were selected based on pre-
vious studies [9, 10] to facilitate comparisons.
The League Development Department gathered temperature data on all official matches
during the 2015 to 2021 period from the official Hydrometeorological Center. Meteorological
data were retrieved from the meteorological stations located nearest the stadiums, available
from the Russian Weather Service (http://aisori-m.meteo.ru/waisori/) and Ogimet (https://
www.ogimet.com/synops.phtml.en) databases. As far as data from the stadiums were con-
cerned, meteorological data were thus taken per hour during the match-day and one hour
prior to kick-off, measured from the centre of the field, as stated by League requirements. Data
on air temperature during matches were retrieved from the stadiums’ meteorological stations
and reported by the official match delegate. Two experts retrospectively examined and verified
the official Hydrometeorological Center data by comparisons with the League delegate reports.
All matches were classified into six groups depending on the ambient temperature at the start
of the match: <-5˚C; -4 to 0˚C; 1 to 5˚C; 6 to 10˚C; 11 to 20˚C; >20˚C. Ranges for tempera-
tures higher than 5˚C were set according to Carling et al. [5], while two additional ranges (-4
to 0˚C and 1 to 5˚C) were added for temperatures up to 5˚C. The lowest temperature threshold
(<-5˚C) was set according to Link and Weber [11]. All physical match performance and
PLOS ONE
Cold exposure and match running performance
PLOS ONE | https://doi.org/10.1371/journal.pone.0288494
July 11, 2023
3 / 9
match temperature data were analysed by the same experienced investigator to ensure inter-
observer reliability was not a limitation of the study.
Statistical analysis
The statistical analysis was conducted using the software R, version 4.2.0 (R Foundation for
Statistical Computing, Vienna, Austria). Data are presented as mean ± standard deviation. Lin-
ear mixed models with random intercept on team IDs were used to compare the examined
physical performance variables across temperature ranges. When there was a significant
(p<0.05) difference between temperature ranges, Tukey’s tests were used to determine which
ranges differed. The estimated differences were standardized by dividing them by the esti-
mated between-team standard deviation to determine the effect size (ES). Absolute ES values
were evaluated as <0.2, trivial; 0.2–0.6, small; 0.6–1.2, moderate; 1.2–2.0, large; 2.0–4.0, very
large; >4.0 extremely large [12].
The effect of low temperature on physical match performance was further explored by tak-
ing temperature as a continuous variable. Linear mixed models with random intercept on
team IDs were used to examine the effect of 1˚C ambient temperature decrease on physical
performance variables. For this analysis, only matches started at ambient temperature < 0˚C
were included. Effect sizes were calculated as Cohen’s d from the coefficients of linear mixed
models coefficient with the lme.dscore function from package EMAtools [13], and interpreted
as very small (<0.2), small (0.2–0.5), medium (0.5–0.8), large (>0.8) [14].
For all models, it was ensured that the assumptions of linearity, homoscedasticity and nor-
mality of residuals were met by visually inspecting histograms of residuals and plots of residu-
als vs. fitted values. For all analyses, statistical significance was set at p<0.05.
Results
In the examined matches, the mean ambient temperature at the start of the match was
11.7 ± 10.2˚C.
Table 1 presents the mean values for the examined physical performance variables in each
temperature range, while Table 2 displays the standardized differences (ES). The total distance
was similar from the <-5˚C to the 6 to 10˚C ranges, while it decreased in the 11 to 20˚C range
Table 1. Mean ± SD values, with 95% Confidence Intervals (CI), for physical match performance variables across temperature ranges.
<-5˚C
-4 to 0˚C
1 to 5˚C
6 to 10˚C
11 to 20˚C
>20˚C
(n = 134)
(n = 222)
(n = 322)
(n = 372)
(n = 726)
(n = 516)
Total distance (km)
Mean ±SD
114.9de ± 4.8
114.3de ± 5.1
114.5de ± 4.7
114.6de ± 4.6
113.7e ± 4.8
112.3 ± 5.5
95% CI
114.1–115.8
113.6–114.9
114.0–115.0
114.1–115.1
113.4–114.0
112.8–111.9
Running distance (km)
Mean ±SD
20.0de ± 1.8
20.0de ± 2.0
20.1de ± 1.8
20.0de ± 2.0
19.3e ± 1.8
18.3 ± 1.8
95% CI
19.7–20.3
19.7–20.3
19.9–20.3
19.8–20.2
19.2–19.5
18.2–18.5
High-speed distance (km)
Mean ±SD
7.7e ± 0.9
7.9de ± 0.9
7.9de ± 0.9
7.9de ± 1.0
7.7e ± 0.9
7.2 ± 0.9
95% CI
7.5–7.8
7.8–8.0
7.8–8.0
7.8–8.0
7.6–7.7
7.1–7.3
Sprint distance (km)
Mean ±SD
1.1abcd ± 0.4
1.2e ± 0.4
1.2e ± 0.3
1.3e ± 0.4
1.3e ± 0.4
1.1 ± 0.4
95% CI
1.0–1.1
1.2–1.3
1.2–1.3
1.2–1.3
1.2–1.3
1.1–1.2
a denotes a significant difference vs. -4 to 0˚C
b denotes a significant difference vs. 1 to 5˚C
c denotes a significant difference vs. 6 to 10˚C
d denotes a significant difference vs. 11 to 20˚C
e denotes a significant difference vs. >20˚C. Number of data is referred to team performance data points
https://doi.org/10.1371/journal.pone.0288494.t001
PLOS ONE
Cold exposure and match running performance
PLOS ONE | https://doi.org/10.1371/journal.pone.0288494
July 11, 2023
4 / 9
with significant small differences vs. all the lower temperature ranges. In the >20˚C range, the
total distance shows a further increase, with significant moderate differences vs. all the lower
temperature ranges. A similar trend was observed for running distance, with no significant dif-
ferences between ranges from <-5˚C to 6 to 10˚C, a decrease in the 11 to 20˚C range, with sig-
nificant moderate differences vs. all the lower temperature ranges and a marked decrease in
the >20˚C range, with significant moderate to large differences vs. all lower temperature
ranges (Tables 1 and 2). The HSR distance was slightly though not significantly lower in the
<-5˚C range when compared to temperature ranges from -4 to 0˚C to 6 to 10˚C. In the 11 to
20˚C range, HSR was significantly lower than in the -4 to 0˚C, 1 to 5˚C, and 6 to 10˚C ranges,
with small differences. In the >20˚C range, HSR distance was significantly lower than all
lower temperature ranges, with moderate to large differences. Sprint distance was similar
across ranges from -4 to 0˚C to 11 to 20˚C, while it was significantly lower in the <-5˚C range
when compared to all higher temperature ranges except >20˚C with moderate differences,
and in the >20˚C range when compared to all lower temperature ranges except <-5˚C, with
differences ranging from small to moderate (Tables 1 and 2).
The linear mixed model analysis performed using ambient temperature as a quantitative
independent variable including matches when temperature was equal to zero or lower, when
examining the estimated fixed effect coefficients, revealed small (d = 0.21–0.23) though non-
significant (p >0.05) increases of total distance and running distance for a 1˚C decrease of
ambient temperature. High-speed running distance showed a very small (d = 0.07) yet non-
significant decrease with decreasing temperature. Conversely, there was a significant (p<0.01)
decrease of 19.2 m (approximately 1.6%) for every 1˚C decrease of ambient temperature
(d = 0.48, small).
Table 2. Standardized differences (ES), with 95% Confidence Intervals (CI), between temperature ranges (values of temperature categories in columns minus values
of temperature categories in rows).
<-5˚C
-4 to 0˚C
1 to 5˚C
6 to 10˚C
11 to 20˚C
Total distance
-4 to 0˚C
0.16 (-0.27 to 0.59)
1 to 5˚C
0.12 (-0.29 to 0.59)
-0.04 (-0.38 to 0.30)
6 to 10˚C
0.08 (-0.31 to 0.48)
-0.07 (-0.41 to 0.27)
-0.03 (-0.33 to 0.27)
11 to 20˚C
0.55* (0.18 to 0.92)
0.39* (0.09 to 0.70)
0.43* (0.17 to 0.70)
0.47* (0.21 to 0.72)
>20˚C
1.18* (0.80 to 1.57)
1.02* (0.70 to 1.35)
1.06* (0.78 to 1.35)
1.10* (0.83 to 1.37)
0.63* (0.40 to 0.86)
Running distance
-4 to 0˚C
0.11 (-0.32 to 0.53)
1 to 5˚C
0.01 (-0.39 to 0.41)
-0.10 (-0.43 to 0.24)
6 to 10˚C
0.10 (-0.29 to 0.49)
-0.01 (-0.34 to 0.32)
0.09 (-0.21 to 0.38)
11 to 20˚C
0.72* (0.35 to 1.08)
0.61* (0.32 to 0.91)
0.71* (0.45 to 0.97)
0.62* (0.37 to 0.87)
>20˚C
1.73* (1.36 to 2.11)
1.63* (1.31 to 1.94)
1.72* (1.45 to 2.00)
1.64* (1.37 to 1.90)
1.01* (0.79 to 1.24)
High-speed distance
-4 to 0˚C
-0.28 (-0.90 to 0.34)
1 to 5˚C
-0.33 (-0.91 to 0.25)
-0.05 (-0.55 to 0.44)
6 to 10˚C
-0.34 (-0.91 to 0.23)
-0.06 (-0.55 to 0.42)
-0.01 (-0.44 to 0.42)
11 to 20˚C
0.21 (-0.33 to 0.74)
0.48* (0.05 to 0.92)
0.54* (0.16 to 0.92)
0.55* (0.18 to 0.91)
>20˚C
1.31* (0.76 to 1.86)
1.59* (1.13 to 2.05)
1.64* (1.24 to 2.05)
1.65* (1.27 to 2.04)
1.11* (0.78 to 1.43)
Sprint distance
-4 to 0˚C
-0.70* (-1.23 to -0.18)
1 to 5˚C
-0.85* (-1.34 to -0.35)
-0.15 (-0.57 to 0.27)
6 to 10˚C
-0.93* (-1.42 to -0.45)
0.23 (-0.63 to 0.18)
-0.08 (-0.45 to 0.29)
11 to 20˚C
-0.84* (-1.29 to -0.39)
-0.13 (-0.50 to 0.24)
0.02 (-0.31 to 0.33)
0.10 (-0.21 to 0.40)
>20˚C
0.18 (-0.29 to 0.65)
0.52* (0.13 to 0.92)
0.67* (0.32 to 1.06)
0.75* (0.42 to 1.08)
0.66* (0.38 to 0.94)
* denotes a significant difference (p<0.05).
https://doi.org/10.1371/journal.pone.0288494.t002
PLOS ONE
Cold exposure and match running performance
PLOS ONE | https://doi.org/10.1371/journal.pone.0288494
July 11, 2023
5 / 9
Discussion
The present study assessed the relationships between low ambient temperature and physical
match behavior in elite soccer players. To this aim, competitive matches played across five con-
secutive seasons in the Russian Premier League were examined. Although there is a mid-sea-
son winter break (mid-December to end-February) during the extremely cold months of the
year in this League, there is still a significant number of matches played under cold or very
cold conditions, mainly in the weeks just before or after the winter break. Thus, an evaluation
into how low temperatures may impact the physical match output of soccer players in a real-
world setting is of significant interest to the support staff tasked with maintaining the well-
being of players.
Together, our results show that a very cold (<-5˚C) ambient temperature at the start of the
match has no impact on total distance covered or distance when running up to 5.5 m/s, while
there is a trend suggesting a slight reduction of HSR distance (5.5 to 7.0 m/s), and an evident
decrease in sprint distance (>7.0 m/s). On average, in matches played at a temperature of -5˚C
or less, team sprint distance was approximately 10–15% lower than in matches played at less
cold or warmer temperatures (Table 1). Moreover, with a mixed-effects regression analysis, a
moderate but relevant decrease of approximately 19.2 m (1.6%) of sprint distance covered by
the team for every 1˚C decrease of ambient temperature when the temperature was 0˚C or less
was reported. Thus, players performed less sprinting at low temperatures. Overall, the results
support the study hypothesis that physical match performance is negatively affected by playing
at low temperatures, though the impact of cold environment was evident, among the examined
variables, distances covered at high- to maximal intensity were notable.
To date, a number of previous studies have investigated how environmental conditions,
including ambient temperature, affects physical performance in soccer players [1, 3, 5, 11, 15–
18]. There is some evidence that total distance covered during a soccer match, as well as HSR
distance, are reduced in hot environments compared to neutral temperature conditions [1, 11,
16]. Though the present study mainly focused on the impact of low temperatures, the sample
included all available matches from the examined Russian Premier League seasons, thus allow-
ing a comparison between physical match performance across all temperature conditions.
Consistently with previous studies [1, 11, 17], it was observed that total distance covered, as
well as running and HSR distances, were reduced in the 11 to 20˚C and more evidently in the
>20˚C condition than in conditions where the temperature was between -4 and 20˚C. Sprint
distance also showed a decrease when the temperature at the start of the match was 20˚C or
higher when compared to lower temperature ranges (Table 1).
Currently, scant literature is available examining the impact of low temperatures on physi-
cal match performance in elite soccer. Carling et al. [5], analyzed physical match performance
in players from a French Ligue 1 team, and showed no detrimental effect of low temperature
on distances covered in the 0.0–4.0, 4.0–5.5, and >5.5 m/s speed ranges. However, the authors
employed <5˚C as the lower temperature condition. Our findings showed no substantial effect
of temperature in the 1 to 5˚C range, on physical match outcomes, while the effects of low tem-
perature became more evident at -5˚C or less. Therefore, the present results are consistent
with those of Carling and colleagues [5]. Link and Weber [11] examined the effects of ambient
temperature on total distance covered by players from 38 teams from the top two German lea-
gues, collected across 1211 league matches. Those authors used a lower temperature range of
<-5˚C, as in the present study, and also reported no substantial differences in the total distance
covered between matches starting at a temperature of -5˚C or less and matches played in the
-4˚ to 13˚C and 14 to 27˚C temperature ranges. This observation is also consistent with our
finding that total distance covered is unaffected by low temperatures (Tables 1 and 2).
PLOS ONE
Cold exposure and match running performance
PLOS ONE | https://doi.org/10.1371/journal.pone.0288494
July 11, 2023
6 / 9
A hypothesis that may partly explain why sprint distance may decrease during soccer
matches played in very cold (<-5˚C) temperatures is that of a reduced sprinting capacity, that
may occur due to the negative effects of low temperature on muscle function and power pro-
duction. Indeed, it has been shown that speed, as well as agility and lower-limb power, is
impaired immediately after the application of different cryotherapy modalities [18–20]. To our
knowledge, only the study of Carlson et al. [21] investigated the effects of whole-body exposure
to low temperature on the outcomes of lower-limb power, agility and sprint tests. Reduced ver-
tical jump and agility performance was observed, although unaffected sprinting performance,
in recreational athletes, after a 15-minute cool (6.1˚C) exposure vs. a thermoneutral (17.2˚C)
environment was reported [21]. However, the temperature administered in this study for the
cool environment is higher than the lower ranges of ambient temperature in the present sam-
ple of soccer matches (<-5˚C). A colder temperature (-14˚C) was utilized by Wiggen and col-
leagues [22] to examine the effect of cold exposure on double poles sprint performance in
cross-country skiers. The authors reported lower performance in terms of power output at
14˚C vs. 6˚C temperatures. Such evidence supports the assumption that very cold ambient
temperature can impair sprinting performance in soccer players, however future studies are
warranted to further explore the effect of low ambient temperature on sprinting performance
in training and competitive match-play in players from different leagues.
A further factor that may potentially have an impact on reduced sprint distance at low tem-
peratures is possibly related to playing surface conditions. At sub-zero temperatures, the play-
ing turf may be frozen or partly frozen and slippery, decreasing stability and traction,
increasing the player ground contact and surface interaction, and therefore making it more
difficult for players to execute maximal or near-maximal actions, including sprints, than in
normal conditions. Additionally, the team playing strategy may be altered due to elements
such as unpredictable ball roll, bounce and ball-speed, leading players to execute more shorter
passes and thus reducing the number of longer, forward passes and subsequent physical
actions involving long (>30 m) sprinting. In this respect, in future investigations it would be
practically interesting to assess how the technical and tactical performance is affected when
matches are performed in cold or very cold conditions.
Limitations
Despite the findings, several limitations of this study were identified. Due to data availability, only
ambient temperature at the start of the match was examined as an indicator of environmental
conditions, while other atmospheric conditions such as atmospheric pressure, wind, wind child,
humidity, heat index, wet-bulb globe temperature, precipitation and cloudiness that may also
influence physical performance were not considered. However, depending on other factors such
as kick-off time, ambient temperature could to some extent increase or decrease throughout the
duration of approximately 2-hours of a soccer match, especially if the kick-off time is in the even-
ing hours. A perspective for future studies is therefore to consider the average ambient tempera-
ture during the match. Furthermore, our study lacked data related to other situational variables
such as match location, that may also potentially modulate the impact of low temperatures on
physical match performance, or others such as match result and quality of the teams that could
also cause different running-based results considering the different scenario of the analyzed team.
For such reasons, it is recommended to consider those variables in future studies.
Conclusion
The present study has shown that sub-zero ambient temperatures, especially when equal to or
lower than -5˚C, are related to changes in the match running performance behavior in elite
PLOS ONE
Cold exposure and match running performance
PLOS ONE | https://doi.org/10.1371/journal.pone.0288494
July 11, 2023
7 / 9
soccer players. A novel finding of our study is that low temperatures are associated with
reduced sprint performance. The present results may be of real practical interest to coaching
staff who are responsible for improving and maintaining the health and well-being of soccer
players that regularly play in cold or very cold conditions.
Author Contributions
Conceptualization: Ryland Morgans, Eduard Bezuglov, Dave Rhodes, Jose Teixeira, Toni
Modric, Sime Versic, Rafael Oliveira.
Formal analysis: Rocco Di Michele.
Investigation: Ryland Morgans, Dave Rhodes, Jose Teixeira, Toni Modric, Sime Versic, Rafael
Oliveira.
Methodology: Ryland Morgans.
Project administration: Ryland Morgans.
Resources: Eduard Bezuglov.
Software: Rocco Di Michele.
Supervision: Ryland Morgans.
Validation: Ryland Morgans, Eduard Bezuglov, Rafael Oliveira.
Visualization: Ryland Morgans, Rocco Di Michele.
Writing – original draft: Ryland Morgans, Rocco Di Michele, Rafael Oliveira.
Writing – review & editing: Ryland Morgans, Eduard Bezuglov, Dave Rhodes, Jose Teixeira,
Toni Modric, Sime Versic, Rafael Oliveira.
References
1.
Ozgunen KT, Kurdak SS, Maughan RJ, et al. Effect of hot environmental conditions on physical activity
patterns and temperature response of football players. Scand J Med Sci Sports 2010; 20: 140–147.
https://doi.org/10.1111/j.1600-0838.2010.01219.x PMID: 21029201
2.
Nassis GP, Brito J, Dvorak J. et al. The association of environmental heat stress with performance:
analysis of the 2014 FIFA World Cup Brazil. Br J Sports Med 2015; 49: 609–613. https://doi.org/10.
1136/bjsports-2014-094449 PMID: 25690408
3.
Chmura P, Konefal M, Andrzejewski M. et al. Physical activity profile of 2014 FIFA World Cup players,
with regard to different ranges of air temperature and relative humidity. Int J Biometerol 2017; 61: 677–
684. https://doi.org/10.1007/s00484-016-1245-5 PMID: 27618828
4.
Draper G, Wright MD, Ishida A. et al. Do environmental temperatures and altitudes affect physical out-
puts of elite football athletes in match conditions? A systematic review of the ‘real world’ studies. Sci
Med Football 2022; In press. https://doi.org/10.1080/24733938.2022.2033823 PMID: 35068376
5.
Carling C, Dupont G, Le Gall F. et al. The effect of a cold environment on physical activity profiles in elite
soccer match-play. Int J Sport Med 2011; 32: 542–545. https://doi.org/10.1055/s-0031-1273711 PMID:
21563033
6.
No M and Kwak HM. Effects of environmental temperature on physiological responses during submaxi-
mal and maximal exercises in soccer players. Integr Med Res 2016; 5: 216–222. https://doi.org/10.
1016/j.imr.2016.06.002 PMID: 28462121
7.
Armstrong LE. Nutritional strategies for football: counteracting heat, cold, high altitude, and jet lag. J
Sports Sci 2006; 24: 723–740. https://doi.org/10.1080/02640410500482891 PMID: 16766501
8.
FIFA. R&D test report. Electronic Performance & Tracking Systems. https://www.fifa.com/technical/
football-technology/resource-hub?id=aca57303eb0449f2835ac891b1beeb24 (2022, accessed 16 May
2023).
9.
Morgans R, Bezuglov E, Orme P. et al. Technical and Physical Performance Across Five Consecutive
Seasons in Elite European Soccer. Int J Sports Sci Coach 2022; 0: 1–9.
PLOS ONE
Cold exposure and match running performance
PLOS ONE | https://doi.org/10.1371/journal.pone.0288494
July 11, 2023
8 / 9
10.
Morgans R, Orme P, Di Michele R. Impact of technical and physical performance on match outcome
over five elite European soccer seasons. Int J Sports Med Phys Fitness 2022; In press. https://doi.org/
10.23736/S0022-4707.22.14018-1 PMID: 35816143
11.
Link D and Weber H. Effect of ambient temperature on pacing in soccer depends on skill level. J
Strength Cond Res 2017; 31: 1766–1770. https://doi.org/10.1519/JSC.0000000000001013 PMID:
25992664
12.
Hopkins W, Marshall S, Batterham A, et al. Progressive statistics for studies in sports medicine and
exercise science. Med Sci Sports Exercise 2009; 41: 3–13. https://doi.org/10.1249/MSS.
0b013e31818cb278 PMID: 19092709
13.
Kleiman E. Data Management Tools for Real-Time Monitoring/Ecological Momentary Assessment Data
[R Package EMAtools version 0.1.3], 2017; Available online: https://cran.r-project.org/web/packages/
EMAtools/index.html (Accessed 25 Jul 2022).
14.
Cohen J. Statistical power analysis for the behavioral sciences (2nd Edition). 2013; Academic Press.
https://doi.org/10.4324/9780203771587.
15.
Chmura P, Liu H, Andrzejewski M. et al. Is there meaningful influence form situational and environmen-
tal factors on the physical and technical activity of elite football players? Evidence from the data of 5
consecutive seasons of the German Bundesliga. PLoS One 2021; 16: 1–16.
16.
Zhou C, Hopkins WG, Mao W. et al. Match performance of soccer teams in the Chinese super league–
effects of situational and environmental factors. Int J Environ Res Public Health 2019; 16: 1–13. https://
doi.org/10.3390/ijerph16214238 PMID: 31683754
17.
Mohr M, Nybo L, Grantham J. et al. Physiological responses and physical performance during football
in the heat. PLoS One 2012; 7: e39202. https://doi.org/10.1371/journal.pone.0039202 PMID:
22723963
18.
Cross KM, Wilson RW, Perrin DH. et al. Functional performance following an ice immersion to the lower
extremity. J Athl Train 1996; 31: 131–136.
19.
Richendollar ML, Darby LA, Brown TM. et al. Ice bag application, active warm-up, and 3 measures of
maximal functional performance. J Athl Train 2006; 41: 364–370. PMID: 17273459
20.
Patterson SM; Udermann BE; Doberstein ST. et al. The effect of cold whirlpool on power, speed, agility,
and range of motion. J Sports Sci Med 2008; 7: 387–394.
21.
Carlson LA, Fowler C, Lawrence MA. et al. Agility and vertical jump performances are impacted by
acute cold exposure. J Strength Cond Res 2019; 33: 1648–1652.
22.
Wiggen ON, Waagaard SH, Heidelberg CT. et al. Effect of cold conditions on double poling sprint per-
formance of well-trained male cross-country skiers. J Strength Cond Res 2013; 27: 3377–3383. https://
doi.org/10.1519/JSC.0b013e3182915e7d PMID: 23539076
PLOS ONE
Cold exposure and match running performance
PLOS ONE | https://doi.org/10.1371/journal.pone.0288494
July 11, 2023
9 / 9
| The relationship between ambient temperature and match running performance of elite soccer players. | 07-11-2023 | Morgans, Ryland,Bezuglov, Eduard,Rhodes, Dave,Teixeira, Jose,Modric, Toni,Versic, Sime,Di Michele, Rocco,Oliveira, Rafael | eng |
PMC3518245 | RESEARCH ARTICLE
Open Access
Comparison of vertical ground reaction forces
during overground and treadmill running.
A validation study
Bas Kluitenberg1*, Steef W Bredeweg1, Sjouke Zijlstra1, Wiebren Zijlstra2,3 and Ida Buist1
Abstract
Background: One major drawback in measuring ground-reaction forces during running is that it is time consuming
to get representative ground-reaction force (GRF) values with a traditional force platform. An instrumented force
measuring treadmill can overcome the shortcomings inherent to overground testing. The purpose of the current
study was to determine the validity of an instrumented force measuring treadmill for measuring vertical
ground-reaction force parameters during running.
Methods: Vertical ground-reaction forces of experienced runners (12 male, 12 female) were obtained during
overground and treadmill running at slow, preferred and fast self-selected running speeds. For each runner, 7 mean
vertical ground-reaction force parameters of the right leg were calculated based on five successful overground
steps and 30 seconds of treadmill running data. Intraclass correlations (ICC(3,1)) and ratio limits of agreement (RLOA)
were used for further analysis.
Results: Qualitatively, the overground and treadmill ground-reaction force curves for heelstrike runners and
non-heelstrike runners were very similar. Quantitatively, the time-related parameters and active peak showed
excellent agreement (ICCs between 0.76 and 0.95, RLOA between 5.7% and 15.5%). Impact peak showed modest
agreement (ICCs between 0.71 and 0.76, RLOA between 19.9% and 28.8%). The maximal and average loading-rate
showed modest to excellent ICCs (between 0.70 and 0.89), but RLOA were higher (between 34.3% and 45.4%).
Conclusions: The results of this study demonstrated that the treadmill is a moderate to highly valid tool for the
assessment of vertical ground-reaction forces during running for runners who showed a consistent landing strategy
during overground and treadmill running. The high stride-to-stride variance during both overground and treadmill
running demonstrates the importance of measuring sufficient steps for representative ground-reaction force values.
Therefore, an instrumented treadmill seems to be suitable for measuring representative vertical ground-reaction
forces during running.
Keywords: Running, Kinetics, Biomechanics, Validity, Overuse injuries
Background
One major drawback in measuring ground-reaction
forces during running is that it is time consuming to get
representative ground-reaction force (GRF) values with a
traditional force platform. A single force platform is only
capable of measuring GRFs of one single stance phase
per trial [1,2]. Therefore, multiple force platforms are
necessary for measuring consecutive steps which is space
consuming and expensive. The limited length of a run-
way, also makes it difficult to simulate natural running
at a constant speed in a laboratory situation [3]. For de-
tection of small differences in GRFs during running,
however, it is important to record sufficient steps during
a stable running pattern [4]. An instrumented treadmill
capable of measuring GRFs can overcome the limitations
inherent to overground GRF testing during running
at a short runway. With an instrumented treadmill
it is possible to measure GRFs of multiple steps during
* Correspondence: [email protected]
1Center for Sports Medicine, University Medical Center Groningen,
Hanzeplein 1, Groningen, GZ 9713, The Netherlands
Full list of author information is available at the end of the article
© 2012 Kluitenberg et al.; licensee BioMed Central Ltd. This is an Open Access article distributed under the terms of the
Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use,
distribution, and reproduction in any medium, provided the original work is properly cited.
Kluitenberg et al. BMC Musculoskeletal Disorders 2012, :
http://www.biomedcentral.com/1471-2474//
one trial without interruptions in running speed, result-
ing in a more stable running pattern during the
measurements [3].
In running, most runners make first ground contact
with the posterior part of the foot, this is called heel-
strike running. This running style results in a typical ver-
tical GRF force-time curve that is characterized by two
peaks, the impact peak and the active peak, as depicted in
Figure 1. Magnitude of the impact peak is speed depen-
dent and occurs during the first 10% of stance (10-30ms)
[5]. The active peak is reached approximately during mid-
stance and can last up to 200ms. The absence of a separ-
ate impact peak in the force-time curve is typical for a
non-heelstrike runner, as depicted in Figure 1 [6]. Besides
a vertical component, GRFs also have an anterior-
posterior and medio-lateral component. During running,
the anterior-posterior force component shows a typical
braking and propulsive phase while the medio-lateral
force component is characterized by more variability [7].
Compared to the vertical GRF component, anterior-
posterior and medio-lateral forces are small [7].
An underlying assumption when using a treadmill for
running analysis is that running on a treadmill is similar
to
overground
running.
A
comparison
of
spatio-
temporal variables during overground and treadmill
running was made in several studies. During treadmill
running, runners tend to run with a shortened stride
length and an increased stride rate [3,8,9]. Despite of
these spatio-temporal differences, only small differences
in knee flexion and a more flattened landing style during
treadmill running were observed [3,10]. An overground-
treadmill comparison with respect to GRFs was made in
only two studies [1,3]. No systematic errors or extraor-
dinary differences in vertical GRFs were found. Impact
peaks and loading rates, however, have not been studied
in these previous studies.
The purpose of this study was to determine the valid-
ity of a custom made instrumented force measuring
treadmill to measure vertical GRF parameters during
running. Validation of the treadmill was performed by
comparing overground and treadmill measured vertical
GRF parameters during running.
Methods
Participants
Twenty-four experienced runners (12 male, 12 female)
between 18 and 35 years old participated in this study.
The runners were voluntarily recruited by contacting
two local track and field clubs. The criteria for inclusion
in this study included a minimal training frequency of
two times a week for at least a period of one year. Run-
ners who reported an injury at time of measurement
were excluded. Both heelstrike and non-heelstrike run-
ners were allowed to participate in this study. All partici-
pants signed an informed consent before measurements
started. The study was approved by the Medical Ethical
Committee at the University Medical Center Groningen,
The Netherlands; M12.112668.
Overground measurements
During the overground measurements, GRFs were mea-
sured at three different individual speed conditions. Par-
ticipants were instructed to run at their preferred speed
(running speed for a normal endurance run), slower
speed (running speed during a warming-up), and a faster
speed (10km race speed) respectively. GRFs were col-
lected with a force platform (0.60m x 0.40m) which was
mounted in the middle of a 17.5m long runway. The
sample frequency of the force platform was set at
1000Hz. Running speed was monitored with two pairs of
photocells placed 2.5m before and after the force
platform.
0
20
40
60
80
100
0
0.5
1
1.5
2
2.5
3
% Of Stance
Vertical-GRF (BW)
Fz2
Fz1
LR
tFz1
tFz2
Contact Time (CT)
0
20
40
60
80
100
0
0.5
1
1.5
2
2.5
3
% Of Stance
Vertical-GRF (BW)
Fz2
tFz2
LR
Contact Time (CT)
Heelstrike landing
Non-Heelstrike landing
Figure 1 Outcome measures in a typical vertical ground-
reaction force (GRF) curve for a heelstrike runner and a non-
heelstrike runner. Figure is created from personal data.
Kluitenberg et al. BMC Musculoskeletal Disorders 2012, :
Page 2 of 8
http://www.biomedcentral.com/1471-2474//
Before the actual overground measurements started,
the participants performed several accommodation runs.
During these accommodation runs, the exact start pos-
ition for the measurements was determined. The start
position was based on the position of foot placement at
the force platform. Foot strike of the right foot should
be completely at the force platform without an alteration
in running pattern. An alteration can indicate aiming for
the force platform, which modifies the GRF pattern [11].
Position of foot placement and running pattern were
evaluated on sight. When participants were able to run
several trials at the same speed, while landing with the
right foot completely placed at the force platform, with-
out visible alterations in running pattern, the actual
measurements started. Since the participants were tested
at three different speed conditions, accommodation runs
were performed for each speed condition (preferred,
slow and fast). The accommodation runs for the pre-
ferred speed were combined with a short warming-up
and took longer (approximately 10 min), where the
subsequent accommodation runs took approximately
5 minutes.
During the actual measurements, GRF data were cap-
tured until five clean strikes of the right leg within a 5%
speed range were recorded for all speed conditions.
Trials with visible alterations in running pattern were
not included in these five clean strikes. Afterwards, the
mean running speed of the five steps was calculated for
each speed condition.
Treadmill measurements
In this study, an instrumented treadmill (Entred, Force-
link, Culemborg, The Netherlands) with a running sur-
face of 1.60m by 0.60m that was driven by a 1.8 kW
motor was used to measure vertical GRFs during
running. The treadmill was equipped with three strain
gage
force
transducers
(ACB-500kg, Vishay
Revere
Transducers, Breda, The Netherlands) which were con-
nected to bridge
amplifiers. The force
transducers
were mounted on a stiff plate which was enforced with a
non-deformable frame and were positioned as shown in
Figure 2. The signals from the amplifiers were sampled
at 1000 Hz, digitized into a 16-bit signal by an AD con-
verter (PCI-6220, National Instruments, Austin, TX,
USA) and were connected to a computer.
Before the treadmill measurements started, partici-
pants started with an accommodation run of 10 minutes
at 10 km·h-1. After this accommodation period, partici-
pants were tested at three different individual speed con-
ditions (slow, preferred and fast). Treadmill speed was
matched to the average overground running speed for
each speed condition because GRF parameters are speed
dependent [7]. The three speed conditions lasted three
minutes and were offered in random order. GRFs were
recorded during the last 30 seconds of each speed condi-
tion. When treadmill measurements were finished, parti-
cipants were given the opportunity for cooling-down at
the treadmill. All measurements were conducted while
participants were running in their personal running
shoes.
Data analysis
Vertical force data from both the force platform and the
treadmill were processed using custom programs written
in MATLAB R2010a (The MathWorks, Inc, Natick,
MA). All steps which were recorded during the treadmill
measurement were entered into the analysis. A 13-point
moving average low-pass filter with a cut-off frequency
of 33.3Hz was used to filter the GRF data that was
recorded during the overground and treadmill measure-
ments. Foot strikes in the overground and treadmill data
were detected with a threshold of 30 Newton for impact
and toe-off phase. Outcome measures for all right foot
steps were identified, as described in Table 1. For each
speed condition outcome measures of each participant
were averaged. A distinction between heelstrike and
S1
S2
S3
Front
0.60m
1.60m
Figure 2 Positioning of the three strain gage force transducers
S1, S2 and S3.
Kluitenberg et al. BMC Musculoskeletal Disorders 2012, :
Page 3 of 8
http://www.biomedcentral.com/1471-2474//
non-heelstrike landing patterns was made based on the
existence of an impact peak Fz1. Peak values Fz1 and
Fz2 and the loading-rate were normalized to bodyweight.
Statistical analysis
A within-subject repeated measures design was used to
determine the validity of the instrumented treadmill for
measuring vertical GRF parameters during running.
Therefore, a two-way mixed-effects, consistency, single
measure
intraclass
correlation
coefficient
(ICC(3,1))
model was used to examine the agreement between
overground and treadmill measured GRF-parameters.
Interpretation of the intraclass coefficients were as fol-
lows: poor (0 – 0.39), modest (0.4 – 0.74), or excellent
(0.75 – 1) [12]. ICCs were calculated by using SPSS
(SPSS inc. Version 18.0, Chicago, IL, U.S.A.). Besides the
intraclass correlations, Bland-Altman plots were used to
examine the agreement between overground and tread-
mill measurements [13]. These plots were made for each
outcome measure and each speed condition. The limits
of agreement (LOA) were calculated (mean difference
+/− 1.96 times the standard deviation of the difference).
Also ratio limits of agreement (RLOA) were calculated
to
express
the
LOA
as
percentage
of
the
mean
overground-treadmill value. The upper and lower LOA
and the RLOA provide insight into how much random
variation may be influencing the measurements.
Results
Ground-reaction force (GRF) parameters of a different
landing strategy cannot be compared, therefore only
GRF parameters of participants who showed a consistent
landing strategy during overground and treadmill run-
ning within a speed condition were examined. During
overground running at preferred speed, 19 participants
showed a heelstrike (HS) landing, while 16 of these run-
ners showed a HS landing during treadmill running.
This shows that 82.4% of the runners used a similar
landing strategy during treadmill running at preferred
speed. Results for the two other speeds can be found in
Table 2.
Qualitatively,
the
overground
and
treadmill
GRF
curves for both HS and NHS running at slow, preferred
and fast running speeds, were very similar, as can be
seen in Figure 3. In Table 3 a quantitative evaluation of
the vertical GRF-parameters of both HS and NHS run-
ners can be found. The levels of agreement between
overground and treadmill running for the time related
variables (tFz1, tFz2 and CT) were excellent (ICCs be-
tween 0.76 and 0.95 and RLOAs between 5.7% and
15.5%). Also the active peak (Fz2) measured with both
devices showed excellent agreement (ICCs between 0.77
and 0.89, RLOAs between 7.8% and 9.9%). Modest
agreement was found for the impact peak, Fz1 (ICCs be-
tween 0.71 and 0.76, RLOAs between 19.9% and 28.8%).
Maximal loading rate (LR) and average loading rate
(ALR) also showed modest to excellent intraclass corre-
lations (ICCs between 0.70 and 0.89), however the ratio
limits of agreement were higher (RLOA values between
34.3% and 45.4%).
Discussion
The instrumented treadmill is capable of measuring ver-
tical ground-reaction forces (GRFs) during running and
seems to be a usable tool for simulating overground run-
ning kinetics. The results of this study demonstrated
that the instrumented treadmill is a highly valid tool for
the assessment of the vertical GRF parameters: tFz1,
tFz2, CT and Fz2 and moderately valid for the assess-
ment of Fz1, LR and ALR for runners who showed a
consistent landing strategy during overground and tread-
mill running. A qualitative evaluation of the overground
and treadmill vertical GRF curves as shown in Figure 3,
Table 1 Definition of outcome measures, as displayed in
Figure 1
Outcome measure
Description
Fz1
Local maximum in the vertical GRF data,
normalized to body weight (BW).
Fz2
Maximum value in the vertical GRF data,
normalized to BW.
LR
The steepest part of the vertical GRF curve,
from stance to impact peak. Expressed
in BW/s.
ALR
Average loading rate, the slope of the line
from 20% to 80% of Fz1. Expressed in BW/s.
tFz1
Time from heelstrike to Fz1 in ms.
tFz2
Time from heelstrike to Fz2 in ms.
CT
Contact time, from heelstrike to toe-off in ms.
Outcome measures for the overground and treadmill data were identified with
the same routine. Foot strikes were detected with a threshold of 30 Newton
for both heelstrike and toe-off.
Table 2 Overground landing strategy compared to treadmill landing strategy, displayed as number of persons and
corresponding percentages of runners who showed a consistent landing strategy
Heelstrike landing
Non-heelstrike landing
Overground
Treadmill
Consistency
Overground
Treadmill
Consistency
Slow
17
14
82.4%
7
5
71.4%
Preferred
19
16
84.2%
5
5
100.0%
Fast
12
12
100.0%
12
6
50.0%
Kluitenberg et al. BMC Musculoskeletal Disorders 2012, :
Page 4 of 8
http://www.biomedcentral.com/1471-2474//
demonstrated that the vertical GRFs for both the heel-
strike (HS) runners and the non-heelstrike (NHS) run-
ners were similar during overground and treadmill
running. The excellent intraclass correlations and low
limits of agreement for contact time (CT), time to im-
pact peak force (tFz1) and time to the active peak (tFz2)
reflect this qualitative similarity. After all, these para-
meters show that the timing of peak values in the verti-
cal GRF curve is not different for overground and
treadmill running. The qualitative similarity of these
GRF curves was also observed in other studies [1,3]. In
the current study, the overground and treadmill mea-
sured active peak (Fz2) showed no noteworthy differ-
ences. This is in accordance with the results of Riley
et al., who also compared overground and treadmill run-
ning kinetics in a group of 20 runners [3]. Overground
and treadmill measured impact peaks (Fz1), maximal
loading rates (LR) and average loading rates (ALR),
showed less consistent results with modest to excellent
intraclass correlations and wider limits of agreement. To
our knowledge this study is the first to compare
overground and treadmill measured impact peaks and
loading rates during running, therefore it is not possible
to evaluate these results with previous studies.
For an overground-treadmill comparison with respect
to vertical GRF parameters, a consistent landing strategy
during both running conditions (overground and tread-
mill) is required. While most runners showed a consist-
ent landing strategy during overground and treadmill
running, some runners switched to another landing
strategy. During slow and preferred running speed, these
inconsistent runners mostly switched from an over-
ground HS landing to a NHS landing during treadmill
running. Considering that this behavior is in line with
the more flattened landing style as observed in a previ-
ous study [14], it is likely that these inconsistencies in
landing strategy are the result of accommodation to
treadmill running. At fast self selected speed, however,
the inconsistent runners switched from a NHS to a HS
landing during treadmill running. These differences in
landing strategy may indicate overground and treadmill
differences in anterior-posterior GRFs which were not
0
20
40
60
80
100
0
0.5
1
1.5
2
2.5
3
% of stance
Vertical-GRF (BW)
Slow running speed (N=14)
0
20
40
60
80
100
0
0.5
1
1.5
2
2.5
3
% of stance
Vertical-GRF (BW)
Preferred running speed (N=16)
0
20
40
60
80
100
0
0.5
1
1.5
2
2.5
3
% of stance
Vertical-GRF (BW)
Fast running speed (N=12)
0
20
40
60
80
100
0
0.5
1
1.5
2
2.5
3
% of stance
Vertical-GRF (BW)
Slow running speed (N=5)
0
20
40
60
80
100
0
0.5
1
1.5
2
2.5
3
% of stance
Vertical-GRF (BW)
Preferred running speed (N=5)
0
20
40
60
80
100
0
0.5
1
1.5
2
2.5
3
% of stance
Vertical-GRF (BW)
Fast running speed (N=6)
Overground
Treadmill
Overground
Treadmill
Non-heelstrike running
Heelstrike running
Figure 3 Average GRF plots from all runners for overground (mean, solid black line; ± SD, dotted black lines) and treadmill
running (mean, solid grey line) at slow, preferred and fast running speed for heelstrike and non-heelstrike runners. Forces are in
body weight (BW).
Kluitenberg et al. BMC Musculoskeletal Disorders 2012, :
Page 5 of 8
http://www.biomedcentral.com/1471-2474//
Table 3 Outcome measures for overground and treadmill running
OG mean ± SD
TM mean ± SD
ICC(3,1) (95%CI)
Mean diff (LOA)
diff (lowLim, upLim)
RLOA
(%)
Fz1 (BW)
HS
Slow
1.67 ± 0.26
1.70 ± 0.23
0.74 (0.37, 0.91)
0.03 (−0.32, 0.38)
20.8
Preferred
1.94 ± 0.45
1.93 ± 0.30
0.71 (0.35, 0.89)
−0.01 (−0.57, 0.55)
28.8
Fast
1.94 ± 0.25
2.06 ± 0.32
0.76 (0.35, 0.92)
0.12 ( −0.28, 0.52)
19.9
Fz2 (BW)
HS
Slow
2.54 ± 0.20
2.53 ± 0.18
0.77 (0.49, 0.91)
−0.02 (−0.27, 0.23)
9.9
Preferred
2.70 ± 0.26
2.65 ± 0.25
0.89 (0.76, 0.96)
−0.03 (−0.25, 0.17)
7.9
Fast
2.77 ± 0.24
2.70 ± 0.22
0.86 (0.67, 0.95)
−0.06 (−0.27, 0.15)
7.8
NHS
Slow
2.56 ± 0.17
2.55 ± 0.20
Preferred
2.61 ± 0.15
2.58 ± 0.13
Fast
2.79 ± 0.15
2.78 ± 0.20
LR (BW/s)
HS
Slow
81.11 ± 25.62
87.28 ± 23.39
0.76 (0.47, 0.90)
3.25 (−28.62, 35.12)
39.9
Preferred
95.34 ± 26.67
105. 33 ± 25.08
0.80 (0.57, 0.91)
6.11 (−26.21, 38.42)
34.3
Fast
104.40 ± 29.29
118.08 ± 33.73
0.70 (0.36, 0.88)
7.17 (−37.69, 52.02)
42.7
NHS
Slow
70.03 ± 14.68
65.09 ± 13.74
Preferred
77.00 ± 22.35
74.25 ± 16.47
Fast
95.81 ± 26.02
87.41 ± 18.74
ALR (BW/s)
HS
Slow
68.89 ± 20.26
73.92 ± 20.22
0.84 (0.63, 0.93)
2.98 (−24.74, 30.70)
45.3
Preferred
82.14 ± 21.38
88.70 ± 20.75
0.89 (0.74, 0.95)
3.60 (−23.01, 30.21)
36.4
Fast
90.70 ± 23.66
100.77 ± 29.10
0.86 (0.67, 0.95)
4.08 (−31.26, 39.42)
45.4
NHS
Slow
33.96 ± 6.07
31.21 ± 5.01
Preferred
47.09 ± 22.92
33.78 ± 4.20
Fast
43.63 ± 13.89
36.00 ± 4.20
tFz1 (ms)
HS
Slow
35 ± 4.08
35 ± 4.86
0.76 (0.40, 0.92)
0.0 ( −5.4, 5.4)
15.5
Preferred
34 ± 4.42
34 ± 3.35
0.82 (0.56, 0.93)
0.3 ( −4.4, 5.0)
13.8
Fast
32 ± 5.00
33 ± 4.88
0.87 (0.61, 0.96)
0.6 ( −3.7, 4.8)
13.0
tFz2 (ms)
HS
Slow
112 ± 13.55
109 ± 10.22
0.84 (0.63, 0.94)
−1.8 (−15.4, 11.8)
12.6
Preferred
102 ± 13.28
100 ± 12.19
0.94 (0.85, 0.97)
−1.6 (−10.1, 6.8)
8.5
Fast
99 ± 10.00
96 ± 11.47
0.87 (0.68, 0.95)
−3.0 (−13.5, 7.5)
11.0
NHS
Slow
102 ± 13.00
103 ± 15.00
Preferred
99 ± 12.00
98 ± 11.00
Fast
92 ± 80 0
91 ± 10.00
CT (ms)
HS
Slow
258 ± 22.00
254 ± 21.13
0.92 (0.80, 0.97)
−4.0 (−21.4, 13.4)
6.9
Preferred
232 ± 23.34
232 ± 20.49
0.92 (0.82, 0.97)
−2.0 (−17.2, 13.2)
6.6
Fast
223 ± 21.00
220 ± 21.14
0.95 (0.87, 0.98)
−3.3 (−15.6, 9.1)
5.7
NHS
Slow
240 ± 17.00
237 ± 19.00
Preferred
229 ± 12.00
222 ± 12.00
Fast
213 ± 12.00
206 ± 12.00
Intraclass correlations, mean-differences with limits of agreement (LOA), and ratio limits of agreement (RLOA) were reported. Both HS and NHS runners were taken
into account in the statistical analysis. Number of participants: HS (slow: N=14, preferred: N=16, fast: N=12), NHS (slow: N=5, preferred: N=5, fast: N=6).
HS: Heelstrike-runner, NHS: Non-Heelstrike-runner, CI: Confidence Interval, LOA: Limit of Agreement, RLOA: Ratio Limit of Agreement, OG: Overground, TM:
Treadmill, BW: Body Weight.
Slow running speed: HS runners: 11.0 ± 1.3 km·h-1, NHS runners: 10.9 ± 1.5 km·h-1.
Preferred running speed: HS runners: 12.7 ± 1.6 km·h-1, NHS runners: 11.8 ± 1.5 km·h-1.
Fast running speed: HS runners: 14.1 ± 2.0 km·h-1, NHS runners: 13.9 ± 1.9 km·h-1.
Kluitenberg et al. BMC Musculoskeletal Disorders 2012, :
Page 6 of 8
http://www.biomedcentral.com/1471-2474//
compared in the current study. The results of this study
demonstrated that the inconsistencies in landing strategy
are smallest during running at preferred speed. There-
fore, to maximize certainty, it can be recommended to
determine landing strategy with a treadmill measure-
ment at preferred running speed.
The use of a treadmill in a research setting has been
subject of much debate. Several factors are mentioned
which may cause biomechanical differences between
overground
and
treadmill
running
[9].
First,
non-
mechanical factors as accommodation to the changed
visual and auditory surroundings or fear during treadmill
running may result in differences between overground
and treadmill running biomechanics [15]. Second, differ-
ences in air resistance may have an effect on treadmill
running form [16]. The effects of air resistance on run-
ning kinematics, however, will only be visible during
running at high speeds [17]. Third, intra-stride belt
speed variations, due to an energy exchange between the
treadmill and the runner, can cause differences in run-
ning kinematics compared to overground running. In
particular low powered treadmills are more sensitive for
opposite forces acting on the belt during running, result-
ing in larger belt speed variations. These variations in
belt speed may lead to biomechanical differences during
treadmill running when compared to overground run-
ning
[15].
Fourth,
during
running,
leg
stiffness
is
adjusted to the stiffness of the running surface [18].
Adjusting leg stiffness results in subtle changes in the
kinematics of the lower extremity [19]. Therefore, differ-
ences in running surface may lead to biomechanical dif-
ferences when comparing overground and treadmill
running.
Several studies compared overground and treadmill
running biomechanics [3,8,14]. Even though runners
tend to run with a shortened stride length and an
increased stride rate during treadmill running [3,8,9],
overground and treadmill running kinematics are re-
markably similar [3,9,14]. Only small differences in knee
and ankle joint kinematics were reported. Nigg et al.
observed a more flattened landing style during treadmill
running [14]. Riley et al. did not find differences in ankle
joint kinematics, but did find differences in minimal and
maximal knee flexion [3]. Maximal knee flexion was
lower and minimal flexion was higher during treadmill
running, which could be a result of the observed de-
crease in flight phase and higher stride rate [3]. Thus,
despite the theoretical factors which may influence
treadmill running biomechanics, only small differences
in overground and treadmill kinematics were observed.
In the current study, also no significant differences in
GRF parameters between overground and treadmill run-
ning were found. These findings are in line with previ-
ous studies where overground and treadmill running
kinetics were compared [1,3]. The between person vari-
ance in Fz1, LR and ALR during both overground and
treadmill running was high, as indicated by the high
standard deviations for these parameters. Stride-to-stride
variance for these parameters was also high, which
demonstrates the importance of measuring sufficient
steps for representative GRF values. This is especially
important for detecting small differences between differ-
ent conditions or persons [20]. Because a treadmill
makes it possible to measure multiple steps during one
test trial, it can be argued that a treadmill measurement
is more suitable for detecting small differences in verti-
cal GRFs during running. However, this assumption was
not assessed in the current study.
Since the treadmill used in the current study only is
capable of measuring vertical GRFs it cannot be used to
assess joint kinetics using the standard inverse dynamics
methodology, because anterior-posterior and medio-
lateral GRFs are also needed for these calculations. It
should also be noted that the inconsistencies in landing
strategy may indicate differences in anterior-posterior
GRFs between overground and treadmill running. Fur-
thermore, this instrumented treadmill would have lim-
ited usefulness for walking studies, because the double
support phase in walking cannot be measured directly.
For measuring GRFs during walking, an instrumented
split-belt treadmill may be more convenient.
A limitation of this study was that participants first
performed the overground measurements after which
the treadmill measurements started. Due to this fixed
order of the measurements, fatigue may have influenced
the later treadmill measurements [21]. Nevertheless, this
influence is expected to be low, since all participants
were experienced runners who did not have to deliver a
maximal performance and participants did not show
signs of exaggerated fatigue during the measurements.
Conclusions
The results of this study demonstrated the treadmill is a
moderate to highly valid tool for the assessment of verti-
cal GRFs during running for runners who showed a con-
sistent landing strategy during overground and treadmill
running. Therefore, an instrumented treadmill can be
used to measure vertical GRF parameters which corres-
pond to normal overground values during running.
In a future study, the treadmill can be used to measure
vertical GRF parameters in a large group of runners, for
instance to identify possible kinetic risk-factors for run-
ning related injuries prospectively.
Abbreviations
GRF: Ground-reaction force; HS: Heelstrike; NHS: Non-heelstrike; Fz1: Impact
peak; Fz2: Active peak; LR: Loading Rate; ALR: Average Loading Rate;
CT: Contact Time; tFz1: Time to impact peak; tFz2: Time to active peak;
Kluitenberg et al. BMC Musculoskeletal Disorders 2012, :
Page 7 of 8
http://www.biomedcentral.com/1471-2474//
BW: Body Weight; ICC: Intraclass correlation; LOA: Limit of Agreement;
RLOA: Ratio Limit of Agreement.
Competing interests
In this study, an instrumented treadmill was used. The research group had
no financial or other interest in the treadmill product or distributor of the
treadmill. The project was not dependent on external financial assistance
and the authors declare that they have no competing interests.
Authors’ contributions
SB, SZ, WZ and IB provided advice on the study design. BK recruited the
participants, was responsible for the data acquisition/analysis and wrote the
article. WZ provided advice on the data analysis. SB, SZ, WZ and IB
contributed to the content of the article. All authors read and approved the
final manuscript.
Author details
1Center for Sports Medicine, University Medical Center Groningen,
Hanzeplein 1, Groningen, GZ 9713, The Netherlands. 2Center for Human
Movement Sciences, University Medical Center Groningen, P.O. Box 196,
Groningen, AD 9700, The Netherlands. 3Institute of Movement and Sport
Gerontology, German Sport University Cologne, Cologne, Germany.
Received: 22 May 2012 Accepted: 1 November 2012
Published: 27 November 2012
References
1.
Kram R, Powell AJ: A treadmill-mounted force platform. J Appl Physiol
1989, 67(4):1692–1698.
2.
Kram R, Griffin TM, Donelan JM, Chang YH: Force treadmill for measuring
vertical and horizontal ground reaction forces. J Appl Physiol 1998,
85(2):764–769.
3.
Riley PO, Dicharry J, Franz J, Della Croce U, Wilder RP, Kerrigan DC: A
kinematics and kinetic comparison of overground and treadmill running.
Med Sci Sports Exerc 2008, 40(6):1093–1100.
4.
Bates B: Comment on ‘The Influence of Running Velocity and Midsole
Hardness on External Impact Forces in Heel-Toe-Running’. J Biomech
1989, 22(8–9):963–965.
5.
Hreljac A: Impact and overuse injuries in runners. Med Sci Sports Exerc
2004, 36(5):845–849.
6.
Williams KR: Biomechanics of running. Exerc Sport Sci Rev 1985,
13(1):389–441.
7.
Munro CF, Miller DI, Fuglevand AJ: Ground reaction forces in running: a
reexamination. J Biomech 1987, 20(2):147–155.
8.
Elliott BC, Blanksby BA: A cinematographic analysis of overground and
treadmill running by males and females. Med Sci Sports 1976, 8(2):84–87.
9.
Schache AG, Blanch PD, Rath DA, Wrigley TV, Starr R, Bennell KL: A
comparison of overground and treadmill running for measuring the
three-dimensional kinematics of the lumbo-pelvic-hip complex.
Clin Biomech (Bristol, Avon) 2001, 16(8):667–680.
10.
Nigg BM: Impact forces in running. Curr Opin Orthop 1997, 8(6):43–47.
11.
Challis J: The variability in running gait caused by force plate targeting.
J Appl Biomech 2001, 17(1):77–83.
12.
Fleiss J: The design and analysis of clinical experiments. New York: John Wiley;
1986.
13.
Bland J, Altman D: Statistical methods for assessing agreement between
two methods of clinical measurement. Lancet 1986, 327(8476):307–310.
14.
Nigg BM, De Boer RW, Fisher V: A kinematic comparison of overground
and treadmill running. Med Sci Sports Exerc 1995, 27(1):98–105.
15.
Savelberg H, Vorstenbosch M, Kamman E, Weijer J, Schambardt H:
Intra-stride belt-speed variation affects treadmill locomotion. Gait Posture
1998, 7(1):26–34.
16.
Pugh L: Oxygen intake in track and treadmill running with observations
on the effect of air resistance. J Physiol (Lond) 1970, 207(3):823–835.
17.
van Ingen Schenau GJ: Some fundamental aspects of the biomechanics
of overground versus treadmill locomotion. Med Sci Sports Exerc 1980,
12(4):257–261.
18.
Ferris D, Liang K, Farley C: Runners adjust leg stiffness for their first step
on a new running surface. J Biomech 1999, 8:787–794.
19.
Dixon S, Collop A, Batt M: Surface effects on ground reaction forces and
lower extremity kinematics in running. Med Sci Sports Exerc 2000,
32(11):1919–1926.
20.
Bates BT, Osternig LR, Sawhill JA, James SL: An assessment of subject
variability, subject-shoe interaction, and the evaluation of running shoes
using ground reaction force data. J Biomech 1983, 16(3):181–191.
21.
Morin J, Samozino P, Millet G: Changes in running kinematics, kinetics,
and spring-mass behavior over a 24-h run. Med Sci Sports Exerc 2011,
43(5):829–836.
doi:10.1186/1471-2474-13-235
Cite this article as: Kluitenberg et al.: Comparison of vertical ground
reaction forces during overground and treadmill running. A validation
study. BMC Musculoskeletal Disorders 2012 :.
Submit your next manuscript to BioMed Central
and take full advantage of:
• Convenient online submission
• Thorough peer review
• No space constraints or color figure charges
• Immediate publication on acceptance
• Inclusion in PubMed, CAS, Scopus and Google Scholar
• Research which is freely available for redistribution
Submit your manuscript at
www.biomedcentral.com/submit
Kluitenberg et al. BMC Musculoskeletal Disorders 2012, :
Page 8 of 8
http://www.biomedcentral.com/1471-2474//
| Comparison of vertical ground reaction forces during overground and treadmill running. A validation study. | 11-27-2012 | Kluitenberg, Bas,Bredeweg, Steef W,Zijlstra, Sjouke,Zijlstra, Wiebren,Buist, Ida | eng |
PMC9718750 | 1
Vol.:(0123456789)
Scientific Reports | (2022) 12:20843
| https://doi.org/10.1038/s41598-022-25253-8
www.nature.com/scientificreports
Premorbid beta blockade in sepsis
is associated with a lower risk
of a lactate concentration
above the lactate threshold,
a retrospective cohort study
Liam Schneider 1, Debra Chalmers 1, Sean O’Beirn 1, Miles Greenberg 2 & Grant Cave 1*
Sepsis and septic shock represent a significant worldwide mortality burden. A lactate greater than
4 mmol/L is associated with increased mortality in septic patients. This is the concentration at the
“lactate threshold” where serum lactate concentrations rise markedly with increased workload in
exercise. Hyperlactatemia in both sepsis and exercise is contributed to by adrenergic agonism which
stimulates aerobic glycolysis, increasing lactate production and decreasing lactate clearance. Our
hypothesis is that in patients with sepsis, treatment with beta blockers in the community will be
associated with a lower probability of initial lactate ≥ 4 mmol/L. This was single centre retrospective
cohort study. We used an in-house SQL Database for all admissions to ICU/HDU for the 2017–2020
calendar years. The dataset was filtered for an APACHE III Diagnosis of sepsis. T-tests were used for
continuous data, Chi squared and Fisher’s exact test were used as appropriate to compare proportions.
Logistic regression was used to investigate covariate effects. Of the 160 patient records analysed, 49
were prescribed beta blockers. A greater proportion of patients not prescribed beta blockers in the
community had a first lactate ≥ 4 mmol/L (p = 0.049). This was robust to regression analysis. There was
no difference in the proportion of patients with lactate ≥ 2 mmol/L (p = 0.52). In our cohort patients
previously prescribed beta blockers were less likely to have a lactate of ≥ 4 mmol/mL. This supports
the proposed mechanism that treatment with beta blockers increases the lactate threshold in sepsis.
Further study is warranted.
Sepsis is a major global health issue accounting for 20% of global mortality and 11 million deaths in 20171. Sepsis
represents a significant proportion of intensive care unit (ICU) caseload2. Hyperlactatemia in sepsis and septic
shock correlates with the severity of sepsis and associated mortality3. Reduction in serial lactate concentrations is
associated with improved outcomes4,5, while goal directed therapy targeting lactate clearance is part of the 2021
surviving sepsis guidelines6. Given the position serum lactate holds in the assessment and management of sepsis,
any premorbid factors which impact on lactate concentrations are of potential interest to the treating clinician.
Lactate kinetics in health and the lactate threshold.
Lactate is produced from pyruvate during gly-
colysis. This reaction regenerates NAD+ for the production of adenosine triphosphate (ATP)7. Lactate is metab-
olised via gluconeogenesis in liver and kidney and oxidation in skeletal muscles8–10. At rest, lactate clearance is
evenly distributed between these two mechanisms; during moderate to high exercise 60–80% of lactate clearance
occurs in skeletal muscle11. The lactate threshold refers to the level of exercise intensity at which serum lactate
accumulation rapidly increases12. This has been attributed to anaerobic metabolism and/or to an imbalance
between lactate production and lactate clearance in the absence of tissue hypoxia13,14. The lactate threshold can
be taken as 4 mmol/L with reasonable accuracy15. This threshold in health has a correlate in critical illness, with
a marked increased mortality in sepsis when initial lactate is greater than 4 mmol/L16. In healthy subjects exercis-
ing at the lactate threshold, the amount of lactate metabolised is reduced compared to moderate exercise17. This
suggests that part of the marked increase in lactate with increased exercise intensity above the lactate threshold
OPEN
1Hawkes Bay Hospital Intensive Care Unit, Hastings, New Zealand. 2University of Notre Dame, Freemantle,
Australia. *email: [email protected]
2
Vol:.(1234567890)
Scientific Reports | (2022) 12:20843 |
https://doi.org/10.1038/s41598-022-25253-8
www.nature.com/scientificreports/
is due to decreased metabolic clearance of lactate, an effect which may be mediated by beta agonism. Adrenergic
stimulation of phosphofructokinase and Na/K ATPase result in increased conversion of glucose to pyruvate
and cytosolic ATP to ADP, respectively. This increases production of lactate from pyruvate and ADP via lactate
dehydrogenase in the cellular cytosol, resulting in decreased uptake of lactate by non-exercising tissues under
adrenergic stimulation. This proposed mechanism is illustrated below in Fig. 1.
This physiology in health could also occur in sepsis—an adrenergically mediated decrease in cellular metabo-
lism of lactate could hypothetically account for part of the hyperlactataemia in sepsis. Any such effect (and the
effect of its blockade) may be more marked near the lactate threshold of 4 mmol/L.
Hyperlactatemia in sepsis.
Hypoperfusion and the resultant tissue hypoxia is traditionally held to explain
the hyperlactatemia seen in sepsis7,18,19. Studies measuring partial pressures of oxygen in septic patients have
however not demonstrated tissue hypoxia20–24, which has led to the consideration of other mechanisms7.
Activation of beta-2 adrenergic receptors, stimulated as part of a stress response is one such explanation7,25.
This pathway has been experimentally blocked at various points, with a resultant reduction in lactate22,26,27. Esmo-
lol infusion in septic patients was found to reduce lactate, and in beta blocker overdose a lower than expected
lactate concentration is seen for the degree of haemodynamic compromise28,29.
Other mechanisms have been proposed. Impaired oxygen utilisation rather than inadequate oxygen delivery
(DO2–VO2 mismatch) has been suggested, though there is little research correlating DO2–VO2 mismatch to lac-
tate levels30–33. Gattinoni et al. hypothesised a combination of tissue hypoxia and inadequate oxygen utilisation
as an explanation, finding that high lactate correlated in septic patients with either the highest or lowest central
venous oxygen saturation (ScVO2)34. Other proposed mechanisms are the Warburg effect in immune activation
and microcirculatory dysfunction35–37. It has been suggested that many or all of these suggested mechanisms
play a role in hyperlactatemia in sepsis38.
Premorbid beta blockade and lactate levels in sepsis.
Five observational studies have assessed the
effect of pre-morbid beta-blockade on lactate levels in patients presenting with sepsis39–43. Three trials found a
significant reduction in lactate levels with beta-blockade39,41,43, while two showed no difference40,42. A recent
meta-analysis of these studeis found lactate levels to be lower in patients on beta-blockers44.
If the effect of beta blockade is mediated by an alteration of the lactate threshold, the effects on serum lactate
would be expected to be greater in patient cohorts with higher lactates. Figure 2 represents data from these five
studies.
Figure 1. The proposed effect of beta adrenergic stimulation on lactate metabolism.
3
Vol.:(0123456789)
Scientific Reports | (2022) 12:20843 |
https://doi.org/10.1038/s41598-022-25253-8
www.nature.com/scientificreports/
All these trials have limitations—the majority of studies are retrospective observational in design; only one
study was multicenter; two trials looked at lactate concentrations as a secondary outcome; inclusion criteria and
sepsis definitions were variable between studies as was the timing of lactate measurement39–43. However, the pat-
tern of data from the studies supports the theory that the effect of beta blockers on lactate is more pronounced
in populations with higher lactate levels.
Effect of premorbid beta-blockers on mortality.
Pre morbid beta blockers may confer a mortality
benefit in patients presenting with sepsis40,43–47, however the evidence is not homogenous48,49. The use of ultra-
short acting beta blockers infusions in patient admitted with sepsis has shown a mortality benefit28,50–56. A meta-
analysis including ten studies found premorbid beta blockade was associated with lower short term mortality
in patients admitted with sepsis44. The proposed mechanisms of a mortality benefit include direct and indirect
cardio protective effects; enhanced microvascular circulation due to reduction in coagulopathy and indirect
immune modulatory effects42,43,47,48. In this regard evidence of a potentially beneficial effect on lactate metabo-
lism of beta blockade would be of interest.
Aim.
To assess whether previous beta blocker prescription affected the probability that the first lactate in
patients admitted from the Emergency Department to our intensive care unit with sepsis was above the lactate
threshold.
Hypothesis.
The hypothesis was that in patients admitted with sepsis, treatment with beta blockers in the
community will be associated with a lower probability of a lactate ≥ 4 mmol/L.
Methods
Setting.
This was a retrospective cohort study conducted in the intensive care unit (ICU) of the Hawkes
Bay Fallen Soldiers Memorial Hospital in Hastings, New Zealand. The hospital has 364 beds and approximately
1000 ICU and High Dependency Unit (HDU) admissions a year. The unit can provide mechanical ventilation
and continuous renal replacement therapy, and cares for both adult and paediatric patients with medical and
surgical conditions. Approval for the audit was granted by the Hawkes Bay DHB audit registration committee
and the Northern B Health and Disability Ethics Committee of New Zealand. As only de-identified date was
used a consent waiver was given by both committees. All research was performed in accordance with relevant
guidelines/regulations. We used an In-house SQL Database that tracked all admissions to ICU/HDU for the
2017–2020 calendar years, which also allows collection of the ANZICS CORE Dataset. This was then filtered
by a diagnosis of sepsis. A keyword search of free-text fields that supplemented the APACHE III Diagnosis that
contained terms such as “Sepsis”, or “Septic Shock”, was carried out to identify those patients admitted with a
co-diagnosis of sepsis. The data was then reviewed, and all duplicate or non-sepsis admissions were removed.
Initial serum lactate level at our centre was measured on an ABL 800 Flex blood gas analyser (Radiometer
Medical ApS, Bronshoj, Denmark). Serum lactate was defined as the first lactate measured during an Emergency
Department (ED) presentation. Only patients admitted directly from the ED were included. Serum lactate,
current medications, presenting vital signs, illness severity scores, laboratory data and mortality outcome were
extracted from patients’ electronic medical record and the unit’s clinical database.
Inclusion.
A single investigator (GC) blinded to beta blocker treatment status evaluated the electronic medi-
cal record to assess whether the clinical or microbiologic picture was consistent with infection.
Exclusion.
A single investigator (LS) reviewed the Emergency Department electronic medical record was
reviewed and a qSOFA score calculated for each patient prior to evaluation of beta blocker status. Patients with
0
0.5
1
1.5
2
2.5
3
0
1
2
3
4
5
6
Difference in latate between BB and non BB
Average lactate in trial
Average lactate in trial vs Mean difference
BB/non BB (95% CI)
Figure 2. Average lactate in individual trials vs. difference in mean lactate for those prescribed and not
previously prescribed beta blockers.
4
Vol:.(1234567890)
Scientific Reports | (2022) 12:20843 |
https://doi.org/10.1038/s41598-022-25253-8
www.nature.com/scientificreports/
qSOFA < 2 were excluded from analysis as patients with a qSOFA score < 2 are identified as low risk of sepsis57.
Patients who did not have lactate measured were also excluded from analysis.
Calculation of sample size.
We used unpublished data from previous work where 25% fewer patients
who were prescribed beta blockers in the community had an initial lactate > 4 mmol/L when compared with
those not prescribed beta blockers (20% vs 45%). The study was powered under the assumptions that there
would be the same proportion of beta blocked patients in the population and the proportions of patients with
lactate > 4 mmol/L would be the same as in our previous work41. Under these assumptions, our study was > 80%
powered at and alpha of 0.05 with 180 patient records included in the analysis. We anticipated that abstraction
of four calendar years of data would provide these patient numbers.
Statistical analysis.
Our data were analysed using the Graphpad Prism version 9. Continuous data are
presented as means with 95% confidence intervals. Proportions are presented as percentages with 95% confi-
dence intervals. Students T-test was used for continuous data, Chi squared and Fisher’s exact test as appropriate
to compare proportions. Logistic regression was used to investigate for significance and magnitude of covariate
effects. Criterion for covariate entry into multiple regression modelling were a statistically significant difference
in distribution of the variable between groups, an effect demonstrated in previously published work or a plausi-
ble covariate effect and a p value < 0.1 on univariate regression.
Ethics approval and consent to participate.
Approval for the use of de-identified data from this study
was given by the Hawkes Bay hospital clinical audit committee and the Northern B Health and Disability Ethics
Committee of New Zealand.
Results
293 patient records were identified for audit. Of these, 129 were excluded for a qSOFA < 2 and a further 3 were
excluded as the clinical situation and/or microbiology did not fit with the diagnosis of sepsis. One patient record
of the remaining 161 did not have their lactate measured and was excluded from analysis.
Baseline characteristics.
The baseline characteristics for the 160 patients included in the analysis are pre-
sented in Table 1. At baseline the beta blocker group was older and had lower lactate than those not exposed to
beta blockers. The site of sepsis by group is shown in Table 2. Fewer beta blocker patients had a respiratory source
of sepsis. There was no statistically significant difference in mean lactate between those with a respiratory source
of sepsis and those with other sites (mean difference 0.48, 95% CI − 1.04 to 2.0 mmol/L) nor between those with
APACHE classified chronic cardiovascular disease (mean difference 0.08, 95% CI − 1.75 to 1.91 mmol/L).
Primary endpoint.
A greater proportion of patients not prescribed beta blockers in the community had a
first lactate ≥ 4 mmol/L (48% of patients no beta blockers vs 29% prescribed beta blockers, p = 0.049). There was
no significant difference in the proportion of patients with lactate ≥ 2 mmol/L (79% of non-beta blocker patients
versus 83% of beta blocker patients, p = 0.52). These results are displayed graphically in Fig. 3.
Covariates and regression analysis.
Of the covariates in table 3 only APACHE 3 score was significantly
correlated with lactate. Logistic regression was undertaken to evaluate the effect of covariates as per the analysis
Table 1. Baseline characteristics.
Beta blocker
Non-beta blocker
p value for difference
Number (n)
49
111
Male (%)
32 (65)
64 (58)
0.36
Age (years)
71
63
< 0.01
First lactate (mmol/L)
3.54
4.46
0.04
Lowest systolic blood pressure in ED (mmHg)
90
91
0.64
Lowest HR first 24 h in ICU
75.6
78
0.42
SaO2 (%)
92
90
0.74
Highest Creatinine first 24 h (mmol/L)
173
187
0.55
Lowest Haematocrit first 24 h
0.32
0.33
0.34
APACHE III score
75
70
0.31
Number qSOFA score 3 n(%)
7 (14)
16 (14)
0.98
Vasopressors used in ED, n (%)
24 (49)
63 (57)
0.36
Prescribed metformin, n (%)
14 (29)
22 (19)
0.22
Chronic cardiovascular disease (APACHE) n (%)
6 (12)
5 (5)
0.07
Chronic respiratory disease (APACHE) n (%)
1(2)
7(6)
0.67
Mortality, n (%)
8 (16)
17 (15)
0.87
5
Vol.:(0123456789)
Scientific Reports | (2022) 12:20843 |
https://doi.org/10.1038/s41598-022-25253-8
www.nature.com/scientificreports/
plan in the methods. The odds ratio for beta blocker prescription for first lactate being ≥ 4 mmol/L regressed for
the covariates age, APACHE 3, metformin prescription, site of sepsis being respiratory and lowest haemotocrit
(Hct Lo) in the first 24 h of ICU admission was 0.31 (0.13–0.71). Of note the upper bound of the 95% CI for the
odds ratio was < 1. This finding was robust to regression with covariates individually, the removal from the model
of age (which exhibited a linear correlation with Apache 3), and addition of lowest systolic blood pressure in the
Emergency Department to the model.
Table 2. Site of sepsis by group.
Site of sepsis
Beta blocker, n (%)
Non-beta blocker, n (%)
p value
Respiratory
7(14)
32 (29)
0.05
Skin/soft tissue/joint
16 (33)
26 (23)
0.22
Genitourinary
12 (25)
18 (16)
0.22
Unknown
5 (10)
24 (22)
0.08
Biliary
5 (10)
3 (3)
n/s
Intra-abdominal
3 (6)
3 (3)
n/s
Cardiac
0 (0)
2 (2)
n/s
Vascular catheter
1 (2)
0 (0)
n/s
CNS
0 (0)
1 (1)
n/s
Figure 3. Percentages of those prescribed and not prescribed beta blockers with lactate greater than 2 and
4 mmol/L at presentation.
Table 3. Covariate analysis.
Covariate
95% CI odds ratio, p value
Male sex
0.8–1.75 to 0.21, p 0.82
Age (years)
0.98–1.01, p 0.99
Lowest recorded SpO2 in ED (%)
0.96–1.07, p 0.36
Lowest haematocrit first 24 h
0.99 to 1.1, p 0.053
APACHE III score
1.013 to1.04, p < 0.001
Vasopressors used in ED
0.39 to 1.4, p 0.44
SBP in ED
0.96–1, p 0.11
Prescribed metformin
0.8–3.8, p 0.31
Site of sepsis respiratory
0.31–1.4, p = 0.6
Cardiovascular disease (APACHE)
0.35–4.4, p = 0.89
6
Vol:.(1234567890)
Scientific Reports | (2022) 12:20843 |
https://doi.org/10.1038/s41598-022-25253-8
www.nature.com/scientificreports/
Discussion
In this patient cohort pre-morbid beta blocker treatment was associated with a lower initial lactate, driven by a
reduction in the proportion of patients with a lactate of ≥ 4 mmol/L. This effect was robust to regression analysis.
There was no significant difference in the proportion of beta blocker/non beta blocker exposed patients with
lactate ≥ 2 mmol/L. Extension from our findings holds obvious caveats in that our methodology permits identi-
fication of association only, and lactate threshold is a concept from exercise physiology that has not been proven
to have an effect in the clinical context. Nonetheless, these findings provide inferential support for premorbid
beta blockade reducing serum lactate in sepsis by increasing the proportion of patients below the concentration
where lactate production and metabolism uncouple in response to metabolic stress. Restated, our findings offer
support for the view that beta blockers increase the lactate threshold in sepsis.
This data fits with the pattern observed the previous five studies as demonstrated in Fig. 2, that the effect
of premorbid beta-blockers on initial lactate was most significant in patient populations with higher mean
lactate concentrations. The papers which found no effect analysed cohorts with an average initial lactate
of ≤ 2 mmol/L40,42; we did not identify any association for premorbid beta blockade with the probability of lac-
tate being ≥ 2 mmol/L. The three other studies which demonstrated an effect of beta blockade when looking at
cohorts with higher average lactates39,41,43. Our proposed mechanism to explain the findings seen in our study,
while hypothetical, offers a unifying explanation for the current heterogeneity of evidence in this area.
Our study has several limitations. The most significant is that seeking an association based on a hypothesized
mechanism can establish the association while the mechanism remains hypothetical. While there is a mark-
edly increased mortality with lactate ≥ 4 mmol/L in sepsis the lactate threshold is a concept proven in exercise
physiology rather than established in clinical medicine. Our study design is a retrospective observational and as
such can only demonstrate an association rather than prove causation for the effects of premorbid beta blockade
on serum lactate concentrations. Additionally, our study is single centre creating limits on external validity.
The mean initial lactate was 3.54 mmol/L for the beta blocked patients compared to 4.46 mmol/L in those not
premorbidly prescribed beta blockade. Both concentrations are above a threshold which would trigger clini-
cal action—viewed from this perspective the clinical significance of the concentration difference is uncertain.
This study is not powered to assess whether statistically significant difference in lactates concentration affected
clinical outcomes. A qSOFA score of ≥ 2 was used as part of the inclusion criteria. The qSOFA score has been
found to have a decreased predictive value in ICU compared to SOFA score but a better predictive value outside
ICU57. A flaw with the qSOFA is the potential for inter-user variability in the recording of scores particularly in
the altered mental status variable. In addition, some of the cohort was excluded due to incomplete data from the
Emergency Department admission. As with other studies, it is only possible to ascertain whether patients were
prescribed beta blockers at the time of to their admission, the actual compliance in the cohort is unknown. The
reason for beta blocker prescription is not available for this patient cohort and as such the role of any underlying
cardiac dysfunction is difficult to quantify. There may also be an effect of other unmeasured variables such as
amount of fluid resuscitation prior to initial lactate measurement. We abstracted patient data in blocks of calendar
years and anticipated 4 years would provide 180 records for study. The 5% reduction in power from analysis of
160 subjects from this period did not result in a type I error—our results were positive. Underpowered studies
which return positive results additionally tend to overestimate the magnitude of effect—a lower proportion of
underpowered studies are expected to be positive with the tendency to exhibit more extreme results. This was
again not the case in our work as the overall difference in proportion of patients with lactate ≥ 4 in was lower
than that powered for (19% vs 25%).
Further research is recommended into the effect of beta blockade on the lactate threshold and its significance.
Our group has commenced bench top mechanistic work in a prospective study of the effect esmolol infusions
on lactate in animal models of sepsis aiming to further examine the proposed mechanism of the effect of beta
blockade on lactate.
Conclusion
In our cohort patients previously prescribed beta blockers presenting with sepsis were less likely to have a lactate
of ≥ 4 mmol/ml. This is in keeping with the pattern of results seen in the current literature and supports the pro-
posed mechanism that treatment with beta blockers increases the lactate threshold. Further study is warranted
as such a mechanism could have clinical significance.
Data availability
The datasets used and/or analysed during the current study are available from the corresponding author on
reasonable request.
Received: 13 June 2022; Accepted: 28 November 2022
References
1. World Health Organization. Global report on the epidemiology and burden of sepsis: Current evidence, identifying gaps and
future directions. World Health Organization. (2020) [cited 2022 Jun 5]. https:// apps. who. int/ iris/ handle/ 10665/ 334216.
2. Fleischmann-Struzek, C. et al. Incidence and mortality of hospital- and ICU-treated sepsis: Results from an updated and expanded
systematic review and meta-analysis. Intensive Care Med. 46(8), 1552–1562 (2020).
3. Shapiro, N. I. et al. Serum lactate as a predictor of mortality in emergency department patients with infection. Ann. Emerg. Med.
45(5), 524–528 (2005).
4. Nichol, A. et al. Dynamic lactate indices as predictors of outcome in critically ill patients. Crit. Care 15(5), R242 (2011).
5. Jansen, T. C. et al. Early lactate-guided therapy in intensive care unit patients. Am. J. Respir. Crit. Care Med. 182(6), 752–761 (2010).
7
Vol.:(0123456789)
Scientific Reports | (2022) 12:20843 |
https://doi.org/10.1038/s41598-022-25253-8
www.nature.com/scientificreports/
6. Evans, L. et al. Surviving sepsis campaign: International guidelines for management of sepsis and septic shock 2021. Crit. Care
Med. 49(11), e1063–e1143 (2021).
7. Garcia-Alvarez, M., Marik, P. & Bellomo, R. Sepsis-associated hyperlactatemia. Crit. Care 18(5), 503 (2014).
8. Consoli, A., Nurjhan, N., Reilly, J. J., Bier, D. M. & Gerich, J. E. Contribution of liver and skeletal muscle to alanine and lactate
metabolism in humans. Am. J. Physiol. 259(5 Pt 1), E677-684 (1990).
9. Bergman, B. C. et al. Active muscle and whole body lactate kinetics after endurance training in men. J. Appl. Physiol. (1985) 87(5),
1684–1696 (1999).
10. Bergman, B. C. et al. Endurance training increases gluconeogenesis during rest and exercise in men. Am. J. Physiol. Endocrinol.
Metab. 278(2), E244–E251 (2000).
11. Mazzeo, R. S., Brooks, G. A., Schoeller, D. A. & Budinger, T. F. Disposal of blood [1–13C]lactate in humans during rest and exercise.
J. Appl. Physiol. (1985) 60(1), 232–241 (1986).
12. Ga, B. Anaerobic threshold: Review of the concept and directions for future research. Med. Sci. Sports Exerc. (1985) [cited 2022
Mar 21]. 17(1). https:// pubmed. ncbi. nlm. nih. gov/ 38849 59/.
13. Wasserman, K. & Koike, A. Is the anaerobic threshold truly anaerobic?. Chest 101(5 Suppl), 211S-218S (1992).
14. Brooks, G. A. The lactate shuttle during exercise and recovery. Med. Sci. Sports Exerc. 18(3), 360–368 (1986).
15. Heuberger, J. A. A. C. et al. Repeatability and predictive value of lactate threshold concepts in endurance sports. PLoS ONE 13(11),
e0206846 (2018).
16. Bou Chebl, R. et al. Serum lactate is an independent predictor of hospital mortality in critically ill patients in the emergency
department: A retrospective study. Scand. J. Trauma Resusc. Emerg. Med. 14(25), 69 (2017).
17. Messonnier, L. A. et al. Lactate kinetics at the lactate threshold in trained and untrained men. J. Appl. Physiol. (1985) 114(11),
1593–1602 (2013).
18. Dellinger, R. P. et al. Surviving sepsis campaign: International, 2012. Intensive Care Med. 39(2), 165–228 (2013).
19. Sterling, S. A. et al. Characteristics and outcomes of patients with vasoplegic versus tissue dysoxic septic shock. Shock 40(1), 11–14
(2013).
20. Rosser, D. M., Stidwill, R. P., Jacobson, D. & Singer, M. Oxygen tension in the bladder epithelium rises in both high and low cardiac
output endotoxemic sepsis. J. Appl. Physiol. (1985) 79(6), 1878–1882 (1995).
21. VanderMeer, T. J., Wang, H. & Fink, M. P. Endotoxemia causes ileal mucosal acidosis in the absence of mucosal hypoxia in a
normodynamic porcine model of septic shock. Crit. Care Med. 23(7), 1217–1226 (1995).
22. Levy, B., Gibot, S., Franck, P., Cravoisy, A. & Bollaert, P. E. Relation between muscle Na+K+ ATPase activity and raised lactate
concentrations in septic shock: A prospective study. Lancet 365(9462), 871–875 (2005).
23. Boekstegers, P., Weidenhöfer, S., Kapsner, T. & Werdan, K. Skeletal muscle partial pressure of oxygen in patients with sepsis. Crit.
Care Med. 22(4), 640–650 (1994).
24. Sair, M., Etherington, P. J., Peter Winlove, C. & Evans, T. W. Tissue oxygenation and perfusion in patients with systemic sepsis.
Crit. Care Med. 29(7), 1343–1349 (2001).
25. Revelly, J. P. et al. Lactate and glucose metabolism in severe sepsis and cardiogenic shock. Crit. Care Med. 33(10), 2235–2240
(2005).
26. McCarter, F. D. et al. Adrenergic blockade reduces skeletal muscle glycolysis and Na(+), K(+)-ATPase activity during hemorrhage.
J. Surg. Res. 99(2), 235–244 (2001).
27. Levy, B., Desebbe, O., Montemont, C. & Gibot, S. Increased aerobic glycolysis through beta2 stimulation is a common mechanism
involved in lactate formation during shock states. Shock 30(4), 417–421 (2008).
28. Morelli, A. et al. Effect of heart rate control with esmolol on hemodynamic and clinical outcomes in patients with septic shock: A
randomized clinical trial. JAMA 310(16), 1683–1691 (2013).
29. Mégarbane, B., Deye, N., Malissin, I. & Baud, F. J. Usefulness of the serum lactate concentration for predicting mortality in acute
beta-blocker poisoning. Clin. Toxicol. (Phila). 48(10), 974–978 (2010).
30. Ronco, J. J. et al. Identification of the critical oxygen delivery for anaerobic metabolism in critically ill septic and nonseptic humans.
JAMA 270(14), 1724–1730 (1993).
31. Ronco, J. J. et al. Oxygen consumption is independent of increases in oxygen delivery by dobutamine in septic patients who have
normal or increased plasma lactate. Am. Rev. Respir. Dis. 147(1), 25–31 (1993).
32. Mira, J. P. et al. Lack of oxygen supply dependency in patients with severe sepsis. A study of oxygen delivery increased by military
antishock trouser and dobutamine. Chest 106(5), 1524–1531 (1994).
33. Astiz, M. E., Rackow, E. C., Kaufman, B., Falk, J. L. & Weil, M. H. Relationship of oxygen delivery and mixed venous oxygenation
to lactic acidosis in patients with sepsis and acute myocardial infarction. Crit. Care Med. 16(7), 655–658 (1988).
34. Gattinoni, L. et al. Understanding lactatemia in human sepsis. Potential impact for early management. Am. J. Respir. Crit. Care
Med. 200(5), 582–589 (2019).
35. Cheng, S. C., Joosten, L. A. B. & Netea, M. G. The interplay between central metabolism and innate immune responses. Cytokine
Growth Factor Rev. 25(6), 707–713 (2014).
36. Charlton, M., Sims, M., Coats, T. & Thompson, J. P. The microcirculation and its measurement in sepsis. J. Intensive Care Soc.
18(3), 221–227 (2017).
37. Ince, C. The microcirculation is the motor of sepsis. Crit. Care. 9(Suppl 4), S13–S19 (2005).
38. Gutierrez, G. & Wulf, M. E. Lactic acidosis in sepsis: A commentary. Intensive Care Med. 22(1), 6–16 (1996).
39. Contenti, J., Occelli, C., Corraze, H., Lemoël, F. & Levraut, J. Long-term β-blocker therapy decreases blood lactate concentration
in severely septic patients. Crit. Care Med. 43(12), 2616–2622 (2015).
40. Chan, J. Z. W. et al. Beta-blockers’ effect on levels of lactate in patients with suspected sepsis—The BeLLa study. Am. J. Emerg. Med.
38(12), 2574–2579 (2020).
41. Pham, D. et al. Is lactate lower in septic patients who are prescribed beta blockers? Retrospective cohort study of an intensive care
population. Emerg. Med. Australas. 33(1), 82–87 (2021).
42. Tan, K. et al. Association between premorbid beta-blocker exposure and sepsis outcomes-the beta-blockers in European and
Australian/American Septic Patients (BEAST) Study. Crit. Care Med. 49(9), 1493–1503 (2021).
43. Kuo, M. J. et al. Premorbid β1-selective (but not non-selective) β-blocker exposure reduces intensive care unit mortality among
septic patients. J. Intensive Care 9(1), 40 (2021).
44. Hasegawa, D., Sato, R., Prasitlumkum, N. & Nishida, K. Effect of premorbid beta-blockers on mortality in patients with sepsis: A
systematic review and meta-analysis. J. Intensive Care Med. 23, 8850666211052926 (2021).
45. Guz, D. et al. β-blockers, tachycardia, and survival following sepsis: An observational cohort study. Clin. Infect. Dis. 73(4), e921–
e926 (2021).
46. Macchia, A. et al. Previous prescription of β-blockers is associated with reduced mortality among patients hospitalized in intensive
care units for sepsis. Crit. Care Med. 40(10), 2768–2772 (2012).
47. Singer, K. E. et al. Outpatient beta-blockers and survival from sepsis: Results from a national cohort of Medicare beneficiaries.
Am. J. Surg. 214(4), 577–582 (2017).
48. Arnautovic, J., Mazhar, A., Souther, B., Mikhijan, G., Boura, J., Huda, N. Cardiovascular factors associated with septic shock
mortality risks. Spartan Med. Res. J. [cited 2022 Feb 22]; 3(1). https:// www. ncbi. nlm. nih. gov/ pmc/ artic les/ PMC77 46094/.
8
Vol:.(1234567890)
Scientific Reports | (2022) 12:20843 |
https://doi.org/10.1038/s41598-022-25253-8
www.nature.com/scientificreports/
49. DeMott, J. M., Patel, G. & Lat, I. Effects of chronic antihypertensives on vasopressor dosing in septic shock. Ann. Pharmacother.
52(1), 40–47 (2018).
50. Hasegawa, D. et al. Effect of ultrashort-acting β-blockers on mortality in patients with sepsis with persistent tachycardia despite
initial resuscitation: A systematic review and meta-analysis of randomized controlled trials. Chest 159(6), 2289–2300 (2021).
51. Chacko, C. J. & Gopal, S. Systematic review of use of β-blockers in sepsis. J. Anaesthesiol. Clin. Pharmacol. 31(4), 460–465 (2015).
52. Liu, H. et al. Effect of esmolol in septic shock patients with tachycardia: A randomized clinical trial. Zhonghua Yi Xue Za Zhi
99(17), 1317–1322 (2019).
53. Wang, Z., Wu, Q., Nie, X., Guo, J. & Yang, C. Combination therapy with milrinone and esmolol for heart protection in patients
with severe sepsis: A prospective, randomized trial. Clin. Drug Investig. 35(11), 707–716 (2015).
54. Wang, S. et al. Effect of esmolol on hemodynamics and clinical outcomes in patients with septic shock. Zhonghua Wei Zhong Bing
Ji Jiu Yi Xue. 29(5), 390–395 (2017).
55. Xinqiang, L. et al. Esmolol improves clinical outcome and tissue oxygen metabolism in patients with septic shock through control-
ling heart rate. Zhonghua Wei Zhong Bing Ji Jiu Yi Xue. 27(9), 759–763 (2015).
56. Kakihana, Y. et al. Efficacy and safety of landiolol, an ultra-short-acting β1-selective antagonist, for treatment of sepsis-related
tachyarrhythmia (J-Land 3S): A multicentre, open-label, randomised controlled trial. Lancet Respir. Med. 8(9), 863–872 (2020).
57. Seymour, C. W. et al. Assessment of clinical criteria for sepsis: For the third international consensus definitions for sepsis and
septic shock (sepsis-3). JAMA 315(8), 762–774 (2016).
Acknowledgements
The authors would like to acknowledge Penny Park, ICU research nurse at the Hawkes Bay hospital for her
assistance with data extraction.
Author contributions
G.C. and M.G. were responsible for study conception. G.C. was responsible for analysis. L.S. and S.O. were
responsible for data extraction. L.S. authored the first draft of the manuscript. All authors subsequently con-
tributed to revisions. No individual identifying patient data is included in this manuscript. The requirement for
informed consent was waived by the Hawkes Bay hospital clinical audit committee and the Northen B Health
and Disability Ethics Committee of New Zealand.
Funding
No funding was received to undertake this study.
Competing interests
The authors declare no competing interests.
Additional information
Correspondence and requests for materials should be addressed to G.C.
Reprints and permissions information is available at www.nature.com/reprints.
Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and
institutional affiliations.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International
License, which permits use, sharing, adaptation, distribution and reproduction in any medium or
format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the
Creative Commons licence, and indicate if changes were made. The images or other third party material in this
article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the
material. If material is not included in the article’s Creative Commons licence and your intended use is not
permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from
the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/.
© The Author(s) 2022
| Premorbid beta blockade in sepsis is associated with a lower risk of a lactate concentration above the lactate threshold, a retrospective cohort study. | 12-02-2022 | Schneider, Liam,Chalmers, Debra,O'Beirn, Sean,Greenberg, Miles,Cave, Grant | eng |
PMC6747064 | International Journal of
Environmental Research
and Public Health
Article
Acute Effects of a Speed Training Program on
Sprinting Step Kinematics and Performance
Krzysztof Mackala 1,*, Marek Fostiak 2, Brian Schweyen 3, Tadeusz Osik 4 and Milan Coch 5
1
Department of Track and Field University School of Physical Education in Wroclaw, Poland, Ul.
Paderewskiego 35, 51-612 Wrocław, Poland
2
Department of Track and Field, Gdansk University of Physical Education and Sport, ul. Kazimierza
Górskiego 1, 80-336 Gdansk, Poland
3
PZLA (Polish Track and Field Association), Mysłowicka 4, 01-612 Warszawa, Poland
4
Athletics Department, University of Montana, Adams Center 32 Campus Drive, Missoula, MT 59812, USA
5
Faculty of Sport, University of Ljubljana, Gortanova ulica 22, 1000 Ljubljana, Slovenia
*
Correspondence: [email protected]; Tel.: +48-605-272-433
Received: 4 July 2019; Accepted: 24 August 2019; Published: 28 August 2019
Abstract: The purpose of this study was to examine the effects of speed training on sprint step
kinematics and performance in male sprinters.
Two groups of seven elite (best 100-m time:
10.37 ± 0.04 s) and seven sub-elite (best 100-m time: 10.71 ± 0.15 s) sprinters were recruited.
Sprint performance was assessed in the 20 m (flying start), 40 m (standing start), and 60 m (starting
block start). Step kinematics were extracted from the first nine running steps of the 20-m sprint
using the Opto-Jump–Microgate system. Explosive power was quantified by performing the CMJ,
standing long jump, standing triple jump, and standing five jumps. Significant post-test improvements
(p < 0.05) were observed in both groups of sprinters. Performance improved by 0.11 s (elite) and
0.06 s (sub-elite) in the 20-m flying start and by 0.06 s (elite) and 0.08 s (sub-elite) in the 60-m start
block start. Strong post-test correlations were observed between 60-m block start performance and
standing five jumps (SFJ) in the elite group and between 20-m flying start and 40-m standing start
performance and standing long jump (SLJ) and standing triple jump (STJ) in the sub-elite group.
Speed training (ST) shows potential in the reduction of step variability and as an effective short-term
intervention program in the improvement of sprint performance.
Keywords: sprinting; speed; sprint exercises; step variability; kinematics
1. Introduction
Past investigations have indicated a significant commonality between sprinting performance
(across distances from 20 to 100 m) and explosive power [1–5]. As the largest inhibitor in the sprinting
movement is gravity, sprinters must produce large vertical ground reaction force during step take-off
to achieve maximal velocity [6]. This is achieved via application of plyometric training as it uses
jump-based movements to train and improve explosive power. However, this training modality does
not involve a running component.
A mixture of plyometric characters exercises (skips and bounds) and some speed running drills
created an alternative sprint-specific running methodology, which represents a high-intensity speed
training (ST). ST is treated as a low-volume training strategy and credited with producing significant
gains in maximal running velocity [7]. This form of training is based on short bouts of all-out sprints
separated by rest periods from 90 s to 5–6 min to enhance post-exercise recovery and avoid fatigue,
including central nervous system fatigue [8–10]. Similar in one regard to plyometrics, sprint-specific
running also involves a rapid eccentric movement followed by a short amortization phase that is
Int. J. Environ. Res. Public Health 2019, 16, 3138; doi:10.3390/ijerph16173138
www.mdpi.com/journal/ijerph
Int. J. Environ. Res. Public Health 2019, 16, 3138
2 of 13
then followed by an explosive concentric movement [6]. This enables the synergistic muscles to
engage the myotatic-stretch reflex during the stretch-shortening cycle [11]. Such a training stimulus
activates the elastic properties of the muscle fibers and connective tissue, allowing the muscles to
store more elastic energy [11–13]. This then enhances the release of the accumulated mechanical
energy during step take-off, a process which is easily visible during skipping and speed bounding.
Hence, improvements in elastic energy storage via ST could be interrelated with improvements in
single sprint step performance and, therefore, maximal velocity.
Running velocity is a product of step frequency (SF) and step length (SL) [3,14,15] in which
a step defines half a running cycle or the interval between ground contact of one foot to the moment
when contact is made by the opposing foot [16]. SF and SL are mutually interdependent in which
an increase in SL results in a decrease in SF and vice versa [17–21]. Both variables are determined
by anthropometric characteristics, movement regulation processes, motor abilities, and energetic
processes [16,22]. Hence, the relationship between SF and SL is unique for each sprinter [19,23].
Previous research investigating the interaction between SL and SF in running has generally been
inconclusive. There is much debate as to whether sprinters achieve greater benefits from increased SL
or SF. Nonetheless, the strong interdependency between SL and SF and performance at sub-maximal
and maximal velocities posits that these are important outcome measures when assessing variability in
sprinting technique [3,24]. In addition, little is known on how other kinematic variables such as flight
phase time, ground contact time, and step velocity interact and influence maximal running velocity.
The fundamental goal of this study was to investigate the potential effects of acute ST on sprint
steps kinematics—how these spatial-temporal characteristics can influence sprint running. A secondary
goal was to examine performance including maximal running velocity and power of lower extremity.
Sprinters of different competitive levels were recruited to better generalize the results of the study.
To mitigate the effects of overtraining, only eight sessions were administered in two 10- and 12-day
modules separated with 10 days of recovery. Furthermore, the intervention was administered during
the pre-season period to avoid additional confounding variables. A pre-test/post-test design was
adopted and included a wide battery of tests and exercises commonly applied in sprint training.
Based on this consideration, it was hypothesized that the information of selecting the relevant
biomechanical parameters obtained after high-speed training, may indicate the elite of the sprinters
(seniors and 100 m < 10.50 s) will show less variability of the sprinting step in the 20-m sprint from
a flying start than the sub-elite sprinters (juniors and 100 m > 10.50 s).
2. Materials and Methods
2.1. Participants
Seven elite and seven sub-elite healthy male sprinters with a minimum of four to five years of
regular sprint training were recruited. Sprinters participating in the experiment were members of
the national team. For the purposes of the experiment, they were divided into two groups: Elite and
sub-elite. The main division criterion was age (juniors up to 19 years, seniors above) and the personal
best times in the 100 m sprint (below 10.50 s—elite). Each participant was medically cleared to
participate in sprint training and presented no orthopedic or physiological limitation or injury that
could affect sprint performance. All participants had previous experience (minimum six months)
in sprint-related strength, speed, and plyometrics training before enrollment. Written consent was
obtained after the protocol and procedures were explained in full. Parental consent was also obtained
from those individuals under 18 years of age. Sprinters were instructed to maintain their normal intake
of food and fluids during the study period. Additionally, they were instructed to avoid any strenuous
physical activity 24 h prior to testing as well as refrain from eating 3 h before test commencement.
The study design was approved by the Institutional Ethics Committee.
Int. J. Environ. Res. Public Health 2019, 16, 3138
3 of 13
2.2. Design
The study was performed over a four-week period during the pre-season (April–May) just
prior to the outdoor racing season.
All participants had been engaged in a standard sprint
training regimen (five to seven days of a weekly microcycle) from October to November until
inclusion in the study. This training regimen involved general and specialized fitness exercises
designed to enhance sprinting technique, increase maximal velocity, and develop strength, power,
and endurance. Additionally, many of the participants competed during the indoor competition season
(January–February).
Eight ST sessions were executed in a 10-day and then 12-day module (Table 1). The modules
were separated by 10 days of recovery training with minimal load to promote central nervous system
regeneration. ST in the first 10-day module was executed every other day and in the 12-day module every
third day. The days with training involved the main SST workout and one short supplementary workout
(general fitness) with the order reversed in each subsequent training session. Post-workout recovery
was also provided and predominately involved massage treatment and low-intensity swimming.
Table 1. Training characteristics of the training modules.
Type of Exercise Modality
Training Module
10-day Module
12-day Module
Number of Workouts (n)
Strength (combined with a short
plyometrics session)
3
2
Plyometrics
−
1
Speed
4
4
Speed−endurance
1
2
Tempo
2
1
General fitness (supplementary session)
4
4
Recovery (swimming, massage,
cryotherapy)
5
5
Day off (rest)
−
1
Testing
−
1
Total training workouts per module
36
ST focused primarily on developing sprint technique and maximal velocity by various exercises
that consisted of various skips, bounds, accelerations, starts, and maximal intensity sprints (Table 2).
Training duration was approximately 90 min. This included a 20–30 min warm-up of jogging, stretching,
light jumping, skipping, and submaximal accelerations and a 10–15 min cooldown. Training load
was progressively amplified by varying task complexity (running distance or number of foot contacts
during skip and bound exercises) and by increasing running velocity (from 85–90% to 100%).
Int. J. Environ. Res. Public Health 2019, 16, 3138
4 of 13
Table 2. Speed training (ST) characteristics.
Exercises
Module 1
Module 2
1st
session
2nd
session
3rd
session
4th
session
5th
session
6th
session
7th
session
8th
session
Sub−maximal speed
(85–90%)
10 m skip A + 20 m
acceleration
2 rep.
3 rep.
2 rep.
2 rep.
10 m skip C + 20 m
acceleration
2 rep.
3 rep.
2 rep.
2 rep.
20 m sprint bounding + 20
m acceleration
3 rep.
3 rep.
3 rep.
4 rep.
Falling start + 20 m
build-up
3 rep.
2 rep.
3 rep.
2 rep.
Block starts + 20 m build-up
4 rep.
4 rep.
4 rep.
4 rep.
40 m acceleration
3 rep.
4 rep.
3 rep.
6 rep.
Total distance [m]
540
600
540
640
Maximal speed (ca. 100%)
Falling start + 20 m build
up
2 rep.
2 rep.
2 rep.
2 rep.
Block starts + 20 build up
3 rep.
4 rep.
3 rep.
4 rep.
30 m sprint
3 rep.
4 rep.
3 rep.
4 rep.
40 m sprint
3 rep.
4 rep.
3 rep.
4 rep.
50 m sprint
1 rep.
0 rep.
1 rep.
1 rep
Total distance [m]
360
400
360
450
2.3. Testing Protocol
Maximal running velocity and lower extremity explosive power were measured one day before
the intervention and 48 h after the last training session was completed to ascertain acute training
effects. Pre- and post-testing was executed at the same time of the day and under identical conditions.
No familiarization was provided as all of the testing protocols were well-known to the participants.
A warm-up similar to the one used during the training intervention was administered prior to testing
(light jogging, stretching, skipping drills, light jumps and bounds, and 30- to 50-m accelerations).
The test battery included the countermovement jump (CMJ), standing long jump (SLJ), standing triple
jump (STJ), and standing five jump (SFJ) to assess explosive power. After 60 min of rest, the participants
performed the 20-m sprint (from the flying start) to evaluate sprint step kinematics and then 40 m
(from a standing start) and 60 m (using starting blocks) to determine maximal running velocity.
2.4. Assessment of Lower Extremity Explosive Power
The CMJ was used to determine vertical jumping performance. The OptoJump–Microgate optical
measurement system (Optojump, Bolzano, Italy) determined contact and flight times with an accuracy
of 0.001 s. No restrictions were placed on knee angle during the eccentric phase and the participants
were instructed to perform a dynamic double arm swing to attain maximal height. Three trials were
separated by 1 min of rest with the highest jump selected for analysis.
Horizontal jump performance was measured in the following order: SLJ, STJ, and SFJ. From an
erect position with parallel feet placement, the participant executed the SLJ and was required to land
on both feet in the long jump pit without falling backwards. Jumping distance was measured to the
nearest 1.0 cm. The starting position for the STJ and SFJ was similar. In these jumps, after one or
more arm and leg swings, the participant performed the required number of forward jumps with
each step on an alternating leg and was required to land on both feet in the half-squat position on
a special jumping mattress. Three trials were executed for each jump and the longest distance to the
nearest 1.0 cm was recorded. A 2 min rest was provided between each trial and 5–6 min between each
jump modality.
The reliability of the vertical and horizontal jumping tests was measured using intraclass correlation
coefficients (ICC). A posteriori analysis obtained correlations of 0.92 for CMJ jump height and 0.93 for
Int. J. Environ. Res. Public Health 2019, 16, 3138
5 of 13
SLJ, 0.93 for STJ, and 0.90 for SFJ jump distances. The large coefficients indicate satisfactory test–retest
reliability and may be explained by the extensive familiarization of all participants with executing
these jumps.
2.5. Assessment of Sprint Performance
The 20-m flying start, 40-m standing start, and 60-m block start were performed on an indoor track
integrated with the Brower Timing TC-System (Draper, Utah, USA). The photocells were positioned on
the track at the start and finish according to the sprint distance [3]. In the 20-m flying start, the sprinter
began from a standing start and accelerated as quickly as possible to attain maximal running velocity
within a 20 m run-up. Upon reaching the 20 m mark, the sprinter continued to sprint for exactly
20 m at their maximal velocity. This sprint modality had been previously applied in research and is
considered sufficient to achieve maximal velocity [3,6,16,25,26]. Two trials were executed and separated
by 2 min of rest. The same OptoJump measuring system was used to measure the spatial-temporal
characteristics of the first nine running steps at maximal velocity including step length, step frequency,
ground contact time, flight time, and step velocity. In a track configuration, the measurement system
uses a series of interconnected rods (100 cm x 4 cm x 3 cm) fitted with optical sensors. Each rod (RX bars
and TX bars) is fitted with 32 photocells, arranged 4 cm one from another and 0.2 cm above the ground.
The rods were distributed along the length and width of the track (20 m x 1.22 m). The device was
integrated with a computer for data storage and processing. After completing the 20-m flying start
sprint, the participants performed two trials of the 40 m from a standing start and 60 m from a block
start. Rest intervals of 4 and 6 min were provided between trials, respectively. The fastest time in each
distance was selected for analysis.
2.6. Statistical Analysis
Means (x) and standard deviations (SD) were calculated for all dependent variables. Student’s t-test
was used to examine pre- and post-test differences in running velocity and jumping performance.
Fisher’s least significant difference (LSD) tests were performed post hoc to determine pairwise
differences when significant F ratios were obtained. Variability in the nine steps was quantified
by calculating the SD and confidence intervals (95%CI). The associations between the performance
variables (sprint times and jump distance/heights) were determined by Pearson product–moment
correlations. Additionally, hierarchical cluster analysis using Ward’s method was used to determine
the linkage distances among the kinematic characteristics grouped as elementary determinants of
sprint velocity. A statistical power of 0.90 was determined satisfactory and an alpha level of 0.05 was
accepted as statistically significant (denoted in bold font).
3. Results
Table 3 provides the anthropometric and personal bests in the 60 m and 100 m of the elite and
sub-elite sprinters.
Table 3. Descriptive statistics and Student’s t-test results of group age, anthropometric characteristics,
and personal best (PB) times.
Variables
Sub-Elite
Elite
t
p
x
SD
x
SD
Age (years)
18.71
0.75
24.71
2.43
−6.24
0.000043
Height (cm)
182.00
5.35
179.42
3.91
0.78
0.449165
Body mass (kg)
73.28
4.49
74.43
8.24
−0.32
0.753007
BMI (kg/m2)
22.17
1.10
22.79
0.74
−1.22
0.244110
60 m PB
6.97
0.08
6.69
0.79
6.52
0.000028
100 m PB
10.71
0.15
10.37
0.04
5.71
0.000097
Bold format: significant differences.
Int. J. Environ. Res. Public Health 2019, 16, 3138
6 of 13
Height, body mass, and BMI were similar between the groups. The differences between the elite
and sub-elite group for age and personal bests were significant (p < 0.05).
Table
4
presents
the
pre-
and
post-test
results
in
sprint
and
jump
performance.
Significant differences were observed in all variables in which jumping distances increased and
running times decreased in both the elite and sub-elite sprinters (p < 0.05). Sprint step characteristics
are presented in Figure 1. In the elite sprinters, contact time (CT) decreased post-test in steps four
to seven whereas the lowest value was attained in the ninth step. CT showed a decreasing trend
from the first to ninth step. Among the sub-elite sprinters at pre-test, a trend towards increased CT
between the first and ninth steps was observed. At post-test, increased CT was observed in steps
four to eight. The flight time (FT) in both groups increased post-test from the first to the eighth step.
Greater FT was achieved by the sub-elite sprinters at post-test. Step length (SL) in both the elite and
sub-elite sprinters showed an upward trend at pre-test, with increased SL from the first to the eighth
step, only to become more linear at post-test. Additionally, post-test SL magnitudes increased in
both groups. Step frequency (SF) showed an irregular pre-test trend in both the elite and sub-elite
sprinters. Following the intervention, SF increased in the elite sprinters, particularly in the last three
steps. Changes in step velocity (SV) were more pronounced compared with the other variables in both
groups particularly at pre-test (an increase in two consecutive steps followed by a decrease in the next
two steps).
Table 4. Student’s t test results for the dependent variables.
Variable
x
SD
x
SD
∆x
∆xSD
t
p
Confidence
−95.00%
Confidence
+95.0%
Sub-elite
Pre−test
Post−test
60m–60m_t2 (s)
7.10
0.09
7.02
0.05
0.08
0.04
4.77
0.0030
0.038
0.121
20m flying–20m
flying_2t (s)
2.21
0.08
2.13
0.05
0.07
0.05
3.57
0.0116
0.024
0.129
40m –40m_2t (s)
4.37
0.04
4.32
0.03
0.06
0.01
7.94
0.0000
0.040
0.076
SLJ–SLJ_2t (cm)
2.91
0.06
2.99
0.07
−0.08
0.04
−4.81
0.0029
−0.120
−0.039
STJ–STJ_2t (m)
8.56
0.16
8.80
0.18
−0.24
0.07
−8.67
0.0001
−.0313
−0.175
SFJ–SFJ_2t (m)
14.90
0.62
15.56
0.53
−0.66
0.13
−13.66
0.0000
−0.775
−0.539
CMJ –CMJ_2t
(cm)
76.43
4.89
82.71
5.34
−6.29
1.70
−9.76
0.0000
−7.862
−4.709
Elite
Pre−test
Post−test
60m–60m_t2 (s)
6.79
0.08
6.72
0.08
0.06
0.02
7.17
0.0003
0.040
0.082
20m flying–20m
flying_2t (s)
2.07
0.04
1.97
0.07
0.11
0.07
4.17
0.0058
0.047
0.179
40m –40m_2t (s)
4.12
0.02
4.08
0.01
0.04
0.02
4.58
0.0037
0.018
0.061
SLJ –SLJ_2t (m)
3.15
0.10
3.23
0.11
−0.07
0.05
−3.92
0.0078
−0.118
−0.027
STJ–STJ_2t (m)
9.39
0.52
9.89
0.48
−0.49
0.21
−6.28
0.0007
−0.693
−0.304
SFJ–SFJ_2t (m)
15.81
0.44
16.59
0.57
−0.78
0.50
−4.11
0.0062
−1.244
−0.316
CMJ –CMJ_2t
(cm)
81.57
2.57
87.86
1.07
−6.28
2.06
−8.08
0.0001
−8.189
−4.382
_2t—post−test results, bold format—significant differences.
Int. J. Environ. Res. Public Health 2019, 16, 3138
7 of 13
Figure 1. Sprint step kinematics characterizing the first nine steps (S) in the 20-m flying start sprint.
Post-test variability in the sprint step characteristics was significant for group and step (p < 0.05)
(Table 5). The interactions ST × group, SL × group, ST × SL, and ST × SL × group did not show variability
except SV and CT (p < 0.05). Consequently, the consistent generation of high horizontal velocity in the
run-up resulted in greater running velocity with stable SV when sprinting the 20 m distance.
Table 5. ANOVA results of sprint step kinematics.
Feature
Main Effect
Group
ST
PT ×
Group
Step
Step ×
Group
ST × step
ST × step × group
F
p
F
p
F
p
F
p
F
p
F
p
F
p
CT
0.27
0.6129 4.09
0.0660 2.79
0.1206 2.14
0.0388 0.74
0.6540 0.19
0.9918 2.34
0.0240
FT
9.05
0.0109 0.53
0.4789 0.85
0.3756 10.58
0.0000 0.24
0.9812 0.54
0.8220 0.57
0.8016
SF
4.91
0.0468 0.33
0.5763 0.73
0.4100 0.90
0.5167 1.68
0.1121 0.89
0.5247 1.13
0.3484
SL
0.61
0.4489 0.32
0.5841 0.08
0.7828 18.23
0.0000 0.25
0.9798 0.50
0.8503 0.50
0.8519
SV
56.64
0.0000 2.72
0.1249 0.74
0.4080 8.22
0.0000 2.06
0.0475 0.48
0.8703 0.80
0.6012
SST—sprint-speed training; bold format—significant differences.
Table 6 presents the pre- and post-test correlation coefficients for the performance measures in the
20-m flying start, 40-m standing start, and 60-m block start sprints and vertical and horizontal jumping
tests. At pre-test, the only significant correlation in the sub-elite group was between the 60-m block
start and SLJ (r = −0.76) and SFJ (r = −0.77). In the elite group, a significant relationship was found
between the 60-m block start and the 20-m flying start (r = 0.81) and 40 m standing start (r = −0.76).
No other significant pre-test correlations were found. Post-test analysis revealed correlations between
the 20-m flying start and 40-m standing start and SLJ and STJ performance in the sub-elite group.
In the elite sprinters, only the correlation between 60-m block start and SFJ performance was significant
(r = −0.87). Many of the horizontal jump tests (SLJ, STJ, SFJ) were strongly associated with each other,
however, no significant pre- and post-test correlations were observed between CMJ with any of the
variables in either group.
Int. J. Environ. Res. Public Health 2019, 16, 3138
8 of 13
Table 6. Spearman rank correlation coefficients between sprint performance and lower extremity
explosive power variables.
Sub-Elite
Variable
Elite
[7]
[6]
[5]
[4]
[3]
[2]
[1]
[1]
[2]
[3]
[4]
[5]
[6]
[7]
-
−0.77
−0.76
0.85
-
-
60 m [1]
-
0.81
−0.76
-
−0.87 *
-
-
-
-
−0.76 *
0.90 *
-
-
-
-
-
-
-
-
-
-
-
−0.85 *
-
-
-
-
20 m flying
start [2]
-
-
-
-
-
-
-
-
-
−0.76 *
-
-
-
-
40 m [3]
-
-
-
-
-
-
-
−0.79
0.82 *
0.79 *
-
-
-
-
SLJ [4]
-
-
-
-
-
0.85
*
-
-
−0.90
-
-
-
-
-
STJ [5]
-
-
-
-
-
0.79
-
-
0.79 *
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
SFJ [6]
-
-
-
-
-
-
-
CMJ [7]
Italic means pre-test result, bold with * means post-test result.
Hierarchical cluster analysis of the grouped variables is illustrated in Figure 2 (pre-test) and
Figure 3 (post-test). Comparison of the two dendrograms did not reveal any differences between the
emerging clusters. At both time points, the individual clusters were grouped similarly to form two
large characteristic aggregations. This suggests a congruency of the kinematic variables and a relatively
loose relationship with no significant effect of one cluster on the other. Additionally, the Euclidean
distances of the clusters in the pre-test and post-test clusters were similar.
Figure 2. Dendrogram clustering of pre-test sprint step kinematics.
Int. J. Environ. Res. Public Health 2019, 16, 3138
9 of 13
Figure 3. Dendrogram clustering of post-test sprint step kinematics.
4. Discussion
Significant post-test improvements in 60-m block start times were observed in both the elite and
sub-elite sprinters by 1.04% and 1.23%, respectively. Improvements were also noted in 20-m flying
start performance by 4.84% and 3.62% in the elite and sub-elite sprinters, respectively. Only marginal
improvements (not statistically significant) were found for 40-m standing start performance, in which
sprinting time improved by 0.98% in the elite group (Table 5). While indicative that ST has a positive
effect on sprint performance, the findings are difficult to interpret due to the lack of similar data in the
literature. However, comparisons can be made with studies that examined the effects of plyometrics
training on maximal running velocity. In this context, the results of the present study are comparable
with those reported by Kreamer et al. [27], Hennessy and Kilty [12], and Mackala and Fostiak [3].
There is strong evidence that plyometrics training enhances the stretch-shortening cycle of muscle
to improve elastic energy storage [28] and generate faster and more powerful movements [29]. ST is
similar in this regard in that it can likewise activate the elastic properties of muscle fibers and connective
tissue to also allow greater elastic energy storage that, after its release, can provide additional impetus
during running [6].
An important question in this regard is whether ST is more effective than plyometrics training in
enhancing explosive power and maximal running velocity. The data from this study can be compared
with previous research that used an identical testing protocol to assess the effects of six sessions of
plyometrics training [3]. When compared with this study, ST shows greater improvements in 20- and
60-m flying start sprint times (by 0.5% to 2%, respectively) and horizontal jump performance (STL and
STJ). In turn, the plyometrics intervention showed greater increases in standing jump (SJ) and CMJ
performance [3]. Some studies reported strong correlations (r) from 0.65 to 0.90 between sprinting
Int. J. Environ. Res. Public Health 2019, 16, 3138
10 of 13
and drop jump (DJ), SJ, CMJ performance, depending on the sprint distance (20–100 m) and type of
jump [12,30,31]. These findings are comparable with the present investigation, in which significant
post-test correlations were observed between 60-m block start times and SLJ in the sub-elite (r = −0.76)
and STJ in the elite (r = −0.85) group. Strong correlations were also observed between 20-m flying start
times and STJ performance (r = −0.85) in the sub-elite group but not in the elite group.
The secondary purpose of the study was to examine the effects of ST on sprint step kinematics
(SL, SF, FT, CT, and SV). Analysis of SL in both groups of sprinters was compared with other research
where high-performance sprinters were investigated [16,22,32–34]. These studies reported that SL
increased with running velocity and with sprint distance. In our study, linear increases in SL (between
steps two and eight in the elite and steps two and six and also step eight in the sub-elite sprinters) were
observed in both groups at both time points. SL increased only in the sub-elite group by approximately
4 cm at post-test with no changes observed in the elite group. SL was maintained at 228 and 230 cm
(sub-elite and elite, respectively) in the last three steps (Figure 1). This may be explained by the fact
that the flying start involved a 20-m run-up and, therefore, a combined distance of 40 m. Therefore, it is
possible that SL was still increasing with each step (build-up phase) that only plateaued at the end of
the 20 m sprint distance when maximal velocity was attained. In turn, the changes in SF were less
pronounced and remained relatively similar in the sub-elite group but slightly decreased from 4.38 to
4.33 Hz in the elite group. Post-test SF in the elite group was similar between the second and sixth step
(difference of only 0.03 Hz), whereas greater variability was observed between the second and sixth
step in the sub-elite group at pre- and post-test.
Considering both SF and SL, the elite sprinters presented greater SF and slightly longer SL than
the sub-elite sprinters (Figure 1). This result suggests that the improvement in sprint performance
via increased running velocity does not demonstrate the classic dependency between SL and SF.
This contradicts the study of Bezodis et al. [19], who reported a weak correlation (r = –0.192) between
SL and running velocity but a strong correlation between SF and running velocity (r = 0.886). In turn,
Hunter et al. (2004) found a strong correlation between sprint velocity and SL (r = 0.73) and only
a weak correlation between sprint velocity and SF (r = –0.14). While this contradicts the previous
finding, it does confirm Delecluse et al. [35] who used regression analysis to find that ca. 85% of
variance in running velocity can be explained by variance in SL. Similarly, Mackala [34] examined
whether an increase in SF or SL would increase running velocity to find that SL was more strongly
associated with running velocity than SF.
Post-test SV increased from 9.10 m/s to 9.26 m/s in the sub-elite and from 9.93 m/s to 9.98 m/s in
the elite group. Single sprint step execution in the elite sprinters showed a linear increase across steps
one to nine at both pre- and post-test with no changes in SL in step four and seven when compared
with earlier steps three and six. In turn, the sub-elite sprinters showed more pronounced variation in
SV and SL, which, in turn, may have perturbed the sprinting movement and thus explain the slower
running velocity (Figure 1). As SV is a product of reduced CT during the support phase, this may
explain the increased running velocity in the elite sprinters. According to Coh et al. [17], and Alcaraz
et al. [36] the most important factor in sprint step efficiency is the support phase, especially the ratio
between the braking and propulsion phases. Therefore, maximal running velocity can be achieved
only if the force impulse is as small as possible during the braking phase and may be possible by
positioning the foot of the push-off leg as close as possible to the vertical projection of the body’s center
of gravity on the surface. Although this was not measured in the present study, this is the most rational
explanation for the increases in running velocity in the 20-m flying start. Hence, the various sprint
distances involved in the SST intervention may have increased execution economy during the support
phase by positioning the center of gravity closer to the fulcrum upon landing, thereby increasing the
velocity of each step.
No significant changes were also observed in CT and FT in either group of sprinters (Figure 1).
The difference between CT and FT was ca. 0.04 s and was relatively linear from step two to step
eight at both pre- and post-test. More variability was observed in post-test FT in the elite sprinters
Int. J. Environ. Res. Public Health 2019, 16, 3138
11 of 13
although the difference between the minimum and maximum values is 0.03 s. CT was reduced in the
elite sprinters and was comparable with values reported in other studies during maximal sprinting
(90–120 ms) [15,37,38] noted a decreasing trend in CT in the first 10 sprint steps after which CT stabilizes.
To better understand the effects of ST on the spatial and temporal variability of sprinting,
hierarchical cluster analysis was applied. Post-test analysis revealed that the variables in the first cluster
show greater Euclidean distances between CT and FT (6.5 at pre-test and 7.5 at post-test). Changes were
also observed in the grouping order regarding the CT, which are arranged in the order of the executed
steps. Post-test changes in the second cluster (SF and SV) revealed closer sub-cluster linkages (2.5 units
at pre-test and 2 at post-test). These results suggest that SF and SV show considerable dependency and
may be associated with improvements in running velocity. Similar conclusions can be assumed for CT
and FT based on the Euclidean distances.
5. Conclusions
In summary,
this study has shown that the application of eight SST sessions are
effective in significantly increasing 60-m block start and 20-m flying start sprint performance.
Significant improvements were also observed in lower extremity explosive power as ascertained by
vertical and horizontal jump testing. Greater increases were observed in the CMJ (mean 7.95% increase)
than in the horizontal jumps (mean 2.5–5.3% increase). Increased 20-m running velocity was associated
with increases SV, as CT and FT did not change after SST and only relatively linear increases were
observed in SL and SF from step two to step eight.
Additionally, sprint-speed training can be recommended as an effective short-term intervention
to improve sprint performance and lower extremity explosive power, particularly when considering
the required training volume of eight sessions (across 22 days). Sprint performance gains can also be
optimized by decreasing variability in sprint step kinematics during maximal velocity running in both
lower and higher performing sprinters.
Author Contributions: Conceptualization, M.K., F.M., O.T.; methodology, M.K., F.M., C.M., software, M.K., F.M.;
validation, C.M., O.T.; formal analysis, M.K., F.M., C.M., O.T., S.B.; investigation, M.K., F.M., O.T.; a resources,
F.M.; data curation, M.K., C.M., S.B.; writing—original draft preparation, MK., F.M.; writing—review and editing,
M.K., SB.; visualization, M.K., F.M., C.M.; supervision, M.K., C.M.
Funding: This research received no external funding.
Conflicts of Interest: The authors have no conflict of interest to declare. The results do not constitute endorsement
of any product or device. The authors would like to thank the sprinters who participated in this study.
References
1.
Charag, S.A.; Pal, R.; Yadav, S. Effect of plyometric training on muscular power and aerobic ability of the
novice sprinters. Asian J. Phys. Educ. Comput. Sci. Sports 2011, 4, 77–81.
2.
Kukolj, M.; Ropret, R.; Ugarkovic, D.; Jaric, S. Anthropometric, strength, and power predictors of sprinting
performance. J. Sports Med. Phys. Fit. 1999, 39, 120–122.
3.
Mackala, K.; Fostiak, M. Acute effects of plyometric intervention—Performance improvement and related
changes in sprinting gait variability. J. Strength Cond. Res. 2015, 29, 1956–1965. [CrossRef] [PubMed]
4.
Markovi´c, G.; Juki´c, I.; Milanovi´c, D.; Metikoš, D. Effects of sprint and plyometric training on muscle function
and athletic performance. J. Strength Cond. Res. 2007, 21, 543–549. [PubMed]
5.
Rimmer, E.; Sleivert, G. Effects of plyometric intervention program on sprint performance. J. Strength
Cond. Res. 2000, 14, 295–301.
6.
Coh, M.; Baˇci´c, V.; Mackala, K. Biomechanical, neuromuscular and methodical aspects of running speed
development. J. Hum. Kinet. 2010, 26, 73–81. [CrossRef]
7.
Bishop, D.J. Fatigue during intermittent-sprint exercise. Clin. Exp. Pharm. Physiol. 2012, 39, 836–841.
[CrossRef]
8.
Billaut, F.; Basset, F.A. Effect of different recovery patterns on repeated-sprint ability and neuromuscular
responses. J. Sports Sci. 2007, 25, 905–913. [CrossRef]
Int. J. Environ. Res. Public Health 2019, 16, 3138
12 of 13
9.
Girard, O.; Mendez-Villanueva, A.; Bishop, D. Repeated-sprint ability: Part I. Factors contributing to fatigue.
Sports Med. 2011, 41, 673–694. [CrossRef]
10.
Mendez-Villanueva, A.; Hamer, P.; Bishop, D. Fatigue in repeated sprint exercise is related to muscle power
factors and reduced neuromuscular activity. Eur. J. Appl. Physiol. 2008, 103, 411–419. [CrossRef]
11.
Harrison, A.J.; Keane, S.P.; Coglan, J. Force-velocity relationship and stretch-shortening cycle function in
sprint and endurance athletes. J. Strength Cond. Res. 2004, 18, 473–479. [PubMed]
12.
Hennessy, L.; Kilty, J. Relationship of the stretch-shortening cycle to sprint performance in trained female
athletes. J. Strength Cond. Res. 2001, 15, 326–333. [PubMed]
13.
Johnson, M.D.; Buckley, J.G. Muscle power patterns in the mid-acceleration phase of Sprinting. J. Sports Sci.
2001, 19, 263–272. [CrossRef] [PubMed]
14.
Bezodis, I. Investigations of the step length-step frequency relationship in sprinting: Applied implications
for performance. In Proceedings of the XXX International Conference on Biomechanics in Sports, Melbourne,
Australia, 2–6 July 2012; pp. 43–49.
15.
Salo, A.; Bezodis, I.N.; Batterham, A.M.; Kerwin, D.G. Elite sprinting: Are athletes individually step-frequency
or step-length reliant? Med. Sci. Sports Exerc. 2011, 43, 1055–1062. [CrossRef] [PubMed]
16.
Hunter, J.P.; Marshall, R.N.; McNair, P.J. Interaction of step length and step rate during sprint running.
Med. Sci. Sports Exerc. 2004, 36, 261–271. [CrossRef] [PubMed]
17.
Coh, M.; Skof, B.; Kugovnik, O.; Dolenec, A. Kinematic-Dynamic Model of Maximal Speed of Young Sprinters.
In Proceedings of the XII International Symposium on Biomechanics in Sports, Budapest, Hungary, 2–6 July
1994.
18.
Babi´c, V.; Coh, M.; Dizdar, D. Differences in kinematics parameters of athletes of different running quality.
Biol. Sport 2011, 28, 15–121. [CrossRef]
19.
Bezodis, I.N.; Salo, A.I.T.; Kerwin, D.G. A longitudinal case study of step characteristics in a world class
sprint athlete. In Proceedings of the XXVI International Conference on Biomechanics in Sports; Kwon, Y.H.,
J. Shim, J., Shim, J.K., Shin, I.S., Eds.; Rainbow Books: Seoul, Korea, 2008; pp. 537–540.
20.
Chatzilazaridis, I.; Panoutsakopoulos, V.; Papaiakovou, G.I. Stride characteristics progress in a 40-M sprinting
test executed by male preadolescent, adolescent and adult athletes. Biol. Exerc. 2012, 8, 59–77. [CrossRef]
21.
Mackala, K.; Mero, A.A. Kinematic analysis of three best 100 m performance ever. J. Hum. Kinet. 2013, 36, 149–160.
22.
Gajer, B.; Thepaut-Mathieu, C.; Lehenaff, D. Evolution of stride and amplitude during course of the 100 m
event in athletics. New Stud. Athl. 1999, 14, 43–50.
23.
Segers, V.; Lenoir, M.; Aerts, P.; De Clercq, D. Kinematics of the transition between walking and running
when gradually changing speed. Gait Posture 2007, 26, 349–361. [CrossRef]
24.
Wilson, C.; Gittoes, M.; Heywood, P. The effect of pace on stride characteristics and variability in sprint
running. In Proceedings of the XXVI International Symposium of Biomechanics in Sports (ISBS), Seoul, Korea,
14–18 July 2008; pp. 456–459.
25.
Kampmiller, T.; Vanderka, M.; Šelinger, P.; Šelingerová, M.; ˇCierna, D. Kinematic parameters of the running
stride in 1- to 18-yeard-old youth. Kinesiol. Slov. 2011, 17, 63–75.
26.
Ozsu, I. Biomechanical structure of sprint start and effect of biological feedback methods on sprint start
performance. Turk. J. Sport Exerc. 2014, 16, 72–79. [CrossRef]
27.
Kraemer, J.W.; Ratamess, A.N.; Volek, S.J.; Mazzetti, A.S.; Gomez, I.A. The effect of the Meridian Shoe on
vertical jump and sprint performances following short-term combined plyometric/sprint and resistance
training. J. Strength Cond. Res. 2000, 14, 228–238.
28.
Cormie, P.; McGuigan, M.R.; Newton, R.U. Developing maximal neuromuscular power: Part 1-biological
basis of maximal power production. Sports Med. 2011, 41, 17–38. [CrossRef] [PubMed]
29.
Fatahi, A.; Sadeghi, H. Resistance, plyometrics and combined training in children and adolescents’ volleyball
players: A review Study. J. Sci. Res. Rep. 2014, 20, 2584–2610. [CrossRef] [PubMed]
30.
Mehmet, K.; Alper, A.; Coskun, B.; Caner, A. Relationship among jumping performance and sprint parameters
during maximum speed phase in sprinters. J. Strength Cond. Res. 2009, 23, 2272–2279.
31.
Wild, J.; Bezodis, N.; Blagrove, R.; Bezodis, I. Biomechanical comparison of accelerative and maximum
velocity sprinting: Specific strength training consideration. U. K. Strength Cond. Assoc. 2011, 21, 23–36.
32.
Bruggemann, G.P.; Koszewski, D.; Muller, D. Biomechanical Research Project: Athens 1997; Final Report;
Meyer & Meyer Sport: Oxford, UK, 1999; pp. 12–41.
Int. J. Environ. Res. Public Health 2019, 16, 3138
13 of 13
33.
Shen, W. The effects of stride length and frequency on the speeds of elite sprinters in 100 meter dash.
In Proceedings of the XVIII International Symposium of Biomechanics in Sports (ISBS), Hong-Kong, China,
25–30 June 2000; pp. 333–336.
34.
Mackala, K. Optimization of performance through kinematic analysis of the different phases of the 100 meters.
New Stud. Athlet. 2007, 22, 7–16.
35.
Delecluse, C.H.; van Coppenolle, H.; Willems, R.; Diels, M.; Goris, M.; van Leempurte, M.; Vuylsteke, M.
Analysis of 100 meter sprint performance as a multi-dimensional skill. J. Hum. Mov. Stud. 1995, 28, 87–101.
36.
Alcaraz, P.E.; Palao, J.M.; Elvira, J.L.L.; Linthorne, N.P. Effects of three types of resisted sprint training devices
on the kinematics of sprinting at maximum velocity. J. Strength Cond. Res. 2008, 22, 890–897. [CrossRef]
37.
Hunter, J.P.; Marshall, R.N.; McNair, P.J. Relationships between ground reaction force impulse and kinematics
of sprint-running acceleration. J. Appl. Biomech. 2001, 21, 31–43. [CrossRef]
38.
Coh, M.; Tomazin, K.; Stuhec, S. The biomechanical model of the sprint start and block acceleration. Facta Univ.
Ser. Phys. Educ. Sport 2006, 4, 103–114.
© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access
article distributed under the terms and conditions of the Creative Commons Attribution
(CC BY) license (http://creativecommons.org/licenses/by/4.0/).
| Acute Effects of a Speed Training Program on Sprinting Step Kinematics and Performance. | 08-28-2019 | Mackala, Krzysztof,Fostiak, Marek,Schweyen, Brian,Osik, Tadeusz,Coch, Milan | eng |
PMC3966785 | Figure
S1.
Gaze
in
two
representative
trials.
Distance between the ball image and the point of gaze as a function of time. A)
Gaze
for
a
participant
who
successfully
caught
the
projected
9ly
ball
(after
2.64
s);
B)
Gaze
for
a
participant
who
indicated
that
the
projected
9ly
ball
was
uncatchable
for
her
(after
1.37
s).
See
also
Movies
S1
and
S2,
which
show
scene
camera
recordings
of
these
trials.
!!
0
75
150
225
0
0.5
1
1.5
2
2.5
3
Caught
Tracking
Other
Distance (pixels)
Time to catch (s)
0
0.5
1
1.5
2
2.5
3
Judged to be uncatchable
Time to 'no' (s)
A
B
| Keeping your eyes continuously on the ball while running for catchable and uncatchable fly balls. | 03-26-2014 | Postma, Dees B W,den Otter, A Rob,Zaal, Frank T J M | eng |
PMC10739691 | 1
Vol.:(0123456789)
Scientific Reports | (2023) 13:22865
| https://doi.org/10.1038/s41598-023-49369-7
www.nature.com/scientificreports
Anaerobic threshold using sweat
lactate sensor under hypoxia
Hiroki Okawara 1,4, Yuji Iwasawa 2,4, Tomonori Sawada 1, Kazuhisa Sugai 3, Kyohei Daigo 2,
Yuta Seki 2, Genki Ichihara 2, Daisuke Nakashima 1, Motoaki Sano 2, Masaya Nakamura 1,
Kazuki Sato 3, Keiichi Fukuda 2 & Yoshinori Katsumata 2,3*
We aimed to investigate the reliability and validity of sweat lactate threshold (sLT) measurement
based on the real-time monitoring of the transition in sweat lactate levels (sLA) under hypoxic
exercise. In this cross-sectional study, 20 healthy participants who underwent exercise tests using
respiratory gas analysis under hypoxia (fraction of inspired oxygen [FiO2], 15.4 ± 0.8%) in addition to
normoxia (FiO2, 20.9%) were included; we simultaneously monitored sLA transition using a wearable
lactate sensor. The initial significant elevation in sLA over the baseline was defined as sLT. Under
hypoxia, real-time dynamic changes in sLA were successfully visualized, including a rapid, continual
rise until volitionary exhaustion and a progressive reduction in the recovery phase. High intra- and
inter-evaluator reliability was demonstrated for sLT’s repeat determinations (0.782 [0.607–0.898] and
0.933 [0.841–0.973]) as intraclass correlation coefficients [95% confidence interval]. sLT correlated
with ventilatory threshold (VT) (r = 0.70, p < 0.01). A strong agreement was found in the Bland–Altman
plot (mean difference/mean average time: − 15.5/550.8 s) under hypoxia. Our wearable device enabled
continuous and real-time lactate assessment in sweat under hypoxic conditions in healthy participants
with high reliability and validity, providing additional information to detect anaerobic thresholds in
hypoxic conditions.
It is presumed that hypoxic training helps improve endurance performance in athletes1. Traditional high-altitude
training refers to a state where atmospheric and oxygen pressure decrease and athletes are exposed to chronic
hypobaric hypoxia for many weeks2. Recent studies on normobaric hypoxic exercise have investigated the impact
of the recently popular live low-train high-altitude interventions on athletes’ lifestyles3,4, as prolonged exposure
to low-pressure conditions is not always feasible (travel time, engagement, and expenses) and can lead to health
problems5,6. Anaerobic threshold (AT) and peak oxygen uptake (peak VO2) should be routinely assessed in
hypoxic conditions to practice efficient fitness training during hypoxia7. To date, the ventilatory threshold (VT),
calculated as a noninvasive index of metabolic response to incremental exercise, has been used to determine AT8,9.
The VT assessment method is beneficial; however, VT assessment requires an expensive analyzer and expertise
due to the difficulty in confirming VT based on the oscillations in minute ventilation and inconsistencies among
several factors10. The difference in expertise is reported to worsen the VT determination agreement11. This is
because the respiratory gas analyzer is not readily available in sports settings. Therefore, there is an urgent need
to apply an innovative and simple system to determine AT with high reliability for fitness training under hypoxia.
Flexible, wearable sensing devices can yield vital information about the underlying physiology of a human
participant in a continuous, real-time, and noninvasive manner12,13. Sampling human sweat, rich in physiologi-
cal information, can enable noninvasive monitoring14. We developed a sweat sensor to monitor sweat lactate
levels (sLA) in real-time during progressive exercise in the clinical setting, investigating its use in detecting AT
in healthy individuals and patients with cardiovascular diseases15. sLA has been reported to not reflect blood
lactate during exercise16,17; however, our research group has examined sLA transitions during incremental load
exercise and reported that the sweat lactate threshold was strongly approximated to AT by focusing on the inflec-
tion point where the value increases rapidly during incremental exercise, not the absolute value15,18.
Our sLA sensor is portable and easy to carry, enabling convenient measurements in various environments,
and the continuous collection of only 1 Hz sLA values promises a simpler determination of the inflection point.
Moreover, the need for invasive collection methods, including blood collection, is undesirable considering human
OPEN
1Department of Orthopaedic Surgery, Keio University School of Medicine, Tokyo, Japan. 2Department of
Cardiology, Keio University School of Medicine, Tokyo, Japan. 3Institute for Integrated Sports Medicine, Keio
University School of Medicine, Tokyo, Japan. 4These authors contributed equally: Hiroki Okawara and Yuji
Iwasawa. *email: [email protected]
2
Vol:.(1234567890)
Scientific Reports | (2023) 13:22865 |
https://doi.org/10.1038/s41598-023-49369-7
www.nature.com/scientificreports/
resources for multi-measurements, the possibility of any person to evaluate, and acceptance of the evaluation
target.
Under hypoxia, some researchers previously reported AT evaluation results7. However, similar to normoxia,
it is problematic that the VT evaluation method was applied to broadly cover the sports setting. Therefore, we
aimed to investigate the validity of AT estimation and reliability of the sLA continuously obtained using our sLA
sensor during exercise under hypoxia in healthy participants.
Results
The baseline characteristics of the healthy participants are summarized in Table 1. The participants were males
(100%) with a median (IQR) age of 21 (20–21) years. The temperature and humidity were 28 ± 1 °C and 62 ± 6%
under hypoxia, respectively.
Figures 1 and 2 show the sLA during exercise in hypoxia. During the exercise tests, dynamic changes in the
sLA were continuously measured and projected onto the wearable device without delay, even under hypoxia.
Because of the lack of sweat, the lactate biosensor measured a negligible current response at the commence-
ment of cycling activity. During exercise, sLA increased drastically, and the sweat rate continuously increased
as cycling continued until volitional exhaustion. This drastic sLA increase was not associated with the onset of
sweating (Fig. 1).
Contrary to sLA, the heart rate and VO2 gradually increased from incremental-load exercise initiation to
its end (Fig. 2). At the end of the exercise period, the sLA continued to decrease relatively slowly, mirroring the
decrease in heart rate. The results under normoxia are shown in Supplementary Figs. 1 and 2.
We easily identified the conversion from steady low lactate values to a continuous increase under hypoxia.
Repeated sLT and VT determinations by the same evaluator demonstrated high intra-evaluator reliability
Table 1. Baseline characteristics of participants. Data are presented as median (IQR). BMI body mass index.
Demographic and anthropometric data
Healthy male participants (n = 20)
Age (years)
21.0 (20.0–21.0)
Height (cm)
174.0 (171.0–176.4)
Body weight (kg)
69.0 (62.5–71.1)
BMI (kg/m2)
22.2 (20.5–23.4)
Body fat/body weight (%)
14.2 (12.3–16.7)
Body muscle/body weight (%)
81.3 (78.6–83.1)
Body water/body weight (%)
60.2 (57.5–62.8)
Figure 1. Imaging of sweat lactate levels, local sweat rate, and blood lactate values during incremental exercise
under hypoxia. Representative graphs of sweat lactate levels (orange), local sweat rate (blue), and blood lactate
values (red) during hypoxic exercise with a stepwise incremental protocol (25 W/min) ergometer are shown. VT
ventilatory threshold, sLT sweat lactate threshold.
3
Vol.:(0123456789)
Scientific Reports | (2023) 13:22865 |
https://doi.org/10.1038/s41598-023-49369-7
www.nature.com/scientificreports/
(intraclass correlation [ICC] [2, 1] measured value [95% confidence interval] normo, 0.893 [0.794–0.952]; hypo,
0.782 [0.607–0.898]; and normo, 0.711 [0.500–0.861]; hypo, 0.919 [0.841–0.964]), respectively (Fig. 3 and Sup-
plementary Fig. 3). Moreover, these were reproducible between both blinded reviewers (ICC [1, 1] measured
value [95% confidence interval]; sLT; normo, 0.898 [0.765–0.958], hypo, 0.933 [0.841–0.973], and VT, normo,
0.933 [0.841–0.973], hypo, 0.836 [0.638–0.931]) as shown in Table 2 and Supplementary Table 1. However, the
intra- and inter-evaluation reliability for bLT was low (ICC [1, 1] measured value [95% confidence interval];
normo, 0.529 [0.134–0.781], hypo, 0.652 [0.314–0.845], ICC [2, 1] measured value [95% confidence interval];
normo, 0.621 [0.363–0.813], hypo, 0.586 [0.331–0.790]) as shown in Fig. 3, Table 2, Supplementary Fig. 3, and
Supplementary Table 1.
The relationships between sLT and VT are shown in Fig. 4A and Supplementary Fig. 4A, describing the
strong relationships between each threshold (normo, r = 0.69; hypo, r = 0.70). The Bland–Altman plot revealed
that the mean difference between each threshold was 4.9 s under normoxia and − 15.5 s under hypoxia, and
there was no bias between the mean values, displaying strong agreements between sLT and VT (Fig. 4B and
Supplementary Fig. 4B).
Discussion
The noninvasive sLA sensor enabled continuous and real-time measurement of sLA during an exercise test
under hypoxia. Furthermore, sLT determination had high intra- and inter-evaluator reliability, and sLT was
strongly correlated with VT. Real-time sweat lactate monitoring could be applied to detect aerobic threshold,
even under hypoxia (Fig. 5).
Lactate levels are measured to track an individual’s performance and exertion level19,20. Blood lactate levels
are measured by athletes or their supporters21,22, but these are not continuous, real-time measurements, limiting
their utility to applications where stationary, infrequent tests are sufficient. In particular, applying bLT relies on
Figure 2. Measured parameters in hypoxia. The graph shows the measured parameters [(a) VO2/body weight,
(b) Heart rate, (c) Sweat lactate, (d) sweat rate] at rest, warm up, VT, and peak in hypoxia. Data are shown as
mean (± standard deviation). VO2 oxygen uptake, VT ventilatory threshold, HR heart rate, sLA sweat lactate, SR
sweat rate.
4
Vol:.(1234567890)
Scientific Reports | (2023) 13:22865 |
https://doi.org/10.1038/s41598-023-49369-7
www.nature.com/scientificreports/
the measurement’s reliability; in this study, the intra- and inter-evaluation reliability for bLT was low. Conversely,
even under hypoxia, our devices captured the sLA during fitness in a real-time, noninvasive, and continuous
manner at 1 Hz instead of cumulative values as in the conventional method, which detects the “timing of change”
in a real-time and sensitive manner. Therefore, it is easy to identify the inflection point (sLT) from the plots of the
sLA values. Using sLT demonstrated lower intra- and inter-observer bias and superior determination accuracy.
Figure 3. Reliability testing of the time at sLT determined by the same evaluator in hypoxia. (a) The graph
shows the relationship between the repeatedly determined sweat lactate threshold (sLT) by the same evaluator.
(b) The graph shows the Bland–Altman plots, which indicate the respective differences between the repeatedly
determined sLT by the same evaluator (y-axis) for each individual against the mean of the time at the repeatedly
determined sLT (x-axis) in hypoxia. R correlation coefficient, p p-value, VT ventilatory threshold, sLT sweat
lactate threshold.
Table 2. Intra-evaluator reliability of sweat lactate threshold determination in hypoxia. ICC intraclass
correlation, sLT sweat lactate threshold, bLT blood lactate threshold, VT ventilatory threshold, SD standard
deviation.
Hypoxia
N
Evaluator 1
Evaluator 2
Evaluator 3
ICC (95% CI)
sLT [s]
Mean
20
553.3
486.3
533.6
0.782 (0.607–0.898)
SD
84.4
89.8
80.8
bLT [s]
Mean
20
581.8
558.7
597.6
0.586 (0.331–0.790)
SD
69.7
58.4
73.7
VT [s]
Mean
20
546.8
547.2
540.9
0.919 (0.841–0.964)
SD
66.2
49.8
59.7
Figure 4. Validity testing of the time at VT and sLT in hypoxia. (a) The graph shows the relationship between
the time from the start of the measurement (seconds) at VT and sLT. (b) The graph shows the Bland–Altman
plots, which indicate the respective differences between the time from the start of measurement (s) at the VT
and sLT (y-axis) for each individual against the mean of the time at the VT and sLT (x-axis) in hypoxia. R
correlation coefficient, VT ventilatory threshold, sLT sweat lactate threshold.
5
Vol.:(0123456789)
Scientific Reports | (2023) 13:22865 |
https://doi.org/10.1038/s41598-023-49369-7
www.nature.com/scientificreports/
Another possible explanation to support this positive result is that several operations, including the exchange
of the sensor chip, cleaning the upper arm which the sensor fixed, and flushing out any residual sweat from the
duct in the perspiration meter, certainly could eliminate the bias due to contaminations from previous experi-
ments or original sweating. sLA has been reported to not reflect blood lactate during exercise16,17; however, our
data showed that the AT point coincided with that in the sLA level during progressive exercise, consistent with
the finding of the previous report15,18. This could be because an increase in lactate production from muscle cells,
reflecting LT, may induce a simultaneous rise in sLA levels through changes in autonomic nervous balance, hor-
mones, acid–base equilibrium, and metabolic dynamics23,24 similar to VT25. A previous study has demonstrated
a rapid increase in blood catecholamine concentrations during incremental exercise loads26. Furthermore, it has
also been indicated that sweat gland metabolism is activated by catecholamines27. Therefore, we are evaluating
the timing of physiological responses to increasing exercise loads using completely different analytes and not
estimating the bLA levels by observing the sweat lactate levels.
Measuring VT and peak VO2 with respiratory gas analysis helps in efficient training under hypoxia. However,
it is often difficult to determine VT because of inconsistencies among the several factors required for detecting
VT, such as the ventilation (VE)/oxygen uptake (VO2) or carbon dioxide production (VCO2)/VO2 slope and oscil-
lations in minute ventilation10. Further, a respiratory gas analyzer is unavailable in a small hypoxic booth because
of its size. Moreover, using a facemask, respiratory gas cannot be collected under hypoxic exercise. In addition,
in a respiratory infection epidemic such as COVID-19, using respiratory gas analyzers has become difficult due
to the possibility of cross-infection. Determining sLT using only sweat-based monitoring could overcome these
problems, and the newly developed device enables AT measurements in various hypoxic environments (a small
private booth and facemask).
It has been reported that sweat rate decreases in hypoxia28. As our sensor showed non-response in the absence
of sweating, evaluating sweat rate is paramount to successfully determining sLT in hypoxia. This study quantified
the amount of sweating per unit area near the sensor; the results showed no difference in the local sweat rate
during exercise under hypoxia compared with that under normoxia. The relationship between the local sweat
rate/response in the sLA sensor, humidity, and temperature during exercise warrants further investigation.
Figure 5. Schematic of the lactate-sensing device under hypoxia. This figure is licensed by © Medical FIG. ICC
interclass correlation coefficients, bLT blood lactate threshold, VCO2 carbon dioxide output, VO2 oxygen uptake,
VT ventilatory threshold, sLT sweat lactate threshold, FiO2 fraction of inspiratory oxygen.
6
Vol:.(1234567890)
Scientific Reports | (2023) 13:22865 |
https://doi.org/10.1038/s41598-023-49369-7
www.nature.com/scientificreports/
The device used in our study is suitable for use in remote monitoring or remote training settings during
isolation measures, such as those taken during a respiratory infection epidemic. Furthermore, real-time assess-
ments of sLA through a wireless data transfer system can offer a rigorous training menu under hypoxia based
on the day-to-day physical conditions of trainees. In addition, exercise under hypoxia has been recognized as
a new therapeutic modality for health promotion and disease prevention or treatment, such as for diabetes29,
cardiovascular diseases30, hypertension31, obesity32, and age-related diseases33. Disease prevention and treatment
can be more efficiently and safely provided by combining sLA sensors with exercise under hypoxia.
The study has some limitations. First, due to the observational study design, we could not exclude the influ-
ence of a selection bias. Second, our study had a relatively small number of cases. Third, the current study
included healthy college-aged male individuals. Recent findings could be applied to various age groups and
genders; however, further research, including females and young athletes, is required considering a sweat func-
tional difference between sexes. Fourth, the sLA sensor used in this study exhibited the current value, not the
sLA concentration. Conversion to concentration from the current value is possible; however, it is sufficient to
display the current values to determine the inflection point based on the constant value of sLA during exercise.
The effect of sLA dilution by high sweat rate on sLT determination is minimal due to the low sweat rate at AT
and, therefore, does not negate our study’s result. Finally, exercise training has been performed under various
hypoxic conditions; however, only a hypoxia of 15.5% was verified. Further verification is required to overcome
these limitations.
In conclusion, the noninvasive sweat lactate sensor enabled continuous and real-time measurement of sweat
lactate during exercise under hypoxia. The sweat lactate threshold can also be reliably determined by non-
experts, even under hypoxia. Real-time sweat lactate monitoring could be used to detect aerobic threshold in
a noninvasive and feasible manner under hypoxia and normoxia. It is expected that these findings enhance the
effectiveness of exercise under hypoxia. This was the first study to show real-time monitoring of sLA during
progressive exercise under hypoxia. Given the difficulty in deciding VT, such as in hypoxia, sLA monitoring
could be beneficial in improving VT detection with high reliability.
Methods
Experimental approach to the problem
We conducted a cross-sectional study with 20 healthy participants who underwent exercise tests with respiratory
gas analysis under hypoxia or normoxia and simultaneously monitored changes in sLA using a wearable lactate
sensor to investigate the capability of sweat lactate sensor to monitor sLA under hypoxia and the relationship
between sLT and VT. In addition, Intraclass correlation was determined for the intra- and inter-evaluator reli-
ability of each threshold in this study.
Subjects
Participants aged 20–80 years were recruited through a web system in June 2021. The exclusion criteria were
patients receiving medication, having comorbidities like hypertension, diabetes, and active lung diseases, and
having low local sweat rates of < 0.4 mg/cm2/min at the upper arm during maximal exercise. This sweat rate
threshold was defined based on previous reports15 and preliminary studies. Twenty healthy participants were
enrolled, including athletes and those with a broad spectrum of aerobic capacities and fitness levels. Notably, all
participants exercised regularly for more than twice weekly.
The study protocol was approved by the Institutional Review Board (IRB) of Keio University School of Medi-
cine (approval number 20190229), and the study was conducted following the principles of the Declaration of
Helsinki. Verbal informed consent was obtained from all participants because the IRB approved using verbal con-
sent following the Japanese guidelines for clinical research. Verbal consent was recorded as an experimental note.
Procedures
The twice exercise tests with a minimum of 2 days intervals were performed using an electromagnetically braked
ergometer (POWER MAX V3 Pro, Konami Sports Co., Ltd., Tokyo, Japan) with respiratory gas analysis under
hypoxia (hypo; a fraction of inspired oxygen [FiO2], 15.4 ± 0.8% equivalent to a simulated altitude of 2500 m) or
normoxia (normo; FiO2, 20.9%). Hypoxic conditions were created in an exercise booth with an oxygen filtration
hypoxic generator (Hypoxico Everest Summit II; WILL Co., Tokyo, Japan) by insufflating nitrogen as a target of
FiO2 15.5%34. During exercise, the sLA was monitored using an sLA sensor (Grace Imaging Inc., Tokyo, Japan)
attached to the upper arm, and the local sweat rate was measured at a sampling rate of 1 Hz in the same area as
the sLA sensor using a perspiration meter (SKN-2000M; SKINOS Co., Ltd., Nagano, Japan). A perspiration meter
ensured the value returned to zero before the new experiment by flushing out any residual sweat from the duct.
Heart rate was monitored using Duranta (Zaiken, Tokyo, Japan), and blood lactate levels were measured using
a standard enzymatic method on a lactate analyzer (Lactate Pro2®, ARKRAY, Kyoto, Japan).
On the day of the exercise test, the participants avoided any prior heavy physical activity. The participants
performed the test upright on an electronically braked ergometer. Following a 2-min rest to stabilize the heart rate
and respiratory condition, the participants performed a 4-min warm-up pedaling at 20 W. Then, they exercised at
increasing intensity until they could no longer maintain the pedaling rate (volitional exhaustion). The resistance
was increased in 25-W increments from 50-W at 1-min intervals. Once the exercise tests were terminated, the
participants were instructed to stop pedaling and remain on the ergometer for 3 min.
The expired gas flow collected through the mask was measured using a breath-by-breath automated system
(Aeromonitor®, Minato Medical Science Co., Ltd., Osaka, Japan). This system was subjected to a three-way cali-
bration process involving a flow volume sensor, gas analyzer, and delay time calibration. The gas analyzer was
calibrated under hypoxia using 8% O2, assuming a minimum oxygen concentration of 8% in exhaled air during
7
Vol.:(0123456789)
Scientific Reports | (2023) 13:22865 |
https://doi.org/10.1038/s41598-023-49369-7
www.nature.com/scientificreports/
hypoxic exercise. Respiratory gas exchange, including VE, VO2, and VCO2, was continuously monitored and
measured using a 10-s average. VT was determined using the ventilatory equivalent, excess carbon dioxide, and
modified V-slope methods10 through manual operating software. First, two of the three experienced researchers
independently and randomly evaluated each participant’s VT using the three methods. The researchers used
all three methods to assess concurrent breakpoints and eliminate false breakpoints. Second, if the VO2 values
determined by the independent researchers were within 3%, then the VO2 values from the two investigators were
averaged. Third, if the VO2 values determined by the independent evaluators were not within 3% of one another,
a third researcher independently determined VO2. The third VO2 value was then compared with that obtained
by the initial investigators. If the adjudicated VO2 value was within 3% of either of the initial investigators, the
two VO2 values were averaged.
Blood lactate values were obtained by auricular pricking and gentle squeezing of the ear lobe to obtain a
capillary blood sample at rest, warm-up, and every minute after the start of progressive intensity. The samples
were immediately analyzed for whole-blood lactate concentrations (mmol/L).
bLT was determined through graphical plots of the bLA value vs. time8. Visual interpretation was indepen-
dently made for each participant by two experienced researchers to locate the first rise from baseline. If the
independent determinations of the stage at LT differed between the two researchers, a third researcher adjudi-
cated the difference by independently determining LT. The three researchers then jointly agreed on the LT point.
The sLA was measured using a sLA sensor, which quantifies lactate concentration as a current value because
it reacts with sLA and generates an electric current15. The sLA sensing system comprises a disposable sensor chip
and a sensor. The sensor chip generates the current value proportional to the lactate concentration by catalyzing
the enzymatic immobilization on its surface to oxidize lactate, which reduces hydrogen peroxide. In addition,
a protective film formed by exposure using a UV lamp allows the achievement of immediate responsiveness
(response delay < 1 s) and sustainability without the enzyme reacting all at once15. The current value can be
obtained as continuous data within 0.1–80 μA in 0.1-μA increments. The sLA sensor responded linearly to the
lactate concentrations, especially in the 0–5 mmol/L range, which were most significant in determining the LT
because the LT had normal lactate values from 2 to 4 mmol/L15.
Moreover, it is also validated that the sLA values obtained from this sensor can show a significant enough difference
to determine the inflection point under various sweat environments35. After calibration using saline for 2 or 3 min, the
sensor chip connected to the sensor device was attached to the superior right upper limb of the participants and cleaned
with an alcohol-free cloth to eliminate the influence of original sweat. In addition, the data were recorded at a sampling
frequency of 1 Hz for mobile applications with Bluetooth connection. The recorded data were converted to moving
average values over 13-s intervals and underwent zero correction using the baseline value. sLT was defined as the first
significant increase in the sLA above baseline based on graphical plots15. Three researchers, independent of those who
analyzed respiratory gas exchange, agreed on the point of sLT.
Statistical analyses
The results are represented as mean ± standard deviation for continuous variables and percentages for categorical
variables, as appropriate. ICC was determined for intra- and inter-evaluator reliability of each threshold36. The
intra-evaluator reliability was tested by one of the blinded reviewers. The inter-observer reliability was tested
by estimating each threshold using three blinded reviewers. The relationship between exercise time at sLT and
VT was investigated using Pearson’s correlation coefficient test. In addition, the Bland–Altman technique was
applied to verify the similarities among the different methods37. The graphical representation of the difference
between the methods and the average WAS compared. Statistical significance was set at two-tailed p-values < 0.05.
All statistical analyses were performed using IBM SPSS Statistics for Windows, version 27.0 (IBM Corporation,
Armonk, NY, USA).
Data availability
The datasets used and/or analyzed during the current study are available from the corresponding author upon
reasonable request.
Received: 5 July 2023; Accepted: 7 December 2023
References
1. Lundby, C., Millet, G. P., Calbet, J. A., Bartsch, P. & Subudhi, A. W. Does “altitude training” increase exercise performance in elite
athletes?. Br. J. Sports Med. 46, 792–795 (2012).
2. Owen, J. R. A preliminary evaluation of altitude training particularly as carried out by some members of the Olympic teams of
Great Britain and of other European countries in 1972. Br. J. Sports Med. 8, 9–17 (1974).
3. Girard, O., Brocherie, F., Goods, P. S. R. & Millet, G. P. An updated panorama of “living low-training high” altitude/hypoxic
methods. Front. Sports Act. Living 2, 26 (2020).
4. Girard, O., Goods, P. S. R. & Brocherie, F. Editorial: Elevating sport performance to new heights with innovative ‘live low-train
high’ altitude training. Front. Sports Act. Living 2, 108 (2020).
5. Khodaee, M., Grothe, H. L., Seyfert, J. H. & VanBaak, K. Athletes at high altitude. Sports Health 8, 126–132 (2016).
6. Pena, E., Brito, J., El Alam, S. & Siques, P. Oxidative stress, kinase activity and inflammatory implications in right ventricular
hypertrophy and heart failure under hypobaric hypoxia. Int. J. Mol. Sci. 21, 6421 (2020).
7. Ofner, M. et al. Influence of acute normobaric hypoxia on physiological variables and lactate turn point determination in trained
men. J. Sports Sci. Med. 13, 774–781 (2014).
8. Faude, O., Kindermann, W. & Meyer, T. Lactate threshold concepts: How valid are they?. Sports Med. 39, 469–490 (2009).
9. Weston, S. B. & Gabbett, T. J. Reproducibility of ventilation of thresholds in trained cyclists during ramp cycle exercise. J. Sci. Med.
Sport 4, 357–366 (2001).
8
Vol:.(1234567890)
Scientific Reports | (2023) 13:22865 |
https://doi.org/10.1038/s41598-023-49369-7
www.nature.com/scientificreports/
10. Gaskill, S. E. et al. Validity and reliability of combining three methods to determine ventilatory threshold. Med. Sci. Sports Exerc.
33, 1841–1848 (2001).
11. Kaczmarek, S. et al. Interobserver variability of ventilatory anaerobic threshold in asymptomatic volunteers. Multidiscip. Respir.
Med. 14, 20 (2019).
12. An, B. W. et al. Smart sensor systems for wearable electronic devices. Polymers (Basel) 9, 303 (2017).
13. Kobsar, D. & Ferber, R. Wearable sensor data to track subject-specific movement patterns related to clinical outcomes using a
machine learning approach. Sensors (Basel) 18, 2828 (2018).
14. Baker, L. B. Sweating rate and sweat sodium concentration in athletes: A review of methodology and intra/interindividual vari-
ability. Sports Med. 47, 111–128 (2017).
15. Seki, Y. et al. A novel device for detecting anaerobic threshold using sweat lactate during exercise. Sci. Rep. 11, 4929 (2021).
16. Green, J. M., Bishop, P. A., Muir, I. H., McLester, J. R. Jr. & Heath, H. E. Effects of high and low blood lactate concentrations on
sweat lactate response. Int. J. Sports Med. 21, 556–560 (2000).
17. Lamont, L. S. Sweat lactate secretion during exercise in relation to women’s aerobic capacity. J. Appl. Physiol. 62, 194–198 (1987).
18. Maeda, Y. et al. Implications of the onset of sweating on the sweat lactate threshold. Sensors 237, 3378 (2023).
19. Buono, M. J., Lee, N. V. & Miller, P. W. The relationship between exercise intensity and the sweat lactate excretion rate. J. Physiol.
Sci. 60, 103–107 (2010).
20. Falk, B. et al. Sweat lactate in exercising children and adolescents of varying physical maturity. J. Appl. Physiol. 71, 1735–1740
(1991).
21. Jansen, T. C. et al. Early lactate-guided therapy in intensive care unit patients: A multicenter, open-label, randomized controlled
trial. Am. J. Respir. Crit. Care Med. 182, 752–761 (2010).
22. Vincent, J. L., Quintairos, E. S. A., Couto, L. Jr. & Taccone, F. S. The value of blood lactate kinetics in critically ill patients: A sys-
tematic review. Crit. Care 20, 257 (2016).
23. Alvear-Ordenes, I., García-López, D., De Paz, J. A. & Gonzáles-Gallego, J. Sweat lactate, ammonia, and urea in rugby players. Int.
J. Sports Med. 26, 632–637 (2005).
24. Shiraishi, Y. et al. Real-time analysis of the heart rate variability during incremental exercise for the detection of the ventilatory
threshold. J. Am. Heart Assoc. 7, e006612 (2018).
25. Peinado, A. B., Rojo, J. J., Calderón, F. J. & Maffulli, N. Responses to increasing exercise upon reaching the anaerobic threshold,
and their control by the central nervous system. BMC Sports Sci. Med. Rehabil. 6, 17 (2014).
26. Tanoue, Y. et al. The ratio of heart rate to heart rate variability reflects sympathetic activity during incremental cycling exercise.
Eur. J. Sport Sci. 22, 1714–1723 (2022).
27. Sato, K. The physiology, pharmacology, and biochemistry of the eccrine sweat gland. Rev. Physiol. Biochem. Pharmacol. 79, 51–131
(1977).
28. DiPasquale, D. M., Kolkhorst, F. W., Nichols, J. F. & Buono, M. J. Effect of acute normobaric hypoxia on peripheral sweat rate. High
Alt. Med. Biol. 3, 289–292 (2002).
29. Mai, K. et al. Hypoxia and exercise interactions on skeletal muscle insulin sensitivity in obese subjects with metabolic syndrome:
Results of a randomized controlled trial. Int. J. Obes. 44, 1119–1128 (2020).
30. Sprick, J. D. & Rickards, C. A. Combining remote ischemic preconditioning and aerobic exercise: A novel adaptation of blood flow
restriction exercise. Am. J. Physiol. Regul. Integr. Comp. Physiol. 313, R497–R506 (2017).
31. Kong, Z., Zang, Y. & Hu, Y. Normobaric hypoxia training causes more weight loss than normoxia training after a 4-week residential
camp for obese young adults. Sleep Breath 18, 591–597 (2014).
32. Park, H. Y., Jung, W. S., Kim, J. & Lim, K. Twelve weeks of exercise modality in hypoxia enhances health-related function in obese
older Korean men: A randomized controlled trial. Geriatr. Gerontol. Int. 19, 311–316 (2019).
33. Kayser, B. & Verges, S. Hypoxia, energy balance and obesity: From pathophysiological mechanisms to new treatment strategies.
Obes. Rev. 14, 579–592 (2013).
34. Kon, M. et al. Effects of acute hypoxia on metabolic and hormonal responses to resistance exercise. Med. Sci. Sports Exerc. 42,
1279–1285 (2010).
35. Muramoto, Y. et al. Estimation of maximal lactate steady state using the sweat lactate sensor. Sci. Rep. 13, 10366 (2023).
36. Rousson, V., Gasser, T. & Seifert, B. Assessing intrarater, interrater and test-retest reliability of continuous measurements. Stat.
Med. 21, 3431–3446 (2002).
37. Bland, J. M. & Altman, D. G. Statistical methods for assessing agreement between two methods of clinical measurement. Lancet
1, 307–310 (1986).
Acknowledgements
We are grateful to Editage for editing this manuscript. This study was funded by the Grant-in-Aid from Scien-
tific Research from the Japan Agency for Medical Research and Development (ID. 21ek0210130h0003) and by a
grant from the Kimura Memorial Heart Foundation Research Grant for 2019, Suzuken Memorial Foundation,
Foundation for Total Health Promotion, and Research Grant for Public Health Science. The funders had no role
in study design, data collection and analysis, and the decision to publish or prepare the manuscript.
Author contributions
The author contributions are stated as follows; the manuscript was drawn by Y.I., H.O., and T.S. The images were
prepared by Y.I., H.O., T.S., and Y.K. The patient information was collected by H.O., T.D., K.S., K.D., Y.S., G.I.,
D.N., and Y.K. A critical revision of the manuscript for key intellectual content and supervision was provided by
M.S., K.S., K.F., and Y.K. All authors have approved all aspects of our work, read, and approved the manuscript.
Competing interests
D.N. is a founder and shareholder of Grace Imaging Inc. All other authors declare no competing interests.
Additional information
Supplementary Information The online version contains supplementary material available at https:// doi. org/
10. 1038/ s41598- 023- 49369-7.
Correspondence and requests for materials should be addressed to Y.K.
Reprints and permissions information is available at www.nature.com/reprints.
Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and
institutional affiliations.
9
Vol.:(0123456789)
Scientific Reports | (2023) 13:22865 |
https://doi.org/10.1038/s41598-023-49369-7
www.nature.com/scientificreports/
Open Access This article is licensed under a Creative Commons Attribution 4.0 International
License, which permits use, sharing, adaptation, distribution and reproduction in any medium or
format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the
Creative Commons licence, and indicate if changes were made. The images or other third party material in this
article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the
material. If material is not included in the article’s Creative Commons licence and your intended use is not
permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from
the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/.
© The Author(s) 2023
| Anaerobic threshold using sweat lactate sensor under hypoxia. | 12-21-2023 | Okawara, Hiroki,Iwasawa, Yuji,Sawada, Tomonori,Sugai, Kazuhisa,Daigo, Kyohei,Seki, Yuta,Ichihara, Genki,Nakashima, Daisuke,Sano, Motoaki,Nakamura, Masaya,Sato, Kazuki,Fukuda, Keiichi,Katsumata, Yoshinori | eng |
PMC5266769 | ORIGINAL RESEARCH ARTICLE
Health and Economic Burden of Running-Related Injuries
in Dutch Trailrunners: A Prospective Cohort Study
Luiz Carlos Hespanhol Junior1 • Willem van Mechelen1,2,3,4 • Evert Verhagen1,3,5
Published online: 25 May 2016
The Author(s) 2016. This article is published with open access at Springerlink.com
Abstract
Background Trailrunning
is
becoming
very
popular.
However, the risk and burden of running-related injuries
(RRI) in trailrunning is not well established.
Objective To
investigate
the
prevalence,
injury
rate,
severity, nature, and economic burden of RRIs in Dutch
trailrunners.
Methods This prospective cohort study included 228
trailrunners aged 18 years or over (range 23–67), and was
conducted between October 2013 and December 2014.
After completing the baseline questionnaire, the Oslo
Sports Trauma Research Center Questionnaire on Health
Problems was administered every 2 weeks to collect data
on RRIs. Participants who reported RRIs were asked about
healthcare utilization (direct costs) and absenteeism from
paid work (indirect costs). RRI was defined as disorders of
the musculoskeletal system or concussions experienced or
sustained during participation in running.
Results The mean prevalence of RRIs measured over time
was 22.4 % [95 % confidence interval (CI) 20.9–24.0], and
the injury rate was 10.7 RRIs per 1000 h of running (95 %
CI 9.4–12.1). The prevalence was higher for overuse
(17.7 %; 95 % CI 15.9–19.5) than for acute (4.1 %; 95 %
CI 3.3–5.0) RRIs. Also, the injury rate was higher for
overuse (8.1; 95 % CI 6.9–9.3) than for acute (2.7; 95 % CI
2.0–3.4) RRIs. The median of the severity score was 35.0
[25–75 %, interquartile range (IQR) 22.0–55.7], and the
median of the duration of RRIs was 2.0 weeks (IQR
2.0–6.0) during the study. The total economic burden of
RRIs was estimated at €172.22 (95 % CI 117.10–271.74)
per RRI, and €1849.49 (95 % CI 1180.62–3058.91) per
1000 h of running. An RRI was estimated to have a direct
cost of €60.92 (95 % CI 45.11–94.90) and an indirect cost
of €111.30 (95 % CI 61.02–192.75).
Conclusions The health and economic burden of RRIs
presented in this study are significant for trailrunners and
for society. Therefore, efforts should be made in order to
prevent RRIs in trailrunners.
Electronic supplementary material The online version of this
article (doi:10.1007/s40279-016-0551-8) contains supplementary
material, which is available to authorized users.
& Luiz Carlos Hespanhol Junior
[email protected]
1
Amsterdam Collaboration on Health and Safety in Sports,
Department of Public and Occupational Health and the
EMGO? Institute for Health and Care Research,
VU University Medical Center Amsterdam, Van der
Boechorststraat 7, 1081 BT Amsterdam, The Netherlands
2
School of Human Movement and Nutrition Sciences, Faculty
of Health and Behavioural Sciences, University of
Queensland, Brisbane, QLD, Australia
3
UCT/MRC Research Unit for Exercise Science and Sports
Medicine (ESSM), Department of Human Biology, Faculty
of Health Sciences, University of Cape Town, Cape Town,
South Africa
4
School of Public Health, Physiotherapy and Population
Sciences, University College Dublin, Dublin, Ireland
5
Australian Centre for Research into Injury in Sport and its
Prevention, Federation University Australia, Ballarat, VIC,
Australia
123
Sports Med (2017) 47:367–377
DOI 10.1007/s40279-016-0551-8
Key Messages
At any given time, one in five trailrunners report
having a running-related injury (RRI).
Of the RRIs in trailrunners, 75.2 % were overuse
injuries, and the prevalence of overuse RRIs was
fourfold higher than acute RRIs.
The indirect cost of RRIs (related to absenteeism
from paid work) was twofold higher than the direct
cost (related to healthcare utilization).
1 Introduction
Physical activity is a cost-effective and cost-saving inter-
vention to improve overall health and gain healthy life-
years [1–4]. There is evidence claiming that physical
activity participation in outdoor environments has a larger
beneficial effect on physical and mental wellbeing than
participation in indoor physical activities [5]. Coinciden-
tally, trailrunning, a mode of running consisting of running
in the outdoors on unpaved and hilly/mountain terrains, is
quickly gaining in popularity worldwide. The trailrunning
community is composed of well trained trailrunners who
participate in ultra-marathon events ([42.2 km), but also
increasingly by trailrunning enthusiasts who partake in
trailrunning events with shorter distances.
Running is a very popular mode of exercise among
people seeking an active lifestyle [6, 7]. Next to being
beneficial for health [8–10], running also carries a risk of
running-related injuries (RRI) with incidence rates ranging
from 7.7 [95 % confidence interval (CI) 6.9–8.7] to 17.8
(95 % CI 16.7–19.1) RRIs per 1000 h of running in
recreational and novice runners, respectively [11]. How-
ever, prospective data on the risk and burden (including
costs) of RRIs in trailrunning are sparse, especially in
cohorts including trailrunning enthusiasts that compose the
general trailrunning population.
Most RRIs have an overuse nature [12] of which the
symptoms can last for several weeks [13]. Also, these
injuries can negatively influence physical activity partici-
pation [14, 15]. Consequently, measuring overuse injuries
next to acute injuries is important to understand the overall
burden of RRIs [15]. However, measuring overuse injuries
is challenging, because of their non-identifiable and gradual
onset, and also due to fluctuation of symptoms over time
[16]. Most studies about running have measured RRIs
leading to consequences, such as time loss (i.e. running
sessions not fully accomplished or completely missed due
to RRIs) and/or medical attention [17]. Defining RRI based
only on these consequences could underestimate the overall
burden of RRIs, since minor injuries not resulting in such
consequences would be neglected [18, 19]. Also, to register
overuse injuries accurately, one needs a long follow-up time
including regular measurement intervals in order to
chart the gradual onset and fluctuations of symptoms related
to overuse RRIs [16, 18]. Such data are sparse in the RRI
literature, and completely missing in trailrunning.
The purpose of this study was therefore to prospectively
investigate the prevalence, injury rate, severity, nature, and
economic burden of acute and overuse RRIs in Dutch
trailrunners. Such data may assist in the development of
RRI prevention programs in this mode of running, and also
may assist in decisions related to allocation of public health
financial resources.
2 Methods
2.1 Participants
This study was composed of a convenience sample of the
general Dutch trailrunning population. Individuals engaged
in trailrunning were invited to partake in the study via flyer
cards
distributed
during
trailrunning
events
in
The
Netherlands, and also by social media channels, newslet-
ters, and the MudSweatTrails (MST) website [20]. The
flyer cards and additional recruitment sources guided the
individuals to the project’s website containing further
information and the option to enroll in the study. Individ-
uals who agreed to participate through online informed
consent, aged 18 years or over, reported running on
unpaved surfaces on a regular basis, and who completed
the baseline questionnaire were included in the study. A
sample size calculation a priori was not possible because of
a lack of information on the prevalence of RRIs repeatedly
measured over time at the commencement of this study.
The study was approved by the medical ethics committee
of the VU University Medical Center Amsterdam, The
Netherlands.
2.2 Study Design
This was a prospective open cohort study conducted
between October 2013 and December 2014. This cohort
was composed of a dynamic sample, i.e., the participants
entered into the study at different time-points and, there-
fore, they had different follow-up periods. However, all
participants were followed for at least 6 months. After
giving informed consent, a link to a secure online baseline
questionnaire was sent by e-mail to the participants. This
368
L. C. H. Junior et al.
123
questionnaire asked about demographics, running experi-
ence, participation in other sports, current medical condi-
tions, previous (last 12 months) RRIs, and current RRIs.
Online follow-up questionnaires were completed every
2 weeks via a secure link sent by email. The aim of these
follow-up questionnaires was to collect data about the
participants’ running exposure (overall exposure and on
unpaved surfaces specifically) and to record any health
problems experienced in the preceding 2 weeks. In case of
a sustained RRI, information about healthcare utilization
and absenteeism from paid work related to the RRI were
also registered through the same follow-up questionnaires
(conditional branching questions). If no response was
received within 1 week, a reminder was sent by e-mail
encouraging the participant to complete the follow-up
questionnaire.
2.3 Health Problems Registration
In order to prospectively register health problems during
the follow-up, the translated and adapted Dutch version of
the Oslo Sports Trauma Research Center (OSTRC) Ques-
tionnaire on Health Problems was included in the follow-
up questionnaires [21, 22]. The OSTRC questionnaire was
proposed and validated to register and monitor sports-re-
lated health problems over time, i.e., acute injuries, overuse
injuries, and illnesses [23]. The internal consistency
(Cronbach’s a) of the OSTRC questionnaire was estimated
at 0.96 and 0.91 for overall problems (including illnesses)
and overuse injuries, respectively [21, 23].
The OSTRC questionnaire consisted of four key ques-
tions on: (1) the extent to which injury, illness, or other
health problems have affected running participation; (2)
running volume; (3) running performance; and (4) the
extent to which the individual has experienced symptoms
during the previous 2 weeks. If no problems were reported
on these four key questions, the questionnaire was finished.
If a problem was reported on any of the four key questions,
the participant was asked to specify whether the problem
was an illness or an injury. In the case of an illness, the
questionnaire was finished. In case of an injury, partici-
pants were asked to report the anatomical location (one
possible answer per RRI), injury type (one possible answer
per RRI), a description of the symptoms (open question),
injury onset, the number of days of time loss (defined as the
number of training sessions not fully accomplished or
completely missed due to injury), and whether the injury
was related to running. In the case of multiple injuries
within the fortnight, the participants were asked to register
the injury that caused most complaints. Other injuries could
be reported
in an open question. Participants were
instructed to report all problems, regardless of whether or
not they had already reported the same problem in previous
follow-up questionnaires.
2.4 Classification of Health Problems
Health problems were classified as injuries if they were
‘‘disorders of the musculoskeletal system or concussions,’’
and were classified as illnesses if they ‘‘involved other
body systems’’ [21]. One investigator who is also a phys-
iotherapist (LCHJ) evaluated each reported injury case by
case. Injuries were classified as RRI when they were
reported as such by the participants, and when the phys-
iotherapist confirmed that they were experienced or sus-
tained during participation in running. Subsequently, RRIs
were subcategorized into acute (the onset could be linked
to a specific injury event) or overuse injuries (could not be
linked to a clearly identifiable event) [21]. The Orchard
Sports Injury Classification System version 10 (OSICS-10)
[24] was used to provide a diagnostic classification for each
RRI.
Substantial health problems were defined as those
leading to moderate or major reductions in training vol-
ume, moderate or major reductions in running perfor-
mance, or complete inability to run, as identified in the
response options of the key questions 2 or 3 of the OSTRC
questionnaire [21].
A recurrent RRI was defined as an RRI at the same
location and of the same type of the index RRI, even if it
concerned re-injuries (after full recovery) or exacerbations
(not full recovery) [25].
2.5 Economic Consequences of Running-Related
Injuries (RRIs)
Participants who had reported an RRI were asked about
their healthcare utilization (direct costs) and days of pro-
ductivity loss related to paid work (indirect costs) due to
RRIs for the duration of their reporting of symptoms. This
information was collected through conditional branching
questions in the follow-up questionnaires. The cost evalu-
ation was performed from a societal perspective, consid-
ering all RRI-related costs regardless of who pays or
benefits [26]. Table 1 provides the cost categories that were
registered and related monetary costs used in this evalua-
tion. All prices were standardized to the year 2009
according to the Dutch Health Insurance Board [27] and
corrected for inflation until the year 2014 [28]. Costs of
absenteeism from paid work were estimated based on the
mean income [27] and working hours of the Dutch popu-
lation according to age and gender [29].
Health and Economic Burden of Running Injuries in Trailrunners
369
123
2.6 Data Analysis
Microsoft Excel 2011 version 14.5.8 (Microsoft Cor-
poration, Redmond, WA, USA) and R version 3.2.3 (R
Foundation for Statistical Computing, Vienna, Austria)
were used to analyze the data. Descriptive analysis was
performed to present baseline and follow-up data. Per-
centages were calculated for categorical variables. The
mean and its 95 % CI were calculated for continuous data
with Gaussian distribution, otherwise the median and the
25–75 % interquartile range (IQR) were calculated.
2.6.1 Prevalence and Injury Rate Calculations
Prevalence repeatedly measured over time is considered
the preferable measure to describe the overall burden of
injuries in sports involving overuse injuries [16]. The
mean prevalence of RRIs repeatedly measured over time
was calculated according to previous recommendations
[16, 21, 23]. For each 2-week period, the prevalence was
calculated by dividing the number of participants report-
ing RRIs during that period by the number of total
questionnaire
respondents
during
the
same
period.
Thereafter, the mean prevalence and its 95 % CI were
calculated by summing all prevalences measured every
2 weeks, divided by the number of 2-week time-periods.
The injury rate was calculated by dividing the number of
RRIs by the sum of total running exposure in hours [18,
30]. The number of RRIs was calculated based on the
number of unique RRIs identified during the follow-up.
Results were expressed as the number of RRIs per 1000 h
of running and its 95 % CI.
2.6.2 Severity
In order to monitor the progress of the RRIs over time, a
severity score ranging from 0 to 100 was calculated for each
RRI based on the response options of the four key questions
of the OSTRC questionnaire [21]. Average severity scores
were calculated by taking the mean of the severity scores
measured every 2 weeks for each RRI. The cumulative
severity score (sum of the severity scores measured every
2 weeks) was calculated as an estimation of the total impact
that each RRI had had over the course of the study. The
average and cumulative time loss were also calculated for
each RRI as the same manner as the severity score.
2.6.3 Costs
Mean direct, indirect, and total costs were estimated per
RRI, per 1000 h of running, and per most commonly
reported RRIs. The participants could present more than
one RRI during the study, resulting in dependent obser-
vations. Therefore, the difference in costs between overuse
and acute RRIs were estimated using linear mixed models
with random intercept at the participant level, adjusted for
the following possible confounders measured at baseline:
age, gender, body mass index (BMI), running experience,
practice of other sports, chronic condition, medication use,
current RRIs, and previous RRIs. As the cost per 1000 h of
running is a rate between cumulative measures at the
population level (i.e., sum of costs divided by the sum of
total running exposure in hours multiplied by 1000),
adjustment for possible confounders was not possible. Cost
data are nonparametric, therefore, 95 % CIs were obtained
by bootstrapping the data with 2000 replications [31–33],
as recommended for economic evaluations [26].
3 Results
3.1 Participants, Response Rate, and Running
Exposure
A total of 228 trailrunners, 171 males (75.0 %) and 57
females (25.0 %), were included in the study. The baseline
results are summarized in Table 2. Five male participants
entered no data in the follow-up questionnaires, corre-
sponding to an attrition rate of 2.2 %. As the participants
entered in the study in different time-points, they had dif-
ferent follow-up periods. However, all participants were
followed for at least 6 months. The median of the follow-
up period was 34.0 weeks (IQR 28.0–36.0), and the
response rate measured every 2 weeks was 77.3 % (IQR
57.6–88.1). The median and IQR for the weekly running
exposure can be found in Table 3. On average, 22.8 %
Table 1 Monetary costs applied in the cost analysis
Description
Cost, €
Healthcare costs (direct costs)
General practitioner (per visit, 10 min)
30.79
General practitioner (per telephone
consultation)
15.40
Medical specialist (per visit)
79.17
Physiotherapist (per visit)
39.59
Costs of productivity loss (indirect costs)
Absenteeism from paid work (per hour)*
31.22 (9.78–43.95)
Prices standardized to the year 2009 according to the Dutch Health
Insurance Board [27] and adjusted for inflation until the year 2014
[28]
* Indirect costs for paid work were estimated based on the mean
income [27] and working hours [29] of the Dutch population
according to age and gender. The value for paid work is the mean
price followed by the minimum and maximal values according to
standardized prices by age and gender, adjusted for inflation [28]
370
L. C. H. Junior et al.
123
(95 % CI 20.1–25.6) of the trailrunners participated in
trailrunning events every 2 weeks. The median of the
distance of the trailrunning events was 28.0 km (IQR
17.5–39.1),
ranging
from
3
(minimum)
to
230 km
(maximum).
3.2 Prevalence, Injury Rate, Severity, and Nature
of RRIs
The absolute number, prevalence, injury rate, and severity
measures of RRIs can be found in Table 4. A total of 148
participants (66.4 %) reported 242 RRIs during the follow-
up. Of the injured participants, 68 (45.9 %) reported mul-
tiple RRIs (i.e., different OSICS-10 diagnostic classifica-
tions). The percentage of injured participants who reported
other RRIs within the 2-week time-period was 4.7 % (IQR
4.0–7.2).
The mean prevalence of RRIs measured every 2 weeks
was 22.4 % (95 % CI 20.9–24.0). For males, the mean
prevalence of RRIs was 23.0 % (95 % CI 21.3–24.7), and
for females this was 20.7 % (95 % CI 18.2–23.2), with a
mean difference of 2.3 percentage points (95 % CI -1.0 to
5.6). The mean prevalence of RRIs was higher for overuse
than for acute RRIs, with a mean difference of 13.6 per-
centage points (95 % CI 10.3 to 16.9).
The injury rate was 10.7 RRIs per 1000 h of running
(95 % CI 9.4–12.1). For males, the injury rate was 11.3
(95 % CI 9.7–12.9), and for females this was 9.1 (95 % CI
6.6–11.6), with an injury rate difference of 2.2 RRIs per
1000 h of running (95 % CI -0.7 to 5.1). The injury rate
Table 2 Baseline data of the participants
All participants
n = 228
Male
n = 171
Female
n = 57
Age, years
43.4 (42.2–44.6)
43.8 (42.4–45.2)
42.4 (39.9–44.8)
Height, cm
178.9 (177.8–180.1)
182.4 (181.4–183.4)
168.4 (166.8–170.0)
Weight, kg
72.5 (71.1–74.0)
76.5 (75.2–77.9)
60.6 (58.9–62.2)
BMI, kg/m2
22.6 (22.3–22.8)
23.0 (22.7–23.3)
21.3 (20.9–21.8)
Total running experience, n (%)
Up to 1 year
7 (3.1 %)
7 (4.1 %)
–
1–2 years
18 (7.9 %)
13 (7.6 %)
5 (8.8 %)
2–5 years
43 (18.9 %)
35 (20.5 %)
8 (14.0 %)
More than 5 years
160 (70.2 %)
116 (67.8 %)
44 (77.2 %)
Trailrunning experience, n (%)
Up to 6 months
22 (9.6 %)
16 (9.4 %)
6 (10.5 %)
6–12 months
38 (16.7 %)
31 (18.1 %)
7 (12.3 %)
1–2 years
59 (25.9 %)
38 (22.2 %)
21 (36.8 %)
2–5 years
71 (31.1 %)
56 (32.7 %)
15 (26.3 %)
More than 5 years
38 (16.7 %)
30 (17.5 %)
8 (14.0 %)
Practice of other sports, n (%)
Yes
152 (66.7 %)
111 (64.9 %)
41 (71.9 %)
No
76 (33.3 %)
60 (35.1 %)
16 (28.1 %)
Chronic condition, n (%)
Yes
40 (17.5 %)
27 (15.8 %)
13 (22.8 %)
No
188 (82.5 %)
144 (84.2 %)
44 (77.2 %)
Current medication use, n (%)
Yes
26 (11.4 %)
16 (9.4 %)
10 (17.5 %)
No
202 (88.6 %)
155 (90.6 %)
47 (82.5 %)
Current RRI, n (%)
Yes
41 (18.0 %)
33 (19.3 %)
8 (14.0 %)
No
187 (82.0 %)
138 (80.7 %)
49 (86.0 %)
Previous RRI (last 12 months), n (%)
Yes
96 (42.1 %)
71 (41.5 %)
25 (43.9 %)
No
132 (57.9 %)
100 (58.5 %)
32 (56.1 %)
Continuous data are given as mean and 95 % confidence interval
BMI body mass index, RRI running-related injury
Health and Economic Burden of Running Injuries in Trailrunners
371
123
was higher for overuse than for acute RRIs, with an injury
rate difference of 5.4 RRIs per 1000 h of running (95 % CI
4.1 to 6.8).
A total of 54.1 % (n = 131) of the RRIs were classified
as substantial (i.e., leading to moderate or major reductions
in training volume, moderate or major reductions in run-
ning performance, or complete inability to run). Fifty-nine
RRIs (24.4 %) neither resulted in time loss nor in medical
attention. Overuse RRIs lasted longer and presented a
higher
cumulative
severity
score
than
acute
RRIs
(Table 4). The most commonly reported RRIs were
Achilles tendon injury (12.8 %, n = 31), calf muscle
injury (10.7 %, n = 26), knee pain undiagnosed (8.7 %,
n = 21), and ankle sprain (7.0 %, n = 17). A breakdown
list with all RRIs reported during this study can be found in
the Electronic Supplementary Material.
Table 3 Running exposure during the follow-up
All participants
n = 223
Male
n = 166
Female
n = 57
Total running exposure
Duration (h/week)
3.5 (2.0–5.0)
3.5 (2.0–5.0)
3.5 (2.0–5.3)
Frequency (times/week)
2.5 (1.5–3.5)
2.5 (1.5–3.5)
2.5 (2.0–3.5)
Distance (km/week)
33.6 (19.5–50.0)
35.0 (20.0–50.0)
32.5 (17.5–50.0)
Running exposure on unpaved surfaces
Duration (h/week)
1.5 (0.5–3.0)
1.5 (0.5–2.8)
1.8 (0.8–3.0)
Frequency (times/week)
1.0 (0.5–2.0)
1.0 (0.5–2.0)
1.5 (0.5–2.0)
Distance (km/week)
15.0 (6.0–28.0)
15.0 (6.0–27.5)
16.0 (7.5–30.0)
Results are given as median and 25–75 % interquartile range (IQR)
Table 4 Absolute number, mean prevalence measured over time (every 2 weeks), injury rate, and severity measures of running-related injuries
(RRIs)
RRIs
Total
Overuse
Acute
Time loss
Medical attention
Overall
Number of RRIs registered
n = 242
n = 182
n = 60
n = 174
n = 72
Prevalence, mean (95 % CI)
22.4 % (20.9–24.0)
17.7 % (15.9–19.5)
4.1 % (3.3–5.0)
15.1 % (14.0–16.2)
5.9 % (5.1–6.7)
Injury rate, number of RRIs per
1000 h of running (95 % CI)
10.7 (9.4–12.1)
8.1 (6.9–9.3)
2.7 (2.0–3.4)
7.7 (6.6–8.9)
3.2 (2.4–3.9)
Severity measures, median (IQR)
Average severity score
35.0 (22.0–55.7)
31.1 (20.0–55.0)
37.0 (28.0–57.2)
43.0 (28.6–63.0)
55.0 (34.5–70.2)
Cumulative severity score
55.5 (28.0–122.0)
63.0 (25.2–122.0)
50.0 (33.8–116.0)
78.0 (37.0–165.0)
132.0 (66.0–278.0)
Average time loss, days
2.0 (0.0–4.7)
2.0 (0.0–4.5)
2.8 (1.0–5.1)
3.3 (1.8–6.0)
4.0 (1.5–7.3)
Cumulative time loss, days
3.0 (0.0–10.0)
3.0 (0.0–10.0)
3.5 (1.0–8.0)
5.0 (3.0–15.5)
12.0 (3.0–28.2)
Duration, weeks
2.0 (2.0–6.0)
4.0 (2.0–6.0)
2.0 (2.0–4.0)
4.0 (2.0–6.0)
6.0 (3.5–10.0)
Substantial
Number of RRIs registered
n = 131
n = 94
n = 37
n = 120
n = 58
Prevalence, mean (95 % CI)
9.9 % (9.1–10.8)
7.3 % (6.5–8.0)
2.3 % (1.4–3.1)
9.4 % (8.6–10.2)
3.7 % (3.1–4.3)
Injury rate, number of RRIs per
1000 h of running (95 % CI)
5.8 (4.8–6.8)
4.2 (3.3–5.0)
1.6 (1.1–2.2)
5.3 (4.4–6.3)
2.6 (1.9–3.3)
Severity measures, median (IQR)
Average severity score
54.5 (39.9–68.3)
54.5 (39.7–68.8)
51.0 (41.2–67.3)
54.8 (41.1–69.2)
59.6 (44.1–76.6)
Cumulative severity score
109.0 (66.0–198)
113.0 (71.2–230.5)
80.0 (50.0–159.0)
117.5 (66.0–226.0)
168.0 (80.0–287.2)
Average time loss, days
4.0 (2.0–6.8)
4.0 (2.0–6.9)
4.0 (2.0–6.0)
4.2 (2.8–7.0)
5.0 (3.0–8.3)
Cumulative time loss, days
7.0 (4.0–20.0)
8.5 (4.0–23.8)
5.0 (3.0–16.0)
9.5 (4.0–21.5)
14.0 (4.0–31.5)
Duration, weeks
4.0 (2.0–8.0)
5.0 (2.0–9.5)
4.0 (2.0–6.0)
4.0 (2.0–8.0)
6.0 (4.0–10.0)
Substantial RRIs were defined as those leading to moderate or major reductions in training volume, moderate or major reductions in running
performance, or complete inability to run
95 % CI 95 % confidence interval, IQR 25–75 % interquartile range
372
L. C. H. Junior et al.
123
3.3 Economic Burden of RRIs
In total, 332 healthcare consultations (21 general practi-
tioner, 47 medical specialist, and 264 physiotherapy con-
sultations) and 102 days of productivity loss related to paid
work were registered. A total (direct plus indirect) cost of
€41,677.13 was calculated for the 242 RRIs. The direct
cost was €14,742.39 (€569.64 related to general practi-
tioner,
€3720.99
related
to
medical
specialist
and
€10,451.76 related to physiotherapy consultations) and the
indirect cost was €26,934.74 (related to absenteeism from
paid work).
The costs per RRI, per 1000 h of running and per most
commonly reported RRIs can be found in Table 5. Overuse
RRIs presented higher physiotherapy costs than acute
RRIs, and acute RRIs presented higher costs related to
general practitioner than overuse RRIs. There were no
statistically significant differences in costs per 1000 h of
running between males and females. Of the four most
commonly reported RRIs, calf muscle injuries presented
the highest direct and indirect costs.
4 Discussion
4.1 Trailrunners and Running Exposure
The sample of the current study was composed by Dutch
trailrunners who were recruited during trailrunning events,
or through trailrunning channels, like the MST website
[20], regardless of age, gender, running experience, com-
petition level, or training exposure (e.g., volume and
intensity). As presented in Table 3, Dutch trailrunners
usually train on paved and unpaved tracks. This could be
explained by the fact that most Dutch trailrunners live in
city areas, and, therefore, they do not have easy and fast
access to trail tracks that usually are composed by rugged,
muddy, and/or mountain terrains. However, trailrunners
need to train on a regular basis to be prepared for the
trailrunning events that usually have longer distances
(median of 28 km in the current study). Therefore, the
sample of trailrunners in the current study can be consid-
ered representative of the general Dutch trailrunning pop-
ulation. Furthermore, the characteristics of the Dutch
trailrunners who participated in this study may also be
similar to recreational trailrunners in other countries.
4.2 Prevalence and Injury Rate of RRIs
The results of this study have shown that the mean
prevalence of RRIs measured every 2 weeks is between
20.9 and 24.0 % (95 % CI) in trailrunners. In other words,
one out of five trailrunners may be expected to sustain
RRIs during a 2-week time-period. The prevalence esti-
mates of this study are not comparable with other studies in
the literature, since this is the first study to report the
prevalence of RRIs repeatedly measured over time in
trailrunners. In addition, previous studies on trailrunning
have used different methods and RRI definitions [34, 35].
This hampers comparisons. For example, the incidence
proportion of lower limb musculoskeletal injuries (22.2 %)
found during the Al Andalus Ultimate Trail 2010 held in
southern Spain [35] was similar to the prevalence repeat-
edly measured over time reported in the current study,
although these are two different measures.
Hespanhol Junior et al. [15] have used similar methods
as the one used in the current study to investigate RRIs in
inexperienced runners training for an event. The study
design, surveillance system, RRI definition and RRI clas-
sifications were the same in both studies, although the
population and the follow-up period were different. The
mean prevalence of all RRIs and the mean prevalence of
overuse RRIs found in the current study were lower than
the mean prevalences reported by Hespanhol Junior et al.
[15]. This may be explained by differences in running
experience [11] and training volume [36] between these
two populations.
As explained in the methods, a priori sample size cal-
culation was not possible because of missing information
on the prevalence of RRIs repeatedly measured over time
in a general trailrunning population at the commencement
of this study. However, the study of Hespanhol Junior et al.
[15] was recently available. Therefore, a post hoc sample
size calculation based on the results reported in Hespanhol
Junior et al. [15] and the results of the current study was
possible. The sample size was estimated based on calcu-
lations for longitudinal studies with repeated measurements
[37]. The prevalence of RRIs repeatedly measured over
time in the study of Hespanhol Junior et al. [15] was
30.8 % (95 % CI 25.6–36.0), and in the current study was
22.4 %
(95 %
CI
20.9–24.0).
Considering
a = 0.05,
b = 0.8, 17 repeated measurements (i.e., median of
34 weeks of follow-up with repeated measurements every
2 weeks), a correlation coefficient of the repeated mea-
surements of 0.24 (calculated in the current study for the
purpose of this sample size calculation), and a response
rate of 77.3 % (reported in the current study), the sample
size calculation suggested a cohort of 152 participants.
Based on this calculation, the sample size of the current
study was appropriate. This calculation may be useful as a
reference for sample size calculations for future longitu-
dinal studies with repeated measurements on RRIs.
Comparisons of injury rates of RRIs across studies are
difficult because of differences in RRI definitions [11, 17,
19]. However, the time loss injury rate in trailrunners found
in the current study [7.7 RRIs per 1000 h (95 % CI
Health and Economic Burden of Running Injuries in Trailrunners
373
123
Table 5 Economic burden of running-related injuries (RRIs) in trailrunners
Overall
Direct cost
Indirect cost
Total
General
practitioner
Medical specialist
Physiotherapy
Absenteeism from
paid work
Cost per RRI, €
All injuries, n = 242
172.22 (117.10 to 271.74)
60.92 (45.11 to 94.90)
2.35 (1.08 to
4.14)
15.38 (8.51 to
27.41)
43.19 (30.96 to 60.20)
111.30 (61.02 to
192.75)
Overuse, n = 182
174.40 (108.52 to 302.65)
69.96 (48.18 to 102.90)
1.44 (0.51 to
3.54)
17.84 (9.57 to
31.88)
50.68 (35.02 to 72.00)
104.44 (50.88 to
205.16)
Acute, n = 60
165.61 (78.19 to 363.54)
33.50 (18.55 to 55.33)
5.13 (2.05 to
10.52)
7.92 (1.32 to
22.43)
20.45 (9.90 to 41.57)
132.11 (44.36 to
301.15)
Difference (overuse
minus acute)
16.60 (-161.98 to 179.61)
31.05 (16.31 to 76.70)*
-3.85 (-10.42 to
-3.66)*
3.79 (-14.19 to
23.45)
27.43 (23.68 to 50.98)*
-14.88 (-210.93 to
99.40)
Cost per 1000 h of running, €
All participants,
n = 223
1849.49 (1180.62 to 3058.91)
654.22 (465.82 to 942.68)
25.28 (11.61 to
44.41)
165.13 (91.35 to
291.60)
463.81 (323.26 to 686.48)
1195.27 (635.76 to
2346.03)
Male, n = 166
1783.44 (1013.66 to 3321.43)
548.35 (353.32 to 966.17)
19.46 (7.41 to
37.99)
142.93 (57.17 to
290.63)
385.96 (238.25 to 631.36)
1235.09 (582.51 to
2622.02)
Female, n = 57
2034.97 (1032.15 to 4630.62)
951.52 (621.73 to 1630.91)
41.63 (10.41 to
104.08)
227.45 (80.28 to
454.90)
682.44 (398.56 to 1137.39)
1083.45 (279.82 to
3134.15)
Difference (males
minus females)
-251.52 (-2671.36 to 1380.54)
-403.17 (-1107.63 to 118.15)
-22.17 (-92.86
to 10.32)
-84.52 (-339.59
to 109.56)
-296.47 (-801.50 to 50.16)
151.64 (-1751.25
to 1485.12)
Cost per most commonly reported RRIs, €
Achilles tendon
injury, n = 31
67.60 (30.08 to 148.10)
46.68 (19.16 to 111.10)
1.99 (0.00 to
12.32)
10.22 (0.00 to
53.23)
34.48 (13.68 to 69.77)
20.91 (0.00 to
128.43)
Calf muscle injury,
n = 26
135.85 (49.22 to 391.91)
56.00 (23.19 to 123.85)
1.18 (0.00 to
6.58)
21.32 (5.66 to
63.18)
33.50 (11.50 to 80.89)
79.86 (0.00 to
384.14)
Knee pain
undiagnosed, n = 21
13.20 (2.83 to 33.33)
13.20 (3.05 to 31.67)
–
–
13.20 (2.92 to 33.73)
–
Ankle sprain, n = 17
82.20 (8.48 to 346.78)
20.96 (5.16 to 50.47)
–
–
20.96 (5.28 to 53.78)
61.24 (0.00 to
319.48)
All costs are presented in euros (€). Mean values are followed by the bias-corrected and accelerated 95 % confidence interval estimated by bootstrapping (2000 replications)
* Significant difference between overuse and acute RRIs
The difference in costs between overuse and acute RRIs were estimated using linear mixed models with random intercept at the participant level, adjusted for the following possible
confounders measured at baseline: age, gender, body mass index (BMI), running experience, practice of other sports, chronic condition, medication use, current RRIs, and previous RRIs
374
L. C. H. Junior et al.
123
6.6–8.9)] was similar to the injury rate in recreational
runners reported by Videbaek et al. [7.7 RRIs per 1000 h
(95 % CI 6.9–8.7)] [11], that was summarized based on
studies with time loss RRI definitions.
According to the literature, overuse RRIs occur more
frequently than acute RRIs [12, 15]. The results of the
current study support this observation for trailrunning,
since the prevalence of overuse RRIs was fourfold higher
than acute RRIs, and the injury rate of overuse RRIs was
threefold higher than acute RRIs. Most of the time, running
can be described as an aerobic physical activity that
requires long duration exertion with few changes in
movement patterns. Therefore, overuse injuries with a
gradual onset mechanism resulting from repetitive micro-
trauma would be more expected in trailrunning than inju-
ries with a sudden onset.
4.3 Severity of RRIs
Severity measures are important to understand the extent to
which sports injuries affect health [38]. A strength of this
study was the continuous and valid method used to monitor
the severity of sports injuries, irrespective of time loss or
medical attention [21]. In fact, 24.4 % of the RRIs reported
in this study neither resulted in time loss nor medical
attention. Therefore, the results of this study support the
hypothesis that measuring RRIs based only on time loss or
medical attention definitions will lead to an underestima-
tion of the burden of RRIs.
The longer duration of overuse RRIs can explain why
the cumulative severity score was higher for overuse than
for acute RRIs. More than half of the RRIs were classified
as substantial, meaning that they caused a moderate or
major reduction in running volume or running perfor-
mance, or had caused a complete inability to participate in
running. This result supports the hypothesis that RRIs may
reach such severity levels that they can lead to dropping out
of running participation [14, 15]. The implication is that
RRIs may lower the motivation to participate in running, a
great ally against the burden of physical inactivity, which is
a leading risk factor for the global disease burden [39] and
mortality [40]. In fact, running is effective in reducing
mortality and disability [8, 9]; however, the adherence to
running participation is essential to reach such health
benefits [9, 10].
4.4 Economic Burden of RRIs
To the best of our knowledge, this is the first study
reporting the total, direct, and indirect costs of RRIs in
trailrunners. The cost per RRI in trailrunners found in the
current study was €172.22 (95 % CI 117.10–271.74),
which was comparable to the cost per RRI found in
runners
training
for
an
event
(€173.72;
95 %
CI
57.17–318.76) [15] and higher than the costs per RRI
found in novice runners (€83.22; 95 % CI 50.42–116.02)
[41]. These cost estimates are lower than the economic
burden generally reported for sports injuries in other
athletic populations [42, 43]. However, comparisons
with other sports and populations should be made with
caution where the study methods and follow-up periods
were different.
Healthcare consultations related to RRIs were threefold
higher than the number of days of productivity loss related
to paid work. However, the indirect cost of RRIs was
twofold higher than the direct cost. Interestingly, the
indirect-direct cost ratio was higher for acute RRIs (indi-
rect cost fourfold higher than direct cost) than for overuse
RRIs (indirect cost 1.5-fold higher than direct cost), indi-
cating that the productivity loss impact may be higher for
acute RRIs. Other studies have also shown higher indirect
than direct costs related to sports injuries [43–47]. These
results indicate that productivity loss is the main contrib-
utor to the economic burden of sports injuries, with a sig-
nificant impact on societal financial resources. As such,
policymakers should always take into account the direct
and especially the indirect costs of sports injuries to drive
their policies.
To put our results into perspective: according to MST,
7500 people participate in trailrunning events organized by
them each year. Based on the results of the current study,
one trailrunner runs approximately 3.5 h per week (i.e.,
182 h per year). Therefore, one could expect to have a total
cost related to RRIs of more than €2.5 million yearly, only
accounting for trailrunners participating in the MST events.
This figure represents around 0.4 % of all annual sports
injury costs in The Netherlands [47]. Although not a large
proportion, if RRIs in trailrunning are prevented, maybe
hundreds of thousands of euros could be saved and redi-
rected to other public health areas. This assumption shows
the financial impact that RRIs in trailrunning could have
for society.
There is sound evidence showing that physical activity
is a cost-effective method to improve overall health, and
gain healthy life-years [1–4]. Evidence also suggests that
the health benefits of running outweigh the related risks
and costs [4, 8–10]. Therefore, running may be advised for
people who seek to improve their health by means of
engaging in strenuous physical activity. Nonetheless, RRIs
are a preventable side effect of such active engagement and
prevention is warranted. Effective prevention of injuries
will not only reduce the individual burden in terms of
injury and costs, but will also improve joyful and contin-
uing participation in running.
Health and Economic Burden of Running Injuries in Trailrunners
375
123
4.5 Limitations
This study was composed of a convenience sample. As
presented in Table 4, most RRIs reported in the current
study were overuse injuries, i.e., those that have a non-
identifiable and gradual onset, and also present fluctuation
of symptoms over time. Consequently, the RRIs reported in
the current study represent all RRIs that could be a result of
running exposure on paved, unpaved, or both surfaces (the
most likely assumption). The RRIs were self-reported and
then classified by a healthcare professional (LCHJ) based
on the RRI description given by the participants. A con-
firmation of the RRI diagnoses during face-to-face con-
sultations was not possible due to logistic reasons. Data
about medicines taken and diagnostic tests due to RRIs
were not collected. This could have lead to an underesti-
mation of the direct costs of RRIs. The cost analysis was an
estimation based on Dutch standardized prices for health-
care utilization [27], and the mean income [27] and
working hours of the Dutch population for absenteeism
from paid work [29], all adjusted for inflation [28]. Despite
the fact that this methodology has been accepted and rec-
ommended [26], it is important to realize that the cost
results were estimated and do not represent actual costs.
5 Conclusions
On average, one out of five trailrunners reported RRIs every
2 weeks. Overuse RRIs represented 75.2 % of all RRIs
registered during the follow-up. A total of 54.1 % of all
RRIs were classified as substantial. The economic burden
(direct plus indirect costs) of RRIs was estimated at €172.22
(95 % CI 117.10–271.74) per RRI, and €1849.49 (95 % CI
1180.62–3058.91) per 1000 h of running. Healthcare uti-
lization (direct costs) contributed to 35.4 % of these costs
and absenteeism from paid work (indirect costs) to 64.6 %.
Acknowledgments Luiz Carlos Hespanhol Junior is a PhD candidate
supported by CAPES (Coordenac¸a˜o de Aperfeic¸oamento de Pessoal
de Nı´vel Superior), process number 0763/12-8, Ministry of Education
of Brazil. The authors wish to thank MudSweatTrails and Marc
Weening for their assistance during the recruitment, and all trail-
runners who participated in this study.
Compliance with Ethical Standards
Funding This study had no funding sources.
Conflict of interest Luiz Carlos Hespanhol Junior and Evert Ver-
hagen declare that they have no conflicts of interest. Willem van
Mechelen declares that he is director-shareholder of VU University
Medical Center spin-off company Evalua Nederland B.V. (http://
www.evalua.nl), and non-executive board member of Arbo Unie B.V.
(http://www.arbounie.nl). Both companies operate on the Dutch
Occupational Health Care market.
Ethical approval All procedures performed in studies involving
human participants were in accordance with the ethical standards of
the institutional and/or national research committee and with the 1964
Helsinki Declaration and its later amendments or comparable ethical
standards.
Informed consent Informed consent was obtained from all individ-
ual participants included in the study.
Open Access This article is distributed under the terms of the
Creative Commons Attribution 4.0 International License (http://
creativecommons.org/licenses/by/4.0/), which permits unrestricted use,
distribution, and reproduction in any medium, provided you give
appropriate credit to the original author(s) and the source, provide a link
to the Creative Commons license, and indicate if changes were made.
References
1. Codogno JS, Turi BC, Kemper HC, et al. Physical inactivity of
adults and 1-year health care expenditures in Brazil. Int J
Public Health. 2015;60(3):309–16. doi:10.1007/s00038-015-
0657-z.
2. Frew EJ, Bhatti M, Win K, et al. Cost-effectiveness of a com-
munity-based physical activity programme for adults (Be Active)
in the UK: an economic analysis within a natural experiment. Br J
Sports
Med.
2014;48(3):207–12.
doi:10.1136/bjsports-2012-
091202.
3. Hagberg LA, Lindholm L. Cost-effectiveness of healthcare-based
interventions aimed at improving physical activity. Scand J
Public Health. 2006;34(6):641–53. doi:10.1080/14034940600
627853.
4. Hatziandreu EI, Koplan JP, Weinstein MC, et al. A cost-effec-
tiveness analysis of exercise as a health promotion activity. Am J
Public Health. 1988;78(11):1417–21.
5. Thompson Coon J, Boddy K, Stein K, et al. Does participating in
physical activity in outdoor natural environments have a greater
effect on physical and mental wellbeing than physical activity
indoors?
A
systematic
review.
Environ
Sci
Technol.
2011;45(5):1761–72. doi:10.1021/es102947t.
6. Stamatakis E, Chaudhury M. Temporal trends in adults’ sports
participation patterns in England between 1997 and 2006: the
Health Survey for England. Br J Sports Med. 2008;42(11):901–8.
doi:10.1136/bjsm.2008.048082.
7. Ottesen L, Jeppesen RS, Krustrup BR. The development of social
capital through football and running: studying an intervention
program
for
inactive
women.
Scand
J
Med
Sci
Sports.
2010;20(Suppl 1):118–31. doi:10.1111/j.1600-0838.2010.01123.x.
8. Chakravarty EF, Hubert HB, Lingala VB, et al. Reduced dis-
ability and mortality among aging runners: a 21-year longitudinal
study. Arch Intern Med. 2008;168(15):1638–46. doi:10.1001/
archinte.168.15.1638.
9. Lee DC, Pate RR, Lavie CJ, et al. Leisure-time running reduces
all-cause and cardiovascular mortality risk. J Am Coll Cardiol.
2014;64(5):472–81. doi:10.1016/j.jacc.2014.04.058.
10. Hespanhol Junior LC, Pillay JD, van Mechelen W, et al. Meta-
analyses of the effects of habitual running on indices of health in
physically inactive adults. Sports Med. 2015;45(10):1455–68.
doi:10.1007/s40279-015-0359-y.
11. Videbaek S, Bueno AM, Nielsen RO, et al. Incidence of running-
related injuries per 1000 h of running in different types of run-
ners: a systematic review and meta-analysis. Sports Med.
2015;45(7):1017–26. doi:10.1007/s40279-015-0333-8.
12. Lopes AD, Hespanhol Junior LC, Yeung SS, et al. What are the
main running-related musculoskeletal injuries?: a systematic
376
L. C. H. Junior et al.
123
review. Sports Med. 2012;42(10):891–905. doi:10.2165/116311
70-000000000-00000.
13. Nielsen RO, Ronnow L, Rasmussen S, et al. A prospective study
on time to recovery in 254 injured novice runners. PLoS One.
2014;9(6):e99877. doi:10.1371/journal.pone.0099877.
14. Kluitenberg B, van Middelkoop M, Diercks RL, et al. The
NLstart2run study: health effects of a running promotion program
in novice runners, design of a prospective cohort study. BMC
Public Health. 2013;13:685. doi:10.1186/1471-2458-13-685.
15. Hespanhol Junior LC, van Mechelen W, Postuma E et al. Health
and economic burden of running-related injuries in runners
training for an event: a prospective cohort study. Scand J Med Sci
Sports. 2015. doi:10.1111/sms.12541.
16. Bahr R. No injuries, but plenty of pain? On the methodology for
recording overuse symptoms in sports. Br J Sports Med.
2009;43(13):966–72. doi:10.1136/bjsm.2009.066936.
17. Yamato TP, Saragiotto BT, Hespanhol Junior LC, et al.
Descriptors
used
to
define
running-related
musculoskeletal
injury: a systematic review. J Orthop Sports Phys Ther.
2015;45(5):366–74. doi:10.2519/jospt.2015.5750.
18. Hespanhol Junior LC, Barboza SD, van Mechelen W, et al.
Measuring sports injuries on the pitch: a guide to use in practice.
Braz J Phys Ther. 2015;19(5):369–80. doi:10.1590/bjpt-rbf.2014.
0110.
19. Kluitenberg B, van Middelkoop M, Verhagen E, et al. The impact
of injury definition on injury surveillance in novice runners. J Sci
Med Sport. 2015;. doi:10.1016/j.jsams.2015.07.003.
20. MudSweatTrails.
Home
page.
Available
at:
http://www.
mudsweattrails.nl. Accessed in 07 March 2016.
21. Clarsen B, Ronsen O, Myklebust G, et al. The Oslo Sports
Trauma Research Center questionnaire on health problems: a new
approach to prospective monitoring of illness and injury in elite
athletes. Br J Sports Med. 2014;48(9):754–60. doi:10.1136/
bjsports-2012-092087.
22. Pluim BM, Loeffen FG, Clarsen B, et al. A one-season
prospective study of injuries and illness in elite junior tennis.
Scand J Med Sci Sports. 2015;. doi:10.1111/sms.12471.
23. Clarsen B, Myklebust G, Bahr R. Development and validation of
a new method for the registration of overuse injuries in sports
injury epidemiology: the Oslo Sports Trauma Research Centre
(OSTRC) overuse injury questionnaire. Br J Sports Med.
2013;47(8):495–502. doi:10.1136/bjsports-2012-091524.
24. Rae K, Orchard J. The Orchard Sports Injury Classification
System (OSICS) version 10. Clin J Sport Med. 2007;17(3):201–4.
doi:10.1097/JSM.0b013e318059b536.
25. Fuller CW, Bahr R, Dick RW, et al. A framework for recording
recurrences, reinjuries, and exacerbations in injury surveillance.
Clin
J
Sport
Med.
2007;17(3):197–200.
doi:10.1097/JSM.
0b013e3180471b89.
26. van Dongen JM, van Wier MF, Tompa E, et al. Trial-based
economic evaluations in occupational health: principles, methods,
and recommendations. J Occup Environ Med. 2014;56(6):
563–72. doi:10.1097/JOM.0000000000000165.
27. van Roijen LH, Tan SS, Bouwmans C. Handleiding voor
kostenonderzoek,
methoden
en
standaard
kostprijzen
voor
economische evaluaties in de gezondheidszorg. College voor
Zorgverzekeringen; 2010.
28. Centraal
Bureau
voor
de
Statistiek.
CBS
StatLine:
Con-
sumentenprijzen; prijsindex 1900=100. Available at: http://
opendata.cbs.nl/Dataportaal/index.html?_la=nl&_catalog=CBS&_
si=&_gu=&_ed=Topics&_td=Perioden&tableId=71905ned&$filter
=(substringof(‘JJ’%2CPerioden))&$select=Perioden%2CInflatie_
2&graphType=bar. Accessed 11 Dec 2015.
29. Centraal Bureau voor de Statistiek. Arbeidsdeelname; kerncijfers
(12-uursgrens).
Available
at:
http://statline.cbs.nl/Statweb/
publication/?DM=SLNL&PA=71738ned&D1=8-10&D2=a&D3=
a&D4=0&D5=l&VW=T. Accessed 11 Dec 2015.
30. Knowles SB, Marshall SW, Guskiewicz KM. Issues in estimating
risks
and
rates
in
sports
injury
research.
J
Athl
Train.
2006;41(2):207–15.
31. Chaudhary MA, Stearns SC. Estimating confidence intervals for
cost-effectiveness ratios: an example from a randomized trial.
Stat
Med.
1996;15(13):1447-58.
doi:10.1002/(SICI)1097-
0258(19960715)15:13\1447::AID-SIM267[3.0.CO;2-V
32. Efron B. Better bootstrap confidence intervals. J Am Stat Assoc.
1987;82(397):171–85. doi:10.1080/01621459.1987.10478410.
33. Efron B, Tibshirani R. Bootstrap methods for standard errors,
confidence intervals, and other measures of statistical accuracy.
Stat Sci. 1986;1(1):54–75. doi:10.2307/2245500.
34. Hoffman MD, Fogard K. Factors related to successful completion
of a 161-km ultramarathon. Int J Sports Physiol Perform.
2011;6(1):25–37.
35. Scheer BV, Murray A. Al Andalus ultra trail: an observation of
medical interventions during a 219-km, 5-day ultramarathon
stage race. Clin J Sport Med. 2011;21(5):444–6.
36. Gabbett TJ. The training-injury prevention paradox: should ath-
letes be training smarter and harder? Br J Sports Med.
2016;50(5):273–80. doi:10.1136/bjsports-2015-095788.
37. Twisk JWR. Sample size calculations. In: Twisk JWR, editor.
Applied longitudinal data analysis for epidemiology. 2 ed.
Cambridge: Cambridge University Press; 2013. p. 237–42.
38. van Mechelen W. The severity of sports injuries. Sports Med.
1997;24(3):176–80. doi:10.2165/00007256-199724030-00006.
39. Lim SS, Vos T, Flaxman AD, et al. A comparative risk assess-
ment of burden of disease and injury attributable to 67 risk factors
and risk factor clusters in 21 regions, 1990–2010: a systematic
analysis for the Global Burden of Disease Study 2010. Lancet.
2012;380(9859):2224–60. doi:10.1016/S0140-6736(12)61766-8.
40. WHO. Global health risks: mortality and burden of disease
attributable to selected major risks. Geneva: World Health
Organization; 2009.
41. Hespanhol Junior LC, Huisstede BM, Smits DW, et al. The
NLstart2run study: economic burden of running-related injuries
in novice runners participating in a novice running program. J Sci
Med Sport. 2015;. doi:10.1016/j.jsams.2015.12.004.
42. King DA, Hume PA, Milburn P, et al. Rugby league injuries in
New Zealand: a review of 8 years of accident compensation
corporation injury entitlement claims and costs. Br J Sports Med.
2009;43(8):595–602. doi:10.1136/bjsm.2009.061481.
43. Verhagen EA, van Tulder M, van der Beek AJ, et al. An eco-
nomic evaluation of a proprioceptive balance board training
programme for the prevention of ankle sprains in volleyball. Br J
Sports Med. 2005;39(2):111–5. doi:10.1136/bjsm.2003.011031.
44. Cumps E, Verhagen E, Annemans L, et al. Injury rate and
socioeconomic costs resulting from sports injuries in Flanders:
data derived from sports insurance statistics 2003. Br J Sports
Med. 2008;42(9):767–72. doi:10.1136/bjsm.2007.037937.
45. Hupperets MD, Verhagen EA, Heymans MW, et al. Potential
savings of a program to prevent ankle sprain recurrence: eco-
nomic evaluation of a randomized controlled trial. Am J Sports
Med. 2010;38(11):2194–200. doi:10.1177/0363546510373470.
46. Janssen KW, Hendriks MR, van Mechelen W, et al. The cost-
effectiveness of measures to prevent recurrent ankle sprains:
results of a 3-arm randomized controlled trial. Am J Sports Med.
2014;42(7):1534–41. doi:10.1177/0363546514529642.
47. Schmikli SL, Backx FJ, Kemler HJ, et al. National survey on
sports injuries in the Netherlands: target populations for sports
injury prevention programs. Clin J Sport Med. 2009;19(2):101–6.
doi:10.1097/JSM.0b013e31819b9ca3.
Health and Economic Burden of Running Injuries in Trailrunners
377
123
| Health and Economic Burden of Running-Related Injuries in Dutch Trailrunners: A Prospective Cohort Study. | [] | Hespanhol Junior, Luiz Carlos,van Mechelen, Willem,Verhagen, Evert | eng |
PMC9250525 | 1
Vol.:(0123456789)
Scientific Reports | (2022) 12:11223
| https://doi.org/10.1038/s41598-022-14616-w
www.nature.com/scientificreports
Effects of different inspiratory
muscle warm‑up loads
on mechanical, physiological
and muscle oxygenation responses
during high‑intensity running
and recovery
Anita B. Marostegan1*, Claudio A. Gobatto1, Felipe M. Rasteiro1, Charlini S. Hartz2,
Marlene A. Moreno2 & Fúlvia B. Manchado‑Gobatto1
Inspiratory muscle warm‑up (IMW) has been used as a resource to enhance exercises and sports
performance. However, there is a lack of studies in the literature addressing the effects of different
IMW loads (especially in combination with a shorter and applicable protocol) on high‑intensity running
and recovery phase. Thus, this study aimed to investigate the effects of three different IMW loads
using a shorter protocol on mechanical, physiological and muscle oxygenation responses during
and after high‑intensity running exercise. Sixteen physically active men, randomly performed four
trials 30 s all‑out run, preceded by the shorter IMW protocol (2 × 15 breaths with a 1‑min rest interval
between sets, accomplished 2 min before the 30 s all‑out run). Here, three IMW load conditions
were used: 15%, 40%, and 60% of maximal inspiratory pressure (MIP), plus a control session (CON)
without the IMW. The force, velocity and running power were measured (1000 Hz). Two near‑infrared
spectroscopy (NIRS) devices measured (10 Hz) the muscle’s oxygenation responses in biceps brachii
(BB) and vastus lateralis (VL). Additionally, heart rate (HR) and blood lactate ([Lac]) were also
monitored. IMW loads applied with a shorter protocol promoted a significant increase in mean and
minimum running power as well as in peak and minimum force compared to CON. In addition, specific
IMW loads led to higher values of peak power, mean velocity (60% of MIP) and mean force (40 and 60%
of MIP) in relation to CON. Physiological responses (HR and muscles oxygenation) were not modified
by any IMW during exercise, as well as HR and [Lac] in the recovery phase. On the other hand, 40%
of MIP presented a higher tissue saturation index (TSI) for BB during recovery phase. In conclusion,
the use of different loads of IMW may improve the performance of a physically active individual in a
30 s all‑out run, as verified by the increased peak, mean and minimum mechanical values, but not in
performance assessed second by second. In addition, 40% of the MIP improves TSI of the BB during
the recovery phase, which can indicate greater availability of O2 for lactate clearance.
High-intensity exercise such as tethered sprints is often applied in training programs1–3. Considering the sig-
nificance of the running power in sports and training programs, investigations have been conducted to improve
the consistency of this parameter by making it measurable using a tethered system capable of monitoring the
force, the velocity, and consequently the running power2,4–6. Furthermore, by applying a high-intensity exer-
cise test, especially with duration around 30 s (i.e., 30 s all-out run), it is possible to assess the peak, mean and
minimum power, as well as the fatigue index7–9, which contribute to training programs applied to athletes and
active participants. The 30 s all-out run is characterized by a predominantly anaerobic stimulus that increases
the physiological responses, such as heart rate and blood lactate production7,8,10, by locomotor muscles (more
OPEN
1Laboratory of Applied Sport Physiology, School of Applied Sciences, University of Campinas, Limeira, São Paulo,
Brazil. 2Postgraduate Program in Human Movement Sciences, Methodist University of Piracicaba, Piracicaba, Sao
Paulo, Brazil. *email:
2
Vol:.(1234567890)
Scientific Reports | (2022) 12:11223 |
https://doi.org/10.1038/s41598-022-14616-w
www.nature.com/scientificreports/
active muscles). Despite the significance of the anaerobic metabolism, the aerobic metabolism also plays an
important role during high-intensity exercise, as observed in the Wingate test11–13 and the 30 s all-out tethered
run7. In addition, the recovery process is essential to reestablish the energy stores, with significant participation
of the aerobic metabolism during and after high-intensity exercise 10,14, acting as an ally in the performance of
subsequent intense efforts15. Recently, our group observed that during an all-out 30 s running test the biceps
brachii (less active muscle) showed increased deoxygenation and a quick adjustment of post-effort oxygen levels
compared to the vastus lateralis10. These findings reinforce the tissue-dependent response, evidencing that the
organism adapts to a stressful stimulus in an integrative way.
Considering the integrative biology during physical effort16 in high-intensity exercise, it is known that the
oxygenation is directed to areas with higher demand, promoting vasoconstriction in less active muscles and vaso-
dilation in more active muscles. In this sense, the accumulation of metabolites (e.g., blood lactate and H + ions) in
the fatigued inspiratory muscles (IM) can initiate the inspiratory muscle metaboreflex, which triggers sympathetic
nerve activity that promotes adrenergic vasoconstriction, provoking competition with locomotor muscles for
oxygenation17–20, and consequently impairing the exercise performance. In line with this, it is reported that IM
plays a significant role during exercise and recovery19–22. Therefore, some studies have focused on IM training,
aiming at reducing the IM fatigue in order to maximize performance and oxygenation redistributions23–25. Moreo-
ver, the warm-up or pre-activation of inspiratory muscles (IMW) using a portable device has been suggested to
prepare this specific musculature before exercise, and consequently improve the physiological responses during
effort26–28, as well as sports performance29–32. Positive effects of IMW were observed during high-intensity, short-
duration exercises, with increased power in the Wingate test33,34. Furthermore, better oxygenation distributions
in more active muscles35 and decreased lactate accumulation in athletes after intermittent running36 were also
observed. However, the available literature lacks studies on the effects of IMW on more and less active muscles,
which are essential for oxygen uptake and metabolite removal, respectively10,37.
Different IMW protocols (i.e., number of sets, repetitions and time between sets) and loads (characterized
by airflow restriction in the inspiration phase) ranging from 5 to 80% of maximal inspiratory pressure (MIP)
have been proposed29,31,32,38–41. Nonetheless, most studies that applied this strategy used efforts based on indi-
vidual MIP, generally composed of 2 sets of 30 repetitions at 40% of MIP. This protocol has been suggested to
improve exercise performance through the preparation of the inspiratory muscles, to minimize the effects of
inspiratory muscle metaboreflex, reducing fatigue and improving oxygen delivery between the locomotor and
respiratory muscles29,31,34,40. The application of higher inspiratory loads such as 60–80% of MIP38,39 to improve
subsequent results is still controversial and lower loads such as 5–15% of MIP are commonly applied as a placebo
condition27,31,35,36,39. Concerning the number of repetitions, recent investigations have suggested a shorter IMW
protocol consisting of 2 sets of 15 inspiratory efforts in order to minimize the time spent during warm-up and
improve its application in a sports/physical training environment32,38,39. However, there is a lack of knowledge
about the effects of different inspiratory loads of IMW on the mechanical and physiological performance in
high-intensity and short-duration running, especially when performing such a short inspiratory protocol.
Therefore, this study aims to investigate the effects of different IMW loads (15%, 40% and 60% of MIP) using
a shorter protocol (2 sets of 15 repetitions with a 1-min rest interval between them) on the mechanical (i.e.,
power, force and velocity), physiological (i.e., heart rate and blood lactate) and muscle oxygenation responses
(in vastus lateralis and biceps brachii, considered the more and less active muscles, respectively) during high-
intensity running and passive recovery. Our hypothesis is that IMW at 40% of MIP will minimize IM fatigue,
allowing redirect the oxygenation to more active muscles with higher percentages of tissue saturation index (TSI),
consequently promoting a positive impact on the mechanical responses during the 30 s all-out run (especially
on mean running power and fatigue index), as well as contributing to a better post-effort blood lactate removal
in relation to other loads. In addition, as studies are controversial and there are gaps about the IMW effects on
high-intensity running, compared with the responses already observed in the literature in different populations
and types of exercise it is estimated that the load with 15% of MIP has a placebo effect and is equal to the control
session, while the 60% load does not improve performance, as it is considered a heavy load that will fatigue the
respiratory muscles.
Materials and methods
Ethics approval.
All procedures were approved by the local Research Ethics Committee of the School of
Medical Sciences of the State University of Campinas (protocol number: 99783318.4.0000.5404) and were in
accordance with the ethical recommendations in the Declaration of Helsinki. Participants were only evaluated
after having received information about the experimental procedures and risks and signing an informed consent
form.
Participants.
Sixteen physically active young men (local sports team players and street running partici-
pants) were evaluated (23 ± 1 years old, 73.2 ± 2.0 kg; 177.4 ± 1.9 cm; 9.0 ± 0.5% body fat, IPAQ at 4539 ± 942
metabolic equivalent-min/week; MIP at 145.6 ± 9.9 cmH2O, peak of global strength index (S-Index) at 139 ± 3
and mean S-Index at 123 ± 4, cmH2O). The analysis with G*Power software showed that a sample size of at least
12 individuals would be necessary to obtain a power of 80% with a significance level α = 5%, based on previously
published data10. The participants were invited to answer questionnaires about their levels of physical activity
(International Physical Activity Questionnaire—IPAQ), sports practice and health history. Only participants
that presented a minimum score to classify them as ‘physically active’ were included in the study42. Individuals
that reported metabolic, cardiovascular, respiratory or orthopedic disease were excluded from this research.
3
Vol.:(0123456789)
Scientific Reports | (2022) 12:11223 |
https://doi.org/10.1038/s41598-022-14616-w
www.nature.com/scientificreports/
Experimental approach to the problem.
The experimental design involved six visits to the laboratory
under similar conditions and at identical daytime (± 60 min), separated by 48–72 h (Fig. 1). On the first day,
after signing the informed consent form and completing the questionnaires, anthropometric and body composi-
tion measurements were performed. In the second visit, MIP and S-Index were determined 1 h apart to prevent
inspiratory fatigue. On the same day, the participants were familiarized with the inspiratory muscle warm-up
protocol and the non-motorized treadmill (NMT), when they were asked to perform five sprints of 10 s. In the
next four visits to the laboratory, all participants were submitted to high-intensity tethered exercise (30 s all-out
run), preceded by different IMW loads protocol (15%, 40% and 60% of MIP plus a control (CON) session with-
out the IMW protocol). These sessions were randomly performed by the individuals under the IMW protocols
used. Upon arrival at the laboratory, the participants were equipped with near-infrared spectroscopy (NIRS)
devices attached to the biceps brachii (BB) and vastus lateralis (VL) muscles, and a heart rate monitor (HR) for
data acquisition throughout the session. The participants remained at rest for 3 min for the determination of
baseline values (BL), including blood lactate concentration ([Lac]) at rest. Then, they were asked to warm-up on
a motorized treadmill for 5 min (7 km/h and 1% inclination) and rest for another 5 min. Subsequently, the IMW
protocol was performed. Two minutes after, the individuals were submitted to 30 s high-intensity tethered run-
ning for data acquisition (i.e., force, velocity and running power). Immediately after the test (T0), blood samples
were collected every 2 min up to 18 min of passive recovery (T2‒T18). In all sessions, the participants were
instructed to have a light meal, not to consume alcohol/caffeine and not to practice moderate-intense exercise
24 h before the tests. The procedures were performed in a controlled and isolated laboratory environment, and
the participants did not receive any information about each intervention.
Figure 1. (A) Experimental protocol illustrating the procedures performed on each day during the whole
study period. (B) Timeline of the protocol sessions. Physiological responses were monitored throughout the
exercise and the 18 min of passive recovery. Near-infrared spectroscopy devices (NIRS, represented by the red
rectangles) provided the biceps brachii and vastus lateralis oxygenation data. S-Index—global strength index;
CON—Control session; WU15, WU40 and WU60—Warm-up with 15%, 40% and 60% of maximal inspiratory
pressure (MIP), respectively.
4
Vol:.(1234567890)
Scientific Reports | (2022) 12:11223 |
https://doi.org/10.1038/s41598-022-14616-w
www.nature.com/scientificreports/
Inspiratory measurements and inspiratory muscle warm‑up.
The analysis was conducted by a
trained researcher who demonstrated the correct performance of the respiratory maneuver. The participants
remained seated on a chair, wearing a nose clip and a plastic mouthpiece connected to an analogical manovacu-
ometer (± 300 cmH2O; GER-AR, São Paulo, SP, Brazil) used to measure maximal pressures. A small hole (2 mm)
was introduced in the rigid mouthpiece in order to prevent glottic closure. The participants were instructed to
complete three to five acceptable and reproducible maximum maneuvers (i.e., differences of 10% or less between
values), with 1 min interval between maneuvers43,44. Each inspiratory effort was sustained for at least 1 s, and the
MIP was considered the highest value between these attempts45.
After 1 h, a dynamic proposal to characterize the strength of the participants’ IM, the global strength index
(S-Index), was assessed by an inspiratory threshold (POWERbreathe K5, IMT Technologies Ltd., Birmingham,
UK), with the participants in standing position and using a nose clip. Thirty dynamic inspirations resistance-
free were performed slowly, with verbal encouragement to inhale a greater air capacity32. During the protocol,
breathing pattern curves were monitored by graphic records provided by Breathe-link® software. At the end of
the test, the algorithm provided the mean and peak values of the S-Index (in units of cmH2O).
The inspiratory muscle warm-up (IMW) protocol loads were also applied using the inspiratory device POW-
ERbreathe K5. The participants initiated every breath from residual volume and were encouraged to continue
the respiratory effort until further excursion of the thorax was not possible, with a diaphragmatic breathing pat-
tern. Subsequently, they were instructed to keep the same inspiratory pressure and the breathing pattern curves
were also monitored by Breathe-link® software. The total protocol was comprised of two sets of 15 inspirations
with a 1-min rest interval between them. The loads were equivalent to 15% (WU15), considered placebo by
the literature32,35,39, 40% (WU40) and 60% (WU60) of MIP. All experimental trials were randomly distributed.
The 30 s all‑out run test and mechanical measurements.
The 30 s all-out run was performed on a
non-motorized treadmill (NMT) (Inbramed Super ATL, Inbrasport, Porto Alegre, Brazil), as detailed by Man-
chado-Gobatto et al.10. Two minutes after the IMW or CON protocols, the participants were asked to run at
maximum intensity for 30 s, tethered by their waist to an inextensible steel cable attached to a load cell (CSL/
ZL-250, MK Controle e Instrumentação Ltda, Brazil) for horizontal force measurement. Other four load cells
(CSAL/ZL-500, MK Controle e Instrumentação Ltda, Brazil) were positioned under the NMT platform to meas-
ure the vertical force (signal frequency at 1000 Hz). A hall-effect sensor in the frontal axis of the NMT provided
pulses for velocity acquisition. Therefore, both vertical and horizontal force components were measured during
the running exercise along with velocity to calculate the power running. The signals were synchronized and the
product between force and velocity resulted in the running power, with the peak, mean and minimum values
relativized to body mass. Fatigue index (FI) was also calculated by the following equation: FI = (peak power—
minimum power)/peak power * 100). The system was calibrated on the test days.
Physiological responses.
Blood lactate concentration and heart rate. For lactate concentrations ([Lac])
at rest, post-effort and every 2 min up to 18 min of passive recovery, blood samples (25 µl) were collected from
the participants’ earlobe with heparinized capillaries, deposited in microtubes (Eppendorf, 1.5 ml containing
50 µl of 1% sodium fluoride—NaF) and frozen at − 20 °C. The [Lac] were determined by a lactate analyzer (YSI-
2300-STAT-Plus™, Yellow Springs, USA). Throughout the protocols, the heart rate (HR) was constantly recorded
(at 1 Hz) (Polar V800, Finland). For all variables, the peak, mean and minimum values were calculated (during
the test we used the 30 s responses, while during passive recovery we considered the mean of 18 min).
Muscle oxygenation by NIRS measurements.
Total hemoglobin ([tHb] = oxyhemoglobin
([O2Hb]) + deoxyhemoglobin [HHb]) and the equilibrium between oxygen supply and consumption were calcu-
lated using the tissue saturation index (TSI = [O2Hb]/([O2Hb] + [HHb]) × 100%)46 throughout the experimental
protocol by two PortaMon devices (Artinis, Medical Systems BV, Zetten, Netherlands) working on the modified
Beer–Lambert law. Each device has three light source transmitters (with two wavelengths of 760 and 850 nm),
positioned at 30, 35 and 40 mm from the receiver. The devices were safely fixed and covered to eliminate back-
ground light after shaving and cleaning the skin surface. While one was positioned in the medial part (belly) of
the BB10,37,47 of the right arm, considered less active during running, the other was allocated in the VL of the right
leg (considered more active), 15 cm above the proximal edge of the patella and 5 cm to the external side10,48–50.
Skinfolds for BB (3.3 ± 0.2 mm) and VL (11.2 ± 1.2 mm) were less than half the distance between the source and
the deepest detector (i.e., 20 mm). Different path lengths (DPF) were used for BB (3.78) and VL (3.83)10. The
signals were smoothed using a 10th order low-pass zero-phase Butterworth filter (cutoff frequency of 0.1 Hz)50,
and recorded and analyzed on Oxysoft® software (Artinis Medical System, Netherlands). The (Δ)[tHb] in micro-
molar units (μM) was obtained by subtracting these values from the final 30 s of the baseline period of 3 min.
Examples of NIRS signals in BB and VL muscles of one participant during the four sessions (CON, WU15,
WU40 and WU60) are displayed in Supplementary file 1, whereas the descriptive graphics of [O2Hb] and [HHb]
during and after the 30 s all-out run are in Supplementary file 2.
Statistical analysis.
All analyses were performed using Statistica software (version 7.0), and the results
are expressed as mean ± standard error of the mean (SEM). Data normality and homogeneity distribution
were tested by Shapiro–Wilk and Levene’s test, respectively. Two-way repeated measures analysis of variance
(ANOVA) was applied to study the effect of IMW (CON x WU15 x WU40 x WU60) and time during the 30 s
of exercise (on mechanical and physiological responses) as well as the 18 min (every 2 min) of passive recovery
(on physiological responses) compared to the baseline condition (BL). Three-way repeated measures ANOVA
investigated the effect of IMW, time, and limb muscles (BB vs VL) on muscle oxygenation during the 30 s all-out
5
Vol.:(0123456789)
Scientific Reports | (2022) 12:11223 |
https://doi.org/10.1038/s41598-022-14616-w
www.nature.com/scientificreports/
run and recovery phase. One-way repeated measures ANOVA was applied to investigate the effects of IMW pro-
tocols on peak, mean, minimum values for force (N/kg), velocity (m/s), running power (W/kg), FI, as well HR
and [Lac] values. Two-way repeated measures ANOVA (effects of IMW and limb muscles) analyzed the peak,
mean and minimum values of muscle oxygenation responses. Lastly, post hoc Newman−Keuls test was used to
detect differences (P ≤ 0.05). The partial eta-squared (ηp
2: 0.01 = small, 0.06 = moderate, 0.14 = large) statistics
provided estimates of the effect sizes.
Results
Analyses during the 30 s all‑out run.
The mechanical and heart rate results during the 30 s all-out
run are shown in Fig. 2. Panel A shows similar running power responses between interventions during the
30 s all-out run test. The two-way repeated measures ANOVA demonstrated a significant effect of time on this
parameter (F(29,1740) = 255.50, P ˂ 0.001, ηp
2 = 0.790). There was an increase in running power in the first second
until a peak power was reached at approximately 6 s and a consequent decrease after this time for all interven-
tions, without any effects of IMW on the studied parameter. Panels B, C, D and E display the peak, mean and
minimum values for power, force, velocity and FI, respectively. For these measurements, the one-way repeated
measures ANOVA revealed an effect of IMW (F(3,45) = 2.98, P = 0.041, ηp
2 = 0.166) on peak power (Panel B), with
higher values in WU60 (34.1 ± 1.0 W/kg, P = 0.029) compared to CON (31.2 ± 1.4 W/kg). For mean power (Panel
B; F(3,45) = 4.5, P = 0.007, ηp
2 = 0.232), minimum power (Panel B; F(3,45) = 3.3, P = 0.028, ηp
2 = 0.181), peak force
(Panel C; F(3,45) = 3.9, P = 0.013, ηp
2 = 0.208), minimum force (Panel C; F(3,45) = 5.5, P = 0.002, ηp
2 = 0.269), all IMW
loads led to higher values in comparison with CON. Additionally, regarding specific IMW effects on mean force
(Panel C; F(3,45) = 3.2, P = 0.028, ηp
2 = 0.180), higher values were observed in WU40 (5,6 ± 0.1 N/kg, P = 0.035) and
WU60 (5.6 ± 0.1 N/kg, P = 0.020) than in CON (5.2 ± 0.1 N/kg), whereas a higher mean velocity value (Panel D;
F(3,45) = 3.5, P = 0.022, ηp
2 = 0.190) was obtained for WU60 (4.5 ± 0.1 m/s) than for CON (4.3 ± 0.1 m/s). Panel
Figure 2. Mechanical and physiological results observed during the 30 s all-out run under control conditions
(CON, black color) and after IMW loads at 15% (WU15, blue color), 40% (WU40, green color) and 60% of MIP
(WU60, red color). ✝ Indicates difference in relation to the control session. (P < 0.05).
6
Vol:.(1234567890)
Scientific Reports | (2022) 12:11223 |
https://doi.org/10.1038/s41598-022-14616-w
www.nature.com/scientificreports/
F represents the HR responses second by second during the 30 s all-out run. The two-way repeated measures
ANOVA pointed to a significant effect of time (F(29,1740) = 350.76, P ˂ 0.001, ηp
2 = 0.854), yet no effects among
interventions were detected. The HR constantly increased, reaching the highest values in the last second of the
test. Furthermore, the peak, mean and minimum HR values were not influenced by IMW (Panel G).
The changes in muscle oxygenation are shown in Fig. 3. Panels A and B illustrate the TSI responses in BB
and VL, respectively. The three-way repeated measures ANOVA revealed effects of limb muscle (F(1,120) = 85.83,
P ˂ 0.001, ηp
2 = 0.419), time (F(29,3480) = 928.57, P ˂ 0.001, ηp
2 = 0.886) and significant interaction between time
x limb muscle (F(29,3480) = 36.39, P ˂ 0.001, ηp
2 = 0.233), but no IMW effects. The peak, mean and minimum
TSI data (Panel C) were analyzed by two-way ANOVA, which indicated the effect of limb muscle (TSI; peak:
F(1,30) = 20.67, P ˂ 0.001, ηp
2 = 0.408; mean: F(1,30) = 33.49, P ˂ 0.001, ηp
2 = 0.527 and minimum: F(1,30) = 71.53, P ˂
0.001, ηp
2 = 0.705) and IMW only for TSI mean values (F(1,90) = 3.72, P = 0.014, ηp
2 = 0.110). Despite these findings,
no interaction between IMW and limb muscles was observed.
The Δ [tHb] for BB and VL are presented in Panels D and E. The three-way repeated measures ANOVA
showed the effects of limb muscle (F(1,119) = 15.03, P ˂ 0.001, ηp
2 = 0.112), time (F(29,3451) = 82.66, P ˂ 0.001,
ηp
2 = 0.410), and an interaction between time x limb muscle (F(29,3451) = 16.86, P ˂ 0.001, ηp
2 = 0.124). However,
no difference was observed through post-hoc analyses (IMW x time x muscle limb). Whereas for peak values
no differences were observed, for mean (F(1,30) = 10.00, P = 0.003, ηp
2 = 0.250) and minimum values (Panel F;
F(1,30) = 23.73, P ˂ 0.001, ηp
2 = 0.442) a limb effect was detected without interaction with IMW.
Analyses in passive recovery.
Physiological responses obtained at baseline (BL) immediately after the
30 s all-out run (T0) and during passive recovery (every 2 min: T2–T18) are displayed in Fig. 4. Panel A shows
that the curves of blood lactate were similar in all interventions, with a clear effect of time (F(10,600) = 1053.76, P
˂ 0.001, ηp
2 = 0.946). A [Lac] peak was observed at 8 min for all participants ([Lac] peak in CON: 17.14 ± 0.62;
WU15: 16.00 ± 0.51; WU40: 16.11 ± 0.56 and WU60: 16.60 ± 0.65, mM). No effects of IMW were observed.
[Lac] values did not return to their initial values (BL) after 18 min of recovery, independently of the IMW
load applied. A similar HR behavior was observed for all conditions (Panel C), with a significant effect of time
(F(10,600) = 1646.51, P ˂ 0.001, ηp
2 = 0.965). The HR peak was obtained immediately after the 30 s all-out run (post-
effort, T0), reaching higher values than T2–T18 and decreasing over time. After 18 min, the HR values did not
Figure 3. Results of tissue saturation indexes (TSI, Panels A–C) and total hemoglobin ([tHb], Panels D–F)
in biceps brachii (BB) and vastus lateralis (VL) at each second during the 30 s all-out run on a non-motorized
treadmill under control conditions (CON, black color) and after the IMW loads with 15% (WU15, blue color),
40% (WU40, green color) and 60% of MIP (WU60, red color), in line graphs. In bar graphs, the dark colors
represent the BB, while the light colors correspond to the VL. MIP = maximum inspiratory pressure.
7
Vol.:(0123456789)
Scientific Reports | (2022) 12:11223 |
https://doi.org/10.1038/s41598-022-14616-w
www.nature.com/scientificreports/
reach the initial values under any condition. Peak, mean and minimum values for [Lac] and HR are shown in
panels B and D, respectively. IMW was not able to influence these results.
Changes in muscle oxygenation during passive recovery are shown in Fig. 5. Panels A and B display the
TSI responses in BB and VL, respectively. The three-way repeated measures ANOVA revealed effects of limb
muscle (F(1,120) = 93.23, P ˂ 0.001, ηp
2 = 0.437), time (F(10,1200) = 643.46, P ˂ 0.001, ηp
2 = 0.843) and a significant
interaction between time x limb muscle (F(10,1200) = 32.43, P ˂ 0.001, ηp
2 = 0.213), but no interaction with IMW.
In all interventions, BB presented lower saturation in relation to BL at T0 and T2, and after 4 min of recovery
(T4) the saturation returned to BL values. An interesting finding was observed in WU40, which reached higher
oxygenation values from T4 to T10 compared to BL. In VL, only WU60 at T2 presented different values from
BL. Moreover, higher saturation was detected in VL than in BB at T0. The two-way repeated measures ANOVA
revealed an effect of limb muscle on BL (F(1,30) = 61.16, P ˂ 0.001, ηp
2 = 0.671), peak (F(1,30) = 17.87, P ˂ 0.001,
ηp
2 = 0.373), mean (F(1,30) = 30.06, P ˂ 0.001, ηp
2 = 0.501) and minimum (F(1,30) = 55.17, P ˂ 0.001, ηp
2 = 0.648)
values (Panel C) without IWM interaction.
Regarding [tHb] (Panel D‒E), there were a significant effect of time (F(10,1200) = 44.73, P ˂ 0.001, ηp
2 = 0.272)
and interaction between time x limb muscle (F(10,1200) = 2.95, P ˂ 0.001, ηp
2 = 0.024). Compared to BL values, the
passive recovery time showed lower values for BB at T0 only in WU40, while higher values were noted at T2–T12
in WU60, subsequently returning to BL values. In VL, differences from BL values were observed at specific
moments in CON (T2‒T8), WU40 (T18) and WU60 (T4‒T18). IMW strategies did not affect peak, mean and
minimum [tHb] values in passive recovery (Panel F).
Discussion
To the best of our knowledge, this is the first study dedicated to investigating the effects of different IMW loads
(15, 40 and 60% of MIP) on mechanical and physiological responses, including oxygenation in more and less
active muscles, during and after high-intensity, short-duration running exercise. Additionally, we studied these
acute inspiratory strategies using a shorter protocol (i.e., lower number of exercise repetitions). Our main findings
revealed some effects of IMW, performed with 2 sets of 15 repetitions with a 1-min rest interval between the sets,
on the high-intensity running effort and recovery, independently of the load applied. Regarding the mechanical
parameters, all IMW promoted a significant increase in mean and minimum running power, as well as in peak
and minimum running force compared to CON. Additionally, when applying specific IMW loads higher values
were observed for peak power, mean velocity (WU60) and mean force (WU40 and WU60) in relation to CON.
The physiological responses, including HR and oxygenation in more and less active muscles during the running
exercise, were not modified by IMW, at least not during the 30 s high-intensity running nor for HR and [Lac]
in the post-effort phase. By comparing the responses in BB and VL, no differences were observed in mechanical
and muscle oxygenation during the 30 s all-out test. In passive recovery, higher TSI values for VL were detected
in the post-effort phase (T0) for all protocols. An interesting finding was observed in WU40, which reached
Figure 4. Physiological results of passive recovery immediately after the 30 s all-out run and every 2 min up to
18 min (mean ± EPM). Results under control conditions (CON, black color) and after IMW protocols with loads
at 15% (WU15, blue color), 40% (WU40, green color) and 60% of MIP (WU60, red color). # corresponds to
difference in relation to baseline (BL) for each protocol. Dashed line means no differences among protocols.
8
Vol:.(1234567890)
Scientific Reports | (2022) 12:11223 |
https://doi.org/10.1038/s41598-022-14616-w
www.nature.com/scientificreports/
higher oxygenation values from T4 to T10 compared to BL. It can be then suggested that the use of different loads
of IMW promotes an improvement in performance corroborated by the increased peak, mean and minimum
mechanical values, but not in the performance assessed second by second. Also, WU40 may improve recovery
phase with higher oxygenation in BB.
Inspiratory muscle warm‑up and performance.
Our choice to investigate the effect of different IMW
loads on the performance of high-intensity, short-duration running exercise was based on previous studies that
indicate the positive effect of IMW, but used different IWM protocols in intermittent running27, in Wingate
tests33,34, in 100 m freestyle swimming29, in specific hockey drills31 and in a simulate judo match32. Additionally,
considering the large use of tethered efforts in physical and sports programs together with the significance of
high-intensity exercise in this context, we focused on the evaluation of the IMW impact on the force, velocity
and running power of 30 s all-out run sessions using different inspiratory loads. As shown in Fig. 2, the same
characteristics were observed for the curve of running power throughout the tests, with no differences among
the IMW load interventions during the 30 s all-out run sessions. As previously mentioned, running power, force
and velocity were improved by the IMW loads, more specifically the WU40 and WU60, which significantly
influenced the mechanical variables compared to CON, suggesting an improvement in running performance for
active participants. Regarding the exercise performance of athletes, the IMW combined with specific warm-up
was capable of reducing the time in 100 m freestyle swimming29, treadmill sprint performance51 and interactions
among the technical-tactical, physical, physiological, and psychophysiological parameters in a simulated judo
match32. Studies that used IMW as the only means of prior effort to main motor task also observed a reduction in
the sensation of dyspnea and an improvement in the distance walked in one badminton-footwork test36, as well
as in one shuttle run test52. Similarly, Özdal and colleagues34 observed an increase in peak and relative power in
Figure 5. Results of tissue saturation indexes (TSI, Panels A–C) and total hemoglobin ([tHb], Panels D–F) in
passive recovery immediately after the 30 s all-out run (T0) and every 2 min up to 18 min (T2–T18) compared
to BL values. Timeline panels show the biceps brachii (BB) and vastus lateralis (VL) under control conditions
(CON, black color) and after the IMW loads with 15% (WU15, blue color), 40% (WU40, green color) and
60% of MIP (WU60, red color). Bar graphs present the peak, mean and minimum values (mean ± SEM), with
dark colors for BB and light colors for VL. MIP = maximum inspiratory pressure. # corresponds to difference
compared to baseline (BL); * means significantly different from BB. Dashed line means no differences among
protocols. Colored line means differences in a specific protocol.
9
Vol.:(0123456789)
Scientific Reports | (2022) 12:11223 |
https://doi.org/10.1038/s41598-022-14616-w
www.nature.com/scientificreports/
anaerobic Wingate test performed by hockey athletes after IMW (2 × 30 at 40% of MIP with a 2-min rest interval
between the sets).
Some additional positive effects of IMW on respiratory response and performance in a shuttle run test27,
on respiratory muscle strength28,53, on performance in a knee flexion–extension protocol accomplished in an
isokinetic test by healthy sedentary participants41 and on long-distance test30 were previously observed. On the
other hand, other studies did not find significant effects of IMW for both active individuals and athletes perform-
ing different types of exercise and tests39,43,46,53–56. It is important to consider the diversity of the IMW protocols
when applied to different populations, sports modalities and exercise tests, which makes it difficult to compare
the results obtained with other findings. Moreover, most investigations do not describe the time interval between
the IMW sets and the time between the IMW application and the test or main exercise. Knowing that the effects
of warm-up can be affected by several factors, such as protocol, load, performance level, type of exercise, time
interval between the conditioning stimulus and the performance testing, etc.57,58, more attention could be paid
to these aspects. In this sense, we focused on a shorter IMW protocol (2 sets of 15 repetitions with a 1-min rest
interval between them, concluded 2 min before the running test), using different inspiratory loads in each session
(without and with 15, 40 and 60% of MIP) applied to active participants. Regarding the respiratory parameters,
even though our participants did not have any experience with respiratory training or inspiratory warm-up,
they reached good MIP values (145.6 ± 9.9 cmH2O), similar to Japanese athletes in triathlon and wrestling (light
category) modalities (145.8 and 147.3 cmH2O, respectively)59.
Inspiratory muscle warm‑up and physiological responses.
The ventilatory responses in high-inten-
sity exercises can affect the perfusion dynamics of the locomotor muscles and tissue saturation indices, rep-
resenting a limitation of exercise performance19,20,60. According to the literature, inspiratory muscle warm-up
can be a strategy to potentialize oxygenation redistribution to more active muscles during physical exercise35.
However, improvements in post-effort recovery process remain unexplored. Thus, we measured for the first
time the physiological responses in exercise and recovery, including oxygenation analysis in more or less active
muscles (which are relevant to providing oxygen and removing metabolites, respectively). Few scientific inves-
tigations, especially in sports, have been conducted to study the IMW potential to minimize respiratory fatigue
and improve the oxygenation redistribution in high-intensity exercise35,40,46,51,61. When studying the oxygena-
tion in the gastrocnemius muscle of female soccer players by submitting them to submaximal cycling test and
intermittent cycling test, Cheng et al.35 demonstrated that the IMW protocol can enhance oxygen saturation in
this tissue. However, the authors did not observe changes in performance, possibly due to the lack of specificity
in the test for these athletes.
In our study, second-by-second NIRS analyses did not reveal the effects of IMW on muscle responses, regard-
less of the load (Fig. 3, Panels A‒B). Despite the evidence of limb muscle effect on TSI peak, mean and mini-
mum values (Fig. 3, Panel C) and [tHb] mean and minimum values (Fig. 3, Panel F), no interaction with IMW
protocols was observed, indicating a similar behavior in BB and VL oxygenation. Whether the IMW promotes a
positive effect on the oxygenation redistribution35, this was not observed in more and less active muscles. Con-
sidering the increased respiratory muscle work and the competition with locomotor muscles for O2 supply17–20,
the analysis of inspiratory muscle oxygenation could provide some insight into the oxygenation between these
muscles. However, the analysis of oxygenation occurs only in accessory and secondary inspiratory muscles62,
and it does not directly reflect the oxygenation of the muscle with the greatest potential for oxygen uptake and
the most affected one by IMW, the diaphragm. Recently, studies addressing the IMW applied to speed skaters on
ice time trial also did not report any improvement in muscle oxygenation variables in the right VL, with some
limitations pointed out by the authors, such as leg compression garments and small sample size40,46. On the other
hand, in high-intensity sprint10 and high-intensity cycling37 a difference in more and less active muscles was
observed, suggesting adjustments in oxygenation during effort in a tissue-dependent manner.
In order to support the high demand of the respiratory muscles during exercise, the VO2 and oxygenation
in this region are increased, and may compromise cardiac output by 14–16% in well-trained individuals63, thus
affecting oxygenation distribution to locomotor muscles64. In high-intensity effort, these locomotor muscles also
use predominantly anaerobic pathways, resulting in lactate production. According to previous studies, inspiratory
muscles may play an important role as lactate consumers21,22. In this context, Lin et al.36 indicated a reduction
in [Lac] in badminton players after IMW. Regarding [Lac], the peak values observed here (~ 16 mM) confirmed
their significant anaerobic contribution in the 30 s all-out run. These findings corroborate previous studies on
exercises characterized by anaerobic contribution9,10,65,66. In our study, both HR and [Lac] were not affected by
IMW interventions, and 18 min post-effort in passive recovery was not sufficient to make these physiological
responses return to baseline values (Fig. 4, Panels A–D).
With respect to passive recovery, a higher decrement in oxygenation was observed immediately after the
exercise, the so-called post-effort phase (T0), with quick adjustments after T2 for both muscles (Fig. 5). Osawa
et al.37 reported that tissue oxygenation did not begin immediately after high-intensity cycling effort and that
deoxygenation occurred for a few seconds. In the present study, TSI percentages started to rise immediately after
the exercise, and after 4 min (T4) they returned to baseline values (Fig. 5, Panels A‒B). Interestingly, only WU40
presented higher TSI values in BB from T4 to T10 in relation to BL. We did not perform correlation analyses,
however, Manchado-Gobatto et al.10 observed a significant correlation between the [Lac] peak and TSI and [tHb]
values in BB. In this regard, higher TSI in BB suggests an important role of this parameter during recovery since
O2 is considered to be essential in oxidative pathways for lactate clearance67. In VL, only WU60 at T2 reached
lower values than BL. Despite our initial hypothesis, which proposed that 40% of MIP would redirect oxygena-
tion to VL with a consequent positive impact on mechanical responses during a tethered 30 s all-out run, in
10
Vol:.(1234567890)
Scientific Reports | (2022) 12:11223 |
https://doi.org/10.1038/s41598-022-14616-w
www.nature.com/scientificreports/
addition to a better blood lactate removal during recovery in relation to other loads, we observed that different
inspiratory loads can improve mechanical parameters and recovery oxygenation.
Recent studies investigated the effects of IMW as a warm-up strategy combined with core warm-ups on recov-
ery period between intermittent exercise and repeated sprints on NMT51 and on recovery periods of sprints on a
cycle ergometer68. Although the authors evaluated recovery, muscle oxygenation was only observed during exer-
cise. In our study, the comparison between BB and VL revealed that only immediately after the 30 s all-out run
the VL presented higher values for all interventions (Fig. 5, Panels A–B). These results corroborate those reported
by Manchado-Gobatto et al.10, who did not use inspiratory strategy to improve the running performance. Finally,
the responses during exercise and recovery are a complex process. Thus, to improve the interpretation of IMW on
running and recovery, integrative analyses could reveal responses beyond conventional statistics. For example,
our group recently observed the improvement in the technical and tactical parameters in a judo simulated fight
using the same shorter IMW protocol used herein, based on a complex network analysis32. In such study, the
centrality metrics revealed that the IWM at 15% of MIP favored the interactions among the psychophysiological,
physical and physiological parameters, while the IWM at 40% of MIP was able to improve performance in the
judo match. Therefore, our next investigations will be considering these findings.
Furthermore, a recent study indicated NIRS measurements as a future physiological marker, showing no
significant differences regarding the respiratory compensation point69. These findings highlight the relationship
between systemic (i.e., ventilatory) and peripheral (i.e., oxygenation of locomotor and non-locomotor muscles)
physiological breakpoints. In this sense, future studies should consider respiratory strategies associated with
the NIRS technique to improve knowledge about the intensity of the training zone. Moreover, the IMW can
attenuate muscle deoxygenation during exercise35 and the NIRS technique can contribute to the monitoring of
oxygenation in clinical practice70, especially in patients with exercise-induced ischemic pain caused by reduced
blood flow to the lower extremities71.
Limitations and strengths
Despite the use of technologies with high-frequency signal acquisition, some limitations regarding our results
must be addressed. First, the all-out run performed on a NMT has shown reliable results in the scientific
literature5,7–10,66,72. However, we did not test the reproducibility of the four IMW interventions – although it is
safe to say that they exhibited similar results in the 30 s all-out run tests (Fig. 2) with no differences in power
running among the IMW loads second by second. We chose to investigate a classic anaerobic test (30 s all-out
run), considering the aerobic component around ~ 18–20% in these efforts7,13. We observed an effect of IMW on
the mechanical parameters, which did not result in greater muscle oxygenation differentiation. It is possible that
by applying another slightly longer exercise protocol or repeated sprints the impact of IMW can be observed on
both mechanical parameters (second by second) and physiological responses. Additionally, we did not use the
gas analyzer to investigate the oxygen uptake due to our experimental design, nor investigated oxygenation of
the inspiratory muscles. Still, we are aware that the association of NIRS measurements and VO2 exchange would
improve our data interpretation, but we have not tested whether this equipment can interfere with breathing
pattern or breath frequency altering the isolated effects of IMW. Another limitation refers to our participants’
characteristics, as only healthy active males, non-athletes performed the test. In future studies, we suggest the
inclusion of female participants, the comparison of IMW with a shorter protocol and the analysis of effects in
both high-performance athletes and non-athletes in different types of exercises, such as repeated-sprint effort.
The strengths of this study include: (i) the investigation of the inspiratory muscle strategy through a shorter
protocol applied with methodological rigor, performed with practical and high-quality inspiratory devices, (ii)
the running effort performed on a non-motorized treadmill able to identify, with high signal capture, minimal
changes in mechanical variables during high-intensity exercise; (iii) monitoring oxygenation responses in dif-
ferent muscle groups, associated or not with respiratory strategies, allows a more integrative interpretation of
this variable during effort and also during recovery.
Conclusion
In summary, different IMW loads with a shorter protocol (2 sets of 15 repetitions with a 1-min rest interval
between sets and 2 min before exercise) applied on high-intensity running exercise suggested an improvement
in performance corroborated by increased peak, mean and minimum mechanical values, but not in power and
oxygenation assessed second by second. With respect to muscle oxygenation, these measurements demonstrated
that the mechanisms by which IMW could possibly exert an effect on performance were not affected by these
protocols, as all interventions showed similar and rapid adjustments of oxygenation responses during exercise
demands. Interestingly, during passive recovery WU40 presented a pronounced TSI value for BB, indicating a
greater availability of O2 for lactate clearance in a tissue-dependent manner.
Data availability
The data that support the findings of this study are available from the corresponding author on reasonable
request. Correspondence and requests for materials should be addressed to F.B.M.G.
Received: 7 February 2022; Accepted: 9 June 2022
References
1. Alcaraz, P. E., Palao, J. M., Elvira, J. L. & Linthorne, N. P. Effects of three types of resisted sprint training devices on the kinematics
of sprinting at maximum velocity. J. Strength Cond. Res. 22(3), 890–897. https:// doi. org/ 10. 1519/ JSC. 0b013 e3181 6611ea (2008).
11
Vol.:(0123456789)
Scientific Reports | (2022) 12:11223 |
https://doi.org/10.1038/s41598-022-14616-w
www.nature.com/scientificreports/
2. Cross, M. R. et al. Training at maximal power in resisted sprinting: Optimal load determination methodology and pilot results in
team sport athletes. PLoS ONE 13(4), e0195477. https:// doi. org/ 10. 1371/ journ al. pone. 01954 77 (2018).
3. Haugen, T., McGhie, D. & Ettema, G. Sprint running: From fundamental mechanics to practice-a review. Eur. J. Appl. Physiol.
119(6), 1273–1287. https:// doi. org/ 10. 1007/ s00421- 019- 04139-0 (2019).
4. Pereira, V. H. et al. Complex network models reveal correlations among network metrics, exercise intensity and role of body
changes in the fatigue process. Sci. Rep. 5, 10489. https:// doi. org/ 10. 1038/ srep1 0489 (2015).
5. Gama, M. C. T., Dos Reis, I. G. M., Sousa, F. A. B. & Gobatto, C. A. The 3-min all-out test is valid for determining critical power
but not anaerobic work capacity in tethered running. PLoS ONE 13(2), e0192552. https:// doi. org/ 10. 1371/ journ al. pone. 01925 52
(2018).
6. Zabaloy, S. et al. Relationships between resisted sprint performance and different strength and power measures in rugby players.
Sports (Basel) 8(3), 34. https:// doi. org/ 10. 3390/ sport s8030 034 (2020).
7. Sousa, F. A. B., Vasque, R. E. & Gobatto, C. A. Anaerobic metabolism during short all-out efforts in tethered running: Comparison
of energy expenditure and mechanical parameters between different sprint durations for testing. PLoS ONE 12(6), e0179378.
https:// doi. org/ 10. 1371/ journ al. pone. 01793 78 (2017).
8. Zagatto, A. M., Miyagi, W. E., Sousa, F. A. & Gobatto, C. A. Relationship between anaerobic capacity estimated using a single effort
and 30-s tethered running outcomes. PLoS ONE 12(2), e0172032. https:// doi. org/ 10. 1371/ journ al. pone. 01720 32 (2017).
9. da Cruz, J. P. et al. Anaerobic and agility parameters of salonists in laboratory and field. Int. J. Sports Med. 41(7), 450–460. https://
doi. org/ 10. 1055/a- 1088- 5429 (2020).
10. Manchado-Gobatto, F. B. et al. New insights into mechanical, metabolic and muscle oxygenation signals during and after high-
intensity tethered running. Sci. Rep. 10(1), 6336. https:// doi. org/ 10. 1038/ s41598- 020- 63297-w (2020).
11. Granier, P., Mercier, B., Mercier, J., Anselme, F. & Préfaut, C. Aerobic and anaerobic contribution to Wingate test performance in
sprint and middle-distance runners. Eur. J. Appl. Physiol. Occup Physiol. 70(1), 58–65. https:// doi. org/ 10. 1007/ BF006 01809 (1995).
12. Bediz, C. S. et al. Comparison of the aerobic contributions to Wingate anaerobic tests performed with two different loads. J. Sports
Med. Phys. Fitness 38(1), 30–34 (1998).
13. Beneke, R., Pollmann, C., Bleif, I., Leithäuser, R. M. & Hütler, M. How anaerobic is the Wingate Anaerobic Test for humans?. Eur.
J. Appl. Physiol. 87(4–5), 388–392. https:// doi. org/ 10. 1007/ s00421- 002- 0622-4 (2002).
14. Dupont, G., Moalla, W., Guinhouya, C., Ahmaidi, S. & Berthoin, S. Passive versus active recovery during high-intensity intermit-
tent exercises. Med. Sci. Sports Exerc. 36(2), 302–308. https:// doi. org/ 10. 1249/ 01. MSS. 00001 13477. 11431. 59 (2004).
15. Brini, S., Ahmaidi, S. & Bouassida, A. Effects of passive versus active recovery at different intensities on repeated sprint performance
and testosterone/cortisol ratio in male senior basketball players. Sci. Sports 35(5), e142–e147. https:// doi. org/ 10. 1016/j. scispo. 2019.
07. 015 (2020).
16. Hawley, J. A., Hargreaves, M., Joyner, M. J. & Zierath, J. R. Integrative biology of exercise. Cell 159(4), 738–749. https:// doi. org/
10. 1016/j. cell. 2014. 10. 029 (2014).
17. Dempsey, J. A., Romer, L., Rodman, J., Miller, J. & Smith, C. Consequences of exercise-induced respiratory muscle work. Respir.
Physiol. Neurobiol. 151(2–3), 242–250. https:// doi. org/ 10. 1016/j. resp. 2005. 12. 015 (2006).
18. Dempsey, J. A. Respiratory determinants of exercise limitation: Focus on phrenic afferents and the lung vasculature. Clin. Chest
Med. 40(2), 331–342. https:// doi. org/ 10. 1016/j. ccm. 2019. 02. 002 (2019).
19. Dempsey, J. A., La Gerche, A. & Hull, J. H. Is the healthy respiratory system built just right, overbuilt, or underbuilt to meet the
demands imposed by exercise?. J. Appl. Physiol. 129(6), 1235–1256. https:// doi. org/ 10. 1152/ jappl physi ol. 00444. 2020 (2020).
20. Rodriguez, F. R., Aughey, R. J. & Billaut, F. The respiratory system during intermittent-sprint work: Respiratory muscle work and
the critical distribution of oxygen. Resp. Physiol. https:// doi. org/ 10. 5772/ intec hopen. 91207 (2020).
21. Chiappa, G. R. et al. Blood lactate during recovery from intense exercise: Impact of inspiratory loading. Med. Sci. Sports Exerc.
40(1), 111–116. https:// doi. org/ 10. 1249/ mss. 0b013 e3181 591de1 (2008).
22. Brown, P. I., Sharpe, G. R. & Johnson, M. A. Loading of trained inspiratory muscles speeds lactate recovery kinetics. Med. Sci.
Sports Exerc. 42(6), 1103–1112. https:// doi. org/ 10. 1249/ MSS. 0b013 e3181 c658ac (2010).
23. Witt, J. D., Guenette, J. A., Rupert, J. L., McKenzie, D. C. & Sheel, A. W. Inspiratory muscle training attenuates the human respira-
tory muscle metaboreflex. J Physiol. 584(Pt 3), 1019–1028. https:// doi. org/ 10. 1113/ jphys iol. 2007. 140855 (2007).
24. Illi, S. K., Held, U., Frank, I. & Spengler, C. M. Effect of respiratory muscle training on exercise performance in healthy individuals:
A systematic review and meta-analysis. Sports Med. 42(8), 707–724. https:// doi. org/ 10. 1007/ BF032 62290 (2012).
25. Lorca-Santiago, J., Jiménez, S. L., Pareja-Galeano, H. & Lorenzo, A. Inspiratory muscle training in intermittent sports modalities:
A systematic review. Int. J. Environ. Res. Public Health. 17(12), 4448. https:// doi. org/ 10. 3390/ ijerp h1712 4448 (2020).
26. Volianitis, S. et al. Inspiratory muscle training improves rowing performance. Med. Sci. Sports Exerc. 33(5), 803–809. https:// doi.
org/ 10. 1097/ 00005 768- 20010 5000- 00020 (2001).
27. Tong, T. K. & Fu, F. H. Effect of specific inspiratory muscle warm-up on intense intermittent run to exhaustion. Eur. J. Appl. Physiol.
97(6), 673–680. https:// doi. org/ 10. 1007/ s00421- 006- 0233-6 (2006).
28. Özdal, M. Acute effects of inspiratory muscle warm-up on pulmonary function in healthy subjects. Respir. Physiol. Neurobiol. 227,
23–26. https:// doi. org/ 10. 1016/j. resp. 2016. 02. 006 (2016).
29. Wilson, E. E. et al. Respiratory muscle specific warm-up and elite swimming performance. Br. J. Sports Med. 48(9), 789–791. https://
doi. org/ 10. 1136/ bjspo rts- 2013- 092523 (2014).
30. Barnes, K. R. & Ludge, A. R. Inspiratory muscle warm-up improves 3,200-m running performance in distance runners. J. Strength
Cond. Res. 35(6), 1739–1747. https:// doi. org/ 10. 1519/ JSC. 00000 00000 002974 (2021).
31. Avci, N., Ozdal, M. & Vural, M. Influence of inspiratory muscle warm-up exercise on field hockey drag-flick and shooting perfor-
mance. Eur. J. Phys. Educ. Sport Sci. 6, 92–104. https:// doi. org/ 10. 46827/ ejpe. v6i11. 3641 (2021).
32. Cirino, C. et al. Complex network model indicates a positive effect of inspiratory muscles pre-activation on performance parameters
in a judo match. Sci. Rep. 11(1), 11148. https:// doi. org/ 10. 1038/ s41598- 021- 90394-1 (2021).
33. Chiappa, G. R. et al. Inspiratory resistive loading after all-out exercise improves subsequent performance. Eur. J. Appl. Physiol.
106(2), 297–303. https:// doi. org/ 10. 1007/ s00421- 009- 1022-9 (2009).
34. Özdal, M., Bostanci, Ö., Dağlioğlu, Ö., Ağaoğlu, S. A. & Kabadayi, M. Effect of respiratory warm-up on anaerobic power. J. Phys.
Ther. Sci. 28(7), 2097–2098. https:// doi. org/ 10. 1589/ jpts. 28. 2097 (2016).
35. Cheng, C. F. et al. Inspiratory muscle warm-up attenuates muscle deoxygenation during cycling exercise in women athletes. Respir.
Physiol. Neurobiol. 186(3), 296–302. https:// doi. org/ 10. 1016/j. resp. 2013. 02. 029 (2013).
36. Lin, H. et al. Specific inspiratory muscle warm-up enhances badminton footwork performance. Appl. Physiol. Nutr. Metab. 32(6),
1082–1088. https:// doi. org/ 10. 1139/ H07- 077 (2007).
37. Osawa, T., Shiose, K. & Takahashi, H. Delayed onset of reoxygenation in inactive muscles after high-intensity exercise. Adv. Exp.
Med. Biol. 977, 255–260. https:// doi. org/ 10. 1007/ 978-3- 319- 55231-6_ 35 (2017).
38. Arend, M., Kivastik, J. & Mäestu, J. Maximal inspiratory pressure is influenced by intensity of the warm-up protocol. Respir. Physiol.
Neurobiol. 230, 11–15. https:// doi. org/ 10. 1016/j. resp. 2016. 05. 002 (2016).
39. Merola, P. et al. High load inspiratory muscle warm-up has no impact on Special Judo fitness test performance. Ido Mov. Cult. 19,
66–74. https:// doi. org/ 10. 14589/ ido. 19.1.7 (2019).
40. Richard, P. & Billaut, F. Effects of inspiratory muscle warm-up on locomotor muscle oxygenation in elite speed skaters during 3000
m time trials. Eur. J. Appl. Physiol. 119(1), 191–200. https:// doi. org/ 10. 1007/ s00421- 018- 4015-8 (2019).
12
Vol:.(1234567890)
Scientific Reports | (2022) 12:11223 |
https://doi.org/10.1038/s41598-022-14616-w
www.nature.com/scientificreports/
41. Kayar, C., Özdal, M. & Vural, M. Acute effect of inspiratory muscle warm-up protocol on knee flexion-extension isokinetic strength.
Eur. J. Physiol. Educ. Sport Sci. https:// doi. org/ 10. 5281/ zenodo. 38590 54 (2020).
42. Ainsworth, B. E. et al. Compendium of physical activities: An update of activity codes and MET intensities. Med. Sci. Sports Exerc.
32(9 Suppl), S498–S504. https:// doi. org/ 10. 1097/ 00005 768- 20000 9001- 00009 (2000).
43. Hartz, C. S., Ferreira, C. R. & Moreno, M. A. Effects of the application of an inspiratory muscular warm-up protocol in the physical
performance of handball athletes. J. Exerc. Physiol. 20(5), 12–22. https:// doi. org/ 10. 1519/ JSC. 0b013 e3181 87456 (2017).
44. Antonelli, C. B. B., Hartz, C. S., Santos, S. D. S. & Moreno, M. A. Effects of inspiratory muscle training with progressive loading on
respiratory muscle function and sports performance in high-performance wheelchair basketball athletes: A randomized clinical
trial. Int. J. Sports Physiol. Perform. 15(2), 238–242. https:// doi. org/ 10. 1123/ ijspp. 2018- 0979 (2020).
45. Neder, J. A., Andreoni, S., Lerario, M. C. & Nery, L. E. Reference values for lung function tests. II. Maximal respiratory pressures
and voluntary ventilation. Braz. J. Med. Biol. Res. 32(6), 719–727. https:// doi. org/ 10. 1590/ s0100- 879x1 99900 06000 07 (1999).
46. Richard, P. & Billaut, F. Combining chronic ischemic preconditioning and inspiratory muscle warm-up to enhance on-ice time-
trial performance in elite speed skaters. Front. Physiol. 9, 1036. https:// doi. org/ 10. 3389/ fphys. 2018. 01036 (2018).
47. Ogata, H. et al. Relationship between oxygenation in inactive biceps brachii muscle and hyperventilation during leg cycling. Physiol.
Res. 56(1), 57–65. https:// doi. org/ 10. 33549/ physi olres. 930888 (2007).
48. Turner, L. A. et al. Inspiratory loading and limb locomotor and respiratory muscle deoxygenation during cycling exercise. Respir.
Physiol. Neurobiol. 185(3), 506–514. https:// doi. org/ 10. 1016/j. resp. 2012. 11. 018 (2013).
49. Kitada, T., Machida, S. & Naito, H. Influence of muscle fibre composition on muscle oxygenation during maximal running. BMJ
Open Sport Exerc. Med. 1(1), e000062. https:// doi. org/ 10. 1136/ bmjsem- 2015- 000062 (2015).
50. Woorons, X., Dupuy, O., Mucci, P., Millet, G. P. & Pichon, A. Cerebral and muscle oxygenation during repeated shuttle run sprints
with hypoventilation. Int. J. Sports Med. 40(6), 376–384. https:// doi. org/ 10. 1055/a- 0836- 9011 (2019).
51. Tong, T. K., Baker, J. S., Zhang, H., Kong, Z. & Nie, J. Effects of specific core re-warm-ups on core function, leg perfusion and
second-half team sport-specific sprint performance: A randomized crossover study. J. Sports Sci. Med. 18(3), 479–489 (2019).
52. Lomax, M., Grant, I. & Corbett, J. Inspiratory muscle warm-up and inspiratory muscle training: Separate and combined effects
on intermittent running to exhaustion. J. Sports Sci. 29(6), 563–569. https:// doi. org/ 10. 1080/ 02640 414. 2010. 543911 (2011).
53. Arend, M., Mäestu, J., Kivastik, J., Rämson, R. & Jürimäe, J. Effect of inspiratory muscle warm-up on submaximal rowing perfor-
mance. J. Strength. Cond. Res. 29(1), 213–218. https:// doi. org/ 10. 1519/ JSC. 00000 00000 000618 (2015).
54. Johnson, M. A., Gregson, I. R., Mills, D. E., Gonzalez, J. T. & Sharpe, G. R. Inspiratory muscle warm-up does not improve cycling
time-trial performance. Eur. J. Appl. Physiol. 114(9), 1821–1830. https:// doi. org/ 10. 1007/ s00421- 014- 2914-x (2014).
55. Thurston, T. S. et al. Effects of respiratory muscle warm-up on high-intensity exercise performance. Sports 3, 312–324. https:// doi.
org/ 10. 3390/ sport s3040 312 (2015).
56. Faghy, M. A. & Brown, P. I. Whole-body active warm-up and inspiratory muscle warm-up do not improve running performance
when carrying thoracic loads. Appl. Physiol. Nutr. Metab. 42(8), 810–815. https:// doi. org/ 10. 1139/ apnm- 2016- 0711 (2017).
57. Sanchez-Sanchez, J. et al. Effects of different post-activation potentiation warm-ups on repeated sprint ability in soccer players
from different competitive levels. J. Hum. Kinet. 61, 189–197. https:// doi. org/ 10. 1515/ hukin- 2017- 0131 (2018).
58. Wilson, J. M. et al. Meta-analysis of postactivation potentiation and power: effects of conditioning activity, volume, gender, rest
periods, and training status. J. Strength Cond. Res. 27(3), 854–859. https:// doi. org/ 10. 1519/ JSC. 0b013 e3182 5c2bdb (2013).
59. Ohya, T., Hagiwara, M., Chino, K. & Suzuki, Y. Maximal inspiratory mouth pressure in Japanese elite male athletes. Respir. Physiol.
Neurobiol. 230, 68–72. https:// doi. org/ 10. 1016/j. resp. 2016. 05. 004 (2016).
60. Katayama, K. et al. Effect of increased inspiratory muscle work on blood flow to inactive and active limbs during submaximal
dynamic exercise. Exp. Physiol. 104(2), 180–188. https:// doi. org/ 10. 1113/ EP087 380 (2019).
61. Ohya, T., Hagiwara, M. & Suzuki, Y. Inspiratory muscle warm-up has no impact on performance or locomotor muscle oxygenation
during high-intensity intermittent sprint cycling exercise. Springerplus 4, 556. https:// doi. org/ 10. 1186/ s40064- 015- 1355-2 (2015).
62. Rodriguez, R. F., Townsend, N. E., Aughey, R. J. & Billaut, F. Respiratory muscle oxygenation is not impacted by hypoxia during
repeated-sprint exercise. Respir. Physiol. Neurobio. 260, 114–121. https:// doi. org/ 10. 1016/j. resp. 2018. 11. 006 (2019).
63. Harms, C. A. et al. Effects of respiratory muscle work on cardiac output and its distribution during maximal exercise. J. Appl.
Physiol. 85(2), 609–618. https:// doi. org/ 10. 1152/ jappl. 1998. 85.2. 609 (1998).
64. Harms, C. A. Effect of skeletal muscle demand on cardiovascular function. Med. Sci. Sports Exerc. 32(1), 94–99. https:// doi. org/
10. 1097/ 00005 768- 20000 1000- 00015 (2000).
65. Engel, F. et al. Hormonal, metabolic, and cardiorespiratory responses of young and adult athletes to a single session of high-intensity
cycle exercise. Pediatr. Exerc. Sci. 26(4), 485–494. https:// doi. org/ 10. 1123/ pes. 2013- 0152 (2014).
66. Gobatto, C. A. et al. Corresponding assessment scenarios in laboratory and on-court tests: Centrality measurements by complex
networks analysis in young basketball players. Sci. Rep. 10(1), 8620. https:// doi. org/ 10. 1038/ s41598- 020- 65420-3 (2020).
67. Brooks, G. A. The science and translation of lactate shuttle theory. Cell Metab. 27(4), 757–785. https:// doi. org/ 10. 1016/j. cmet. 2018.
03. 008 (2018).
68. Cheng, C. F., Hsu, W. C., Kuo, Y. H., Chen, T. W. & Kuo, Y. C. Acute effect of inspiratory resistive loading on sprint interval exercise
performance in team-sport athletes. Respir. Physiol. Neurobiol. 282, 103531. https:// doi. org/ 10. 1016/j. resp. 2020. 103531 (2020).
69. Yogev, A. et al. Comparing the respiratory compensation point with muscle oxygen saturation in locomotor and non-locomotor
muscles using wearable NIRS spectroscopy during whole-body exercise. Front. Physiol. 13, 818733. https:// doi. org/ 10. 3389/ fphys.
2022. 818733 (2022).
70. Brochhagen, J., Coll Barroso, M. T., Baumgart, C., Freiwald, J. & Hoppe, M. W. Non-invasively measured central and peripheral
factors of oxygen uptake differ between patients with chronic heart failure and healthy controls. BMC Cardiovasc. Disord. 20(1),
378. https:// doi. org/ 10. 1186/ s12872- 020- 01661-4 (2020).
71. Treat-Jacobson, D. et al. Optimal exercise programs for patients with peripheral artery disease: A scientific statement from the
American Heart Association. Circulation 139(4), e10–e33. https:// doi. org/ 10. 1161/ CIR. 00000 00000 000623 (2019).
72. Pereira, V. H. et al. Computational and complex network modeling for analysis of sprinter athletes’ performance in track field tests.
Front. Physiol. 9, 843. https:// doi. org/ 10. 3389/ fphys. 2018. 00843 (2018).
Author contributions
F.B.M.G., C.A.G and M.A.M. conceived the main conceptual ideas, designed the project, performed data analysis
and interpretation, and were responsible for funding acquisition. A.B.M. and F.M.R. conducted data collection,
analysis and interpretation, and designed figures and table. A.B.M., C.A.G and F.B.M.G. wrote the main text
of the manuscript. M.A.M. and C.S.H. provided support on inspiratory muscle protocols and performed data
analysis. All authors reviewed the manuscript and have approved the submitted version.
Funding
São Paulo Research Foundation—FAPESP (2009/08535-5, 2012/06355-2, 2016/50250-1, 2018/05821-6,
2019/10666-2 and 2019/20894-2), the National Council for Scientific and Technological Development—CNPq
13
Vol.:(0123456789)
Scientific Reports | (2022) 12:11223 |
https://doi.org/10.1038/s41598-022-14616-w
www.nature.com/scientificreports/
(307718/2018-2, 308117/2018-2) and the Coordination for the Improvement of Higher Education Personnel—
CAPES (Finance Code 001). We also thank all the participants for the voluntary participation.
Competing interests
The authors declare no competing interests.
Additional information
Supplementary Information The online version contains supplementary material available at https:// doi. org/
10. 1038/ s41598- 022- 14616-w.
Correspondence and requests for materials should be addressed to A.B.M.
Reprints and permissions information is available at www.nature.com/reprints.
Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and
institutional affiliations.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International
License, which permits use, sharing, adaptation, distribution and reproduction in any medium or
format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the
Creative Commons licence, and indicate if changes were made. The images or other third party material in this
article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the
material. If material is not included in the article’s Creative Commons licence and your intended use is not
permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from
the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/.
© The Author(s) 2022
| Effects of different inspiratory muscle warm-up loads on mechanical, physiological and muscle oxygenation responses during high-intensity running and recovery. | 07-02-2022 | Marostegan, Anita B,Gobatto, Claudio A,Rasteiro, Felipe M,Hartz, Charlini S,Moreno, Marlene A,Manchado-Gobatto, Fúlvia B | eng |
PMC7893283 | royalsocietypublishing.org/journal/rspb
Research
Cite this article: Bohm S, Mersmann F,
Santuz A, Arampatzis A. 2021 Enthalpy
efficiency of the soleus muscle contributes to
improvements in running economy.
Proc. R. Soc. B 288: 20202784.
https://doi.org/10.1098/rspb.2020.2784
Received: 6 November 2020
Accepted: 5 January 2021
Subject Category:
Morphology and biomechanics
Subject Areas:
biomechanics, physiology
Keywords:
force–length and force–velocity relationship,
enthalpy–velocity relationship, triceps surae,
endurance running, strength training,
tendon stiffness
Author for correspondence:
Sebastian Bohm
e-mail: [email protected]
Electronic supplementary material is available
online at https://doi.org/10.6084/m9.figshare.
c.5271347.
Enthalpy efficiency of the soleus muscle
contributes to improvements in running
economy
Sebastian Bohm1,2, Falk Mersmann1,2, Alessandro Santuz1,2
and Adamantios Arampatzis1,2
1Department of Training and Movement Sciences, Humboldt-Universität zu Berlin, Philippstr. 13,
10115 Berlin, Germany
2Berlin School of Movement Science, Humboldt-Universität zu Berlin, Berlin, Germany
SB, 0000-0002-5720-3672; FM, 0000-0001-7180-7109; AS, 0000-0002-6577-5101;
AA, 0000-0002-4985-0335
During human running, the soleus, as the main plantar flexor muscle, gen-
erates the majority of the mechanical work through active shortening. The
fraction of chemical energy that is converted into muscular work (enthalpy
efficiency) depends on the muscle shortening velocity. Here, we investigated
the soleus muscle fascicle behaviour during running with respect to the
enthalpy efficiency as a mechanism that could contribute to improvements
in running economy after exercise-induced increases of plantar flexor
strength and Achilles tendon (AT) stiffness. Using a controlled longitudinal
study design (n = 23) featuring a specific 14-week muscle–tendon training,
increases in muscle strength (10%) and tendon stiffness (31%) and reduced
metabolic cost of running (4%) were found only in the intervention group
(n = 13, p < 0.05). Following training, the soleus fascicles operated at higher
enthalpy efficiency during the phase of muscle–tendon unit
(MTU)
lengthening (15%) and in average over stance (7%, p < 0.05). Thus, improve-
ments in energetic cost following increases in plantar flexor strength and AT
stiffness seem attributed to increased enthalpy efficiency of the operating
soleus muscle. The results further imply that the soleus energy production
in the first part of stance, when the MTU is lengthening, may be crucial
for the overall metabolic energy cost of running.
1. Introduction
Habitual bipedalism has been recognized as a defining feature of humans [1],
and an exceptional endurance running ability has been linked to the evolution
of the human lineage [2]. Economy, which is the mass-specific rate of oxygen
uptake or metabolic energy consumption at a given speed [3,4], plays a crucial
role in endurance running performance [5]. The cost of generating force and
work through muscles to support and accelerate the body mass is the main
source of metabolic energy expenditure during locomotion [6]. The force–
length–velocity potential of muscles (defined as the fraction of maximum
force according to the force–length [7] and force–velocity relationships [8]) at
which muscles operate during running [9,10] largely dictates the required
active muscle volume and consequently the energetic cost of contraction
[3,9,11].
In human running, the triceps surae is the major contributor to propulsion
and the main plantar flexor muscle group that transmits force through the
Achilles tendon (AT) [12], consuming a significant amount of metabolic
energy [13]. In earlier studies, we provided evidence that both the contractile
capacities of the triceps surae and the mechanical properties of the AT (i.e. its
stiffness) influence running economy [14,15]. We found that the most
© 2021 The Authors. Published by the Royal Society under the terms of the Creative Commons Attribution
License http://creativecommons.org/licenses/by/4.0/, which permits unrestricted use, provided the original
author and source are credited.
economical runners feature a combination of higher plantar
flexor muscle strength and AT stiffness [14], and that a
specific training of muscle strength and AT stiffness can, in
fact, improve running economy [15]. Although the associ-
ation of AT stiffness and energetic cost of running has been
confirmed by other research groups [16,17], the underlying
physiological mechanisms remain unclear.
The soleus is the greatest muscle of the triceps surae [18]
and generates the majority of work/energy to lift and accel-
erate the body [12] by actively shortening throughout the
entire stance phase of running [9,19]. In the first part of the
stance phase, the fascicle shortening is paralleled by a
lengthening of the muscle–tendon unit (MTU) [9], indicating
that a part of the body’s mechanical energy is stored as strain
energy in the AT, but also that the fascicles generate work and
save this work as strain energy in the AT. In the second part
of the stance phase, where the MTU shortens (propulsion
phase), the tendon strain energy is returned to the body
and contributes to the ongoing work generation [9]. The
metabolic cost of generating work by active shortening of
muscles depends on the velocity of the shortening [20]. The
enthalpy efficiency (or mechanical efficiency) quantifies the
fraction of chemical energy from ATP hydrolysis that is con-
verted into mechanical muscular work [21]. The relation of
enthalpy efficiency and shortening velocity shows a steep
increase at low velocities with the peak at around 20% of
the maximum shortening velocity [21,22]. During submaxi-
mal running, the soleus operates below the optimal velocity
for maximal efficiency [9], suggesting that small changes in
the shortening velocity may substantially influence the
enthalpy efficiency of the soleus muscular work production.
The mechanical interaction of the soleus muscle with the
series AT regulates the fascicle shortening dynamics. The AT
takes over a great part of the length changes of the entire
soleus MTU, thereby decoupling the muscle fascicle and
MTU behaviour and, beside the storage and release of
strain energy, allowing the fascicles to operate at velocities
favourable for economical force generation [9,19]. The mech-
anical properties of the tendon in combination with the
strength capacity of the muscle may determine the amount
of fascicle decoupling during the stance phase of running.
However, similar to an increase in muscle strength [23], ten-
dons can adapt to periods of higher mechanical loading by
increasing stiffness [24]. Our earlier findings of improved ener-
getic cost after an exercise-induced increase in AT stiffness and
plantar flexor muscle strength evidenced a direct association
between a balanced adaptation of tendon and muscle and
improvements in running economy [15]. Considering a given
work produced by the soleus muscle during the stance
phase, the energetic cost depends on the enthalpy efficiency
under which this muscular work is generated. Assuming
that a combination of increased plantar flexor strength and
AT stiffness may influence the soleus fascicle shortening pat-
tern, the overall enthalpy efficiency might improve. This
would provide an explaining mechanism to the previously
reported improvements in running economy following effec-
tive
muscle–tendon
training
[15].
To
the
best
of
our
knowledge, no study has experimentally examined the operat-
ing soleus muscle fascicles with respect to the enthalpy
efficiency and its association to the energetic cost of running.
Here, we investigated the effect of a specific muscle-
tendon training, which has been shown to increase plantar
flexor strength and AT stiffness [15], on the enthalpy
efficiency of the operating soleus fascicles during running.
Based on our earlier study [15], we expected an improvement
in running economy after 14 weeks of training. We hypoth-
esized that the training-induced increase in plantar flexor
muscle strength and AT stiffness modulates the soleus fascicle
velocity pattern throughout the stance phase towards vel-
ocities associated with a higher enthalpy efficiency, thereby
reducing the energetic cost of running.
2. Methods
(a) Participants and experimental design
A statistical power analysis was performed a priori and revealed a
required sample size of n = 12 for the intervention group (see
electronic supplementary material for details). Considering
potential dropouts, we recruited 36 participants and randomly
assigned them to an intervention (n = 19) or control group
(n = 17). Inclusion criteria were an age of 20–40 years, at least
two running sessions weekly on a recreational basis and no mus-
cular–tendinous injuries in the previous year. Only habitual
rearfoot-striking runners were considered because it is the most
common foot strike pattern [25] and also to avoid potential con-
founding effects of the strike pattern on our outcome measures.
To quantify the foot strike pattern, we assessed the strike index
[26]
(i.e. centre of pressure position with respect to the heel
relative to foot length at touchdown) during a pre-test session
(0 equals rearfoot-striking, <0.3 inclusion threshold). Twenty-
three participants completed the study, of which 13 were the
intervention group (age 29 ± 5 years, height 178 ± 8 cm, body
mass 73 ± 8 kg, four females) and 10 the control group (31 ±
3 years, 175 ± 10 cm, 70 ± 11 kg, seven females). For the interven-
tion group, the same 14-week muscle–tendon training was added
to the regular ongoing training habits as in our earlier study [15].
Before and after the intervention period, the maximal plantar
flexion moment and AT stiffness as well as energetic cost of run-
ning at 2.5 m s−1 were assessed in both groups. To explain the
expected improvements in energetic cost following the training,
we experimentally determined (i) the foot strike pattern, joint
kinematics and temporal gait parameters as well as (ii) the
soleus MTU and fascicle behaviour in addition to the electromyo-
graphic (EMG) activity during running. We further determined
(iii) the soleus force–fascicle length relationship and force–fasci-
cle velocity relationship in order to calculate the force–length
and force–velocity potential of the fascicles during running (i.e.
fraction of maximum force according to the force–length and
force–velocity curve [9,10,27]) and assessed (iv) the enthalpy effi-
ciency–fascicle velocity relationship to calculate the efficiency of
the soleus muscle during running. Because changes in running
economy were not expected without any intervention [15], the
assessment of the fascicle behaviour was not conducted in the
controls. The university ethics committee approved the study,
and participants gave written informed consent in accordance
with the Declaration of Helsinki.
(b) Exercise protocol
The supervised and biofeedback-based resistance training was
performed for 14 weeks and was characterized by five sets of
four repetitive isometric ankle plantar flexion contractions (3 s
loading and 3 s relaxation) at 90% of the maximum voluntary
contraction (MVC) strength (adjusted every two weeks), three
to four times a week (see electronic supplementary material
for illustration). This loading regimen has been shown to pro-
vide a sufficient magnitude and duration of tendon strain to
promote AT adaptation in addition to increases in plantar
flexor muscle strength [15,24,28].
royalsocietypublishing.org/journal/rspb
Proc. R. Soc. B 288: 20202784
2
(c) Strength of the plantar flexors and Achilles
tendon stiffness
The plantar flexor strength of the right leg was measured using
an inverse dynamics approach. For the determination of AT stiff-
ness, ramp-MVCs were conducted and the force applied to the
AT was calculated as quotient of joint moment and individual
tendon lever arm, which was determined using the tendon-
excursion method. The corresponding AT elongation was ana-
lysed
based
on
the
displacement
of
the
gastrocnemius
medialis-myotendinous junction visualized by ultrasonography.
Stiffness was calculated between 50 and 100% of the maximum
tendon force and strain by dividing elongation by resting
length (see electronic supplementary material for details).
(d) Energetic cost of running
During an 8 min running trial on a treadmill at 2.5 m s−1, expired
gas analysis was conducted and the rate of oxygen consumption
( _VO2) and carbon dioxide production ( _VCO2) was calculated as
the average of the last 3 min [15]. Running economy was then
expressed in units of energy as Energetic cost ¼ 16:89 _VO2 þ
4:84 _VCO2, where energetic cost is presented in (W kg−1) and
_VO2 and _VCO2 in (ml s−1 kg−1) [4,29]. The steady state was visually
confirmed by the rate of _VO2 during each trial, and a respiratory
exchange ratio (RER) of <1.0 was controlled for during the post
analysis (see electronic supplementary material for details).
(e) Joint kinematics and foot strike pattern
Kinematics of the right leg were captured (250 Hz) by a Vicon
motion capture system (Nexus 1.8, Vicon, Oxford, UK) using ana-
tomical-referenced markers [9]. The touchdown and the toe-off
were determined from the kinematic data as consecutive minimum
in knee joint angle over time [30]. The foot strike pattern was ana-
lysed by means of the strike index [26]. A self-developed
algorithm [25] was used to calculate the strike index from the plantar
pressure distribution (120 Hz) captured by the integrated pressure
plate (FDM-THM-S, Zebris Medical GmbH, Isny, Germany).
(f) Soleus muscle-tendon unit length changes,
fascicle behaviour and electromyographic activity
during running
During an additional 3 min running trial at the same speed, kin-
ematics of the ankle joint served to calculate the length change of
the soleus MTU as the product of ankle angle changes and the
previously assessed individual AT lever arm [31]. The initial
soleus MTU length was determined at a neutral joint angle
using the previously reported regression equation by Hawkins &
Hull [32]. Ultrasonic images of the soleus muscle fascicles were
obtained
synchronously at
146 Hz (Aloka
Alpha7,
Tokyo,
Japan). The probe (6 cm linear array, 13.3 MHz) was mounted
over the medial soleus muscle belly. The fascicle length was
post-processed from the images using a semi-automatic tracking
algorithm [33] (figure 1), and corrections were made if necessary.
At least nine steps were averaged [10]. The velocities of MTU and
fascicles were calculated as the first derivative of the lengths over
the time. Synchronized surface EMG of soleus was measured
(1000 Hz) by means of a wireless EMG system (Myon m320RX,
Baar, Switzerland) and is presented as normalized to the
maximum EMG value observed from the individual MVCs [9].
(g) Soleus force–length, force–velocity and efficiency–
velocity relationship
To determine the soleus force–fascicle length relationship (for
details [9]), the participants were placed in the prone position on
the bench of the dynamometer (Biodex Medical, Shirley, NY) with
the knee fixed in a flexed position (figure 1) to restrict the contri-
bution of the bi-articular gastrocnemius muscle to the plantar
flexion moment (approx. 120°) [34]. MVCs were performed with
the right leg in eight different joint angles, and the joint moments
and force acting on the AT were calculated as described in section
2c above. The corresponding soleus fascicle behaviour was cap-
tured synchronously at 30 Hz by ultrasonography, and fascicle
length was measured accordingly (figure 1). The probe remained
attached between the running trial and MVCs. An individual
force–fascicle length relationship was calculated by means of a
second-order polynomial fit (figure 1), giving the maximum force
(Fmax) and optimal fascicle length for force generation (L0).
The force–velocity relationship of the soleus was assessed
using the classical Hill equation [8] and the maximum fascicle
shortening velocity (Vmax) and constants of arel and brel. For
Vmax, we took reported values of human soleus type 1 and 2
fibres [35], adjusted those for physiological temperature [36] and
applied an average fibre-type distribution (81% type 1 fibres and
19% type 2 [9]), giving Vmax as 6.77 L0 s−1 [9]. arel was calculated
as 0.1 + 0.4 × type 2 fibre percentage [37], which equals to 0.175.
The product of arel and Vmax gives brel as 1.182 [37]. Based on
the assessed force–length and force–velocity relationships, it
was possible to calculate the individual force–length and force–
velocity potential of soleus as a function of the fascicle length
(figure 1) and velocity during running [9,10,27].
Furthermore,
we
determined
the
enthalpy
efficiency–
shortening velocity relationship for the soleus fascicles to
calculate the enthalpy efficiency of the soleus as a function of
the
fascicle
velocity
during
running.
We
referred
to
the
fascicle length (mm)
5000
4000
3000
2000
1000
0
force (N)
20
40
60
80
(a)
(b)
Figure 1. (a) Experimental set-up for the determination of the soleus force–
fascicle length relationship. MVCs at eight different joint angles were per-
formed on a dynamometer. During the MVCs, the soleus muscle fascicle
length was measured by ultrasonography as an average (F) of multiple
fascicle portions (short-dashed white lines) identified from the images.
(b) Exemplary force–length relationship of the soleus fascicles obtained
from the MVCs (squares) and the respective second-order polynomial fit
(dashed line).
royalsocietypublishing.org/journal/rspb
Proc. R. Soc. B 288: 20202784
3
experimental efficiency values provided by the paper of Hill [20],
where the values are presented as a function of relative load
which we then transposed to the shortening velocity (normalized
to Vmax) using the classical Hill equation [8]. The corresponding
values of enthalpy efficiency and shortening velocity were fitted
using a cubic spline, giving the right-skewed parabolic-shaped
curve with a peak efficiency of 0.45 at a velocity of 0.18 Vmax.
The resulting function was then used to calculate the soleus
efficiency during running.
(h) Statistics
An analysis of variance for repeated measures including post hoc
analysis (adjusted p-values reported) was performed for the
group comparison. Anthropometric group differences as well
as baseline differences of the plantar flexion moment, AT stiffness
and energetic cost were tested using a t-test for independent
samples. A paired t-test was used to analyse the training effects
on the assessed gait characteristics, kinematics and MTU and fas-
cicle parameters. The level of significance was set to α = 0.05.
Effect sizes (Hedges’s g) assess the strength of the intervention
effects (see electronic supplementary material for details).
3. Results
There were no significant differences in age (p = 0.421), height
( p = 0.361) and body mass ( p = 0.382) between the interven-
tion and control groups. No baseline differences between
groups were observed for the maximum plantar flexion
moment ( p = 0.894), AT stiffness ( p = 0.421) and energetic
cost ( p = 0.143; table 1). Both the plantar flexion moment
and AT stiffness increased significantly in the intervention
group ( p = 0.024, p = 0.048) without significant changes in
the controls ( p = 0.296, p = 0.745; table 1). Furthermore, we
found a significant decrease in the energetic cost of running
following the 14 weeks of training in the intervention group
( p = 0.028) and no significant changes in the control group
( p = 0.688; table 1). Neither group showed any significant
changes in the strike index (intervention p = 0.868, control
p = 0.868), stance time ( p = 0.283, p = 0.283), flight time ( p =
0.981, p = 0.252) and cadence ( p = 0.310, p = 0.384; table 1)
after training, indicating that the training intervention did
not influence the foot strike pattern.
Following the intervention, ankle and knee joint kin-
ematics did not significantly change during the stance
phase, i.e. joint angles at touchdown (ankle p = 0.108, knee
p = 0.064), toe-off ( p = 0.161, p = 0.844), maximal ankle dorsi-
flexion
(p = 0.576)
and
maximal
knee
flexion
(p = 0.138;
table 2 and figure 2). The soleus MTU showed a lengthening–
shortening behaviour during the stance phase, with shortening
starting at 59 ± 2% of the stance phase similarly pre- and post-
intervention (p = 0.266, g = 0.30; see the Statistics section;
figure 3). The training had no effect on the MTU length,
length changes and velocity, neither when averaged over the
entire stance phase (p = 0.943, p = 0.273, p = 0.274) nor over
the
subphase
of
MTU
lengthening
(p = 0.931,
p = 0.893,
p = 0.788) or MTU shortening (p = 0.946, p = 0.470, p = 0.189;
table 3 and figure 3). Despite the MTU lengthening, the
soleus muscle fascicles shortened continuously throughout
the entire stance phase (figure 3). Following the intervention,
the fascicle shortening was not significantly different over the
entire stance phase (p = 0.662) and the phase of MTU
Table 1. Maximal plantar flexion moment and Achilles tendon stiffness as well as energetic cost, foot strike index and temporal step characteristics during
running before and after the training period for the intervention and control group (mean ± s.d., effect size g).
intervention (n = 13)
control (n = 10)
pre
post
g
pre
post
g
moment (Nm kg−1)a
3.12 ± 0.48
3.44 ± 0.37c
0.77
3.10 ± 0.46
2.99 ± 0.32
0.32
stiffness (kN strain−1)a
85 ± 36
111 ± 59c
0.67
73 ± 29
71 ± 28
0.10
energy cost (W kg−1)b
10.6 ± 0.6
10.2 ± 0.7c
0.74
11.2 ± 1.0
11.1 ± 1.0
0.12
strike index
0.08 ± 0.12
0.10 ± 0.16
0.09
0.06 ± 0.03
0.06 ± 0.03
0.05
stance time (ms)
310 ± 23
316 ± 23
0.29
327 ± 17
324 ± 23
0.34
flight time (ms)
53 ± 31
53 ± 24
0.01
50 ± 31
54 ± 31
0.48
cadence (steps min−1)
160 ± 11
159 ± 9
0.39
162 ± 9
161 ± 9
0.26
aSignificant time by group interaction effect (p < 0.05).
bSignificant main effect of time (p < 0.05).
cSignificant difference (post hoc analysis) to pre (p < 0.05).
Table 2. Ankle and knee joint angles at touchdown, toe-off and at the maximal ankle dorsiflexion and knee flexion angle, respectively, during running before
and after the training intervention (mean ± s.d., effect size g, n = 13).
touchdown
toe-off
maximum dorsiflexion/knee flexion
pre
post
g
pre
post
g
pre
post
g
ankle joint (°)
−1.3 ± 5.1
−0.0 ± 6.2
0.45
13.7 ± 7.8
15.1 ± 6.0
0.39
−18.0 ± 3.7
−18.4 ± 4.4
0.15
knee joint (°)
−3.7 ± 3.9
−6.5 ± 6.0
0.53
−11.6 ± 4.5
−11.3 ± 4.6
0.05
−32.8 ± 5.8
−34.6 ± 5.8
0.41
royalsocietypublishing.org/journal/rspb
Proc. R. Soc. B 288: 20202784
4
lengthening (p = 0.106) but in the phase of MTU shortening
(p = 0.016; table 3). L0 (pre 43.1 ± 5.7 mm, post 44.1 ± 8.9 mm,
p = 0.767, g = 0.08) and thus Vmax (pre 291 ± 38 mm s−1, post
298 ± 17 mm s−1, p = 0.767, g = 0.08) were not significantly
altered due to training. The operating fascicle length averaged
over the stance phase (pre 0.87 ± 0.11 L0, post 0.85 ± 0.13 L0,
p = 0.360, g = 0.16), but also during MTU lengthening (pre
0.92 ± 0.12 L0, post 0.91 ± 0.15 L0, p = 0.772, g = 0.07) and short-
ening (pre 0.81 ± 0.10 L0, post 0.76 ± 0.11 L0, p = 0.226, g =
0.32), was not significantly changed following training. Conse-
quently,
the
force–length
potential
was
not significantly
different between pre- and post-training in the different
phases (stance p = 0.172, g = 0.14, MTU lengthening p = 0.713,
g = 0.10, MTU shortening p = 0.640, g = 0.12; figure 4).
After training, the soleus force–velocity potential was sig-
nificantly lower in the phase of MTU lengthening ( p = 0.030,
g = 0.64) and significantly higher when the MTU shortened
( p = 0.045, g = 0.58) with no significant difference over the
entire stance ( p = 0.249, g = 0.31; figure 4). This was the conse-
quence of a tendency towards higher fascicle shortening
velocity during MTU lengthening (pre −0.088 ± 0.054 Vmax,
post −0.129 ± 0.061 Vmax, p = 0.073, g = 0.51) and a signifi-
cantly lower velocity during MTU shortening after training
(pre −0.174 ± 0.057 Vmax, post −0.127 ± 0.008 Vmax, p =
0.007, g = 0.83). Furthermore, the averaged EMG activation
over the phase of MTU shortening ( p = 0.028, g = 0.67) and
the entire stance phase was significantly reduced following
the intervention ( p = 0.017, g = 0.60; figures 3 and 4). Com-
pared with pre-intervention running, the fascicle velocity in
the phase of MTU lengthening was closer to the velocity
for optimal enthalpy efficiency after the training (figure 5).
Consequently, the fascicles operated at a significantly higher
enthalpy efficiency in the phase of MTU lengthening after
the training ( p = 0.006, g = 0.85; figures 5 and 6), while there
was no significant pre–post difference in the phase of MTU
shortening ( p = 0.640, g = 0.12; figure 6). Over the entire
stance phase of running, the efficiency of the fascicle
0
20
40
60
80
100
stance phase (%)
–30
–20
–10
0
10
20
30
ankle angle (°)
pre
post
0
20
40
60
80
100
stance phase (%)
–50
–40
–30
–20
–10
0
10
knee angle (°)
pre
post
dorsiflexion
plantar flexion
flexion
extension
(a)
(b)
Figure 2. (a) Ankle joint angle and (b) knee joint angle during the stance
phase of running before and after the training intervention (mean ± s.e.m.,
n = 13).
20
40
60
80
100
stance phase (%)
0
0.2
0.4
0.6
0.8
1.0
EMGnorm
pre
post
0
20
40
60
80
100
stance phase (%)
20
30
40
50
60
fascicle length (mm)
pre
post
0
20
40
60
80
100
stance phase (%)
280
300
320
340
360
MTU length (mm)
pre
post
*
(a)
(b)
(c)
Figure 3. (a) Soleus MTU length, (b) muscle fascicle length and (c) EMG
activity (normalized to a maximum voluntary isometric contraction, during
the stance phase of running before and after the training intervention
(mean ± s.e.m., n = 13). *Significant difference of the stance phase-averaged
EMG activation between pre and post (p < 0.05).
royalsocietypublishing.org/journal/rspb
Proc. R. Soc. B 288: 20202784
5
shortening was also significantly increased following the
training ( p = 0.025, g = 0.66; figure 6).
4. Discussion
Our current study showed for the first time that specific
muscle–tendon training that increases plantar flexor muscle
strength and AT stiffness facilitates the enthalpy efficiency
of the soleus muscle during the stance phase of running.
The increased enthalpy efficiency was found in the first
part of the stance phase where the soleus muscle produces
work by active shortening and transfers muscular work to
the tendon as strain energy. Furthermore, the results provide
additional evidence that a combination of greater plantar
flexor muscle strength and AT stiffness decreases the energy
cost of running [14,15] and indicates that the soleus enthalpy
efficiency is a contributive determinant.
Following the intervention, the energetic cost of running
was significantly reduced by about 4%, a quantity reported
to be above test–retest typical errors [38] and to substantially
enhance endurance running performance [39]. At the same
time, the soleus, which is the main muscle for work/energy
production during running [12], operated at a significantly
increased (7%) enthalpy efficiency throughout the stance
phase. The enthalpy efficiency quantifies the portion of
energy from ATP hydrolysis used by a muscle that is con-
verted
into
mechanical
muscular
work
[21].
Enthalpy
efficiency depends on the velocity of muscle shortening
with a steep increase at low velocities until the peak at
around 0.18 Vmax and again decreasing at higher shortening
velocities [20,21]. For the whole stance phase, fascicle short-
ening, the force–length potential and the force–velocity
potential of the soleus muscle were not significantly different
before and after the intervention, indicating a similar energy
production through muscular work of the soleus muscle.
During the propulsion phase of running (i.e. MTU shorten-
ing), where both tendon and muscle transfer energy/work
to the skeleton [19,40], the enthalpy efficiency of the operat-
ing soleus muscle was high pre- and post-intervention (94%
and 93% of the maximum efficiency). By contrast, during
the first part of the stance phase (i.e. MTU lengthening),
where energy is transferred from the contractile element to
the tendon, the efficiency was lower during pre-intervention
running (77% of the maximum efficiency). The relevant part
of the soleus fascicle shortening occurred during this first
part of stance (59% of the entire shortening range). In combi-
nation with the high muscle activation (higher during MTU
lengthening than during MTU shortening), this indicates an
important
energy
production
through
muscular
work
during the phase of MTU lengthening.
The exercise-induced increase in plantar flexor muscle
strength and AT stiffness was associated with an alteration
of the operating fascicle velocity profile and a significant
increase of the enthalpy efficiency of the soleus in the phase
of MTU lengthening (88% of maximum), potentially improv-
ing the enthalpy efficiency of muscular work production. The
significant increase of the enthalpy efficiency following train-
ing in the phase of MTU lengthening demonstrates that a
substantial part of the entire muscular work was generated
more economically. In the second part of the stance phase,
where the MTU shortened, the high efficiency was main-
tained after the intervention and, further, the fascicles
operated at a significantly higher force–velocity potential.
This was possible due to a shift of the shortening velocity
around the plateau of the efficiency–velocity curve, from
the descending part before the training to the ascending
part after training (figure 5), without a significant decline in
the efficiency. Consequently, the overall enthalpy efficiency
throughout the stance phase of each step was increased.
The phase of MTU shortening was accompanied by a
reduced soleus EMG activation after the intervention, and
the overall EMG activity during the stance phase was signifi-
cantly lower as well (12%). However, the higher maximum
plantar flexion moment along with no significant changes
in EMGmax during the MVCs (pre 0.409 ± 0.114 mV, post
0.410 ± 0.092 mV, p = 0.300) and antagonistic co-activation
(tibialis anterior EMG pre 0.034 ± 0.016 mV, post 0.034 ±
0.013 mV, p = 0.923) as measures for neural adaption after
training strongly indicate muscle hypertrophy, resulting in a
13% increase of Fmax. Therefore, the reductions in EMG acti-
vation may not correspond to a reduced active muscle
volume. To examine this possibility, we calculated the aver-
age force of the soleus muscle (Fs) during the stance phase,
adopting a ‘Hill-type muscle model’ as a function of the aver-
age force–length potential (λl), force–velocity potential (λv),
Table 3. Soleus MTU length, length changes and velocity as well as muscle fascicle length, fascicle shortening distance and fascicle velocity averaged over the
phase of MTU lengthening, MTU shortening and over the entire stance phase during running before and after the training intervention (mean ± s.d., effect size
g, n = 13).
MTU lengthening
MTU shortening
stance phase
pre
post
g
pre
post
g
pre
post
g
MTU length (mm)
325 ± 20
325 ± 21
0.02
323 ± 20
323 ± 21
0.02
324 ± 20
324 ± 21
0.02
MTU length changes (mm)
18.4 ± 2.0
18.2 ± 3.2
0.04
−33.9 ± 9.3
−32.5 ± 5.6
0.19
−16.4 ± 9.0
−14.8 ± 5.6
0.30
MTU velocity (mm s−1)
97 ± 15
98 ± 22
0.07
−259 ± 52
−244 ± 33
0.36
−173 ± 29
−164 ± 21
0.30
fascicle length (mm)
39.2 ± 4.4
39.0 ± 5.1
0.03
34.5 ± 4.3
33.1 ± 4.5
0.23
37.2 ± 4.3
36.5 ± 4.8
0.12
fascicle shortening (mm)
−5.21 ± 2.68
−6.75 ± 3.08
0.45
−6.49 ± 2.02
−4.98 ± 1.23a
0.72
−11.05 ± 3.32
−11.53 ± 3.47
0.12
fascicle velocity (mm s−1)
−21.2 ± 16.7
−33.4 ± 17.5
0.52
−49.1 ± 16.7
−35.8 ± 10.1a
0.71
−33.0 ± 10.8
−34.6 ± 11.0
0.10
aSignificant difference to pre (p < 0.05).
royalsocietypublishing.org/journal/rspb
Proc. R. Soc. B 288: 20202784
6
EMG activity (α) and Fmax (Fs ¼ lllva FmaxÞ. The average
force of the soleus muscle after the intervention (Fs = 353 ±
122 N) did not show significant differences compared with
the pre-values (Fs = 372 ± 112 N, p = 0.660), indicating a simi-
lar active muscle volume. Similarly, the rate of muscle force
generation during the stance phase (_Fs ¼ Fs=tstance) did not
differ before (_Fs = 1215 ± 413 N s−1) and after the intervention
(_Fs = 1126 ± 400 N s−1, p = 0.498). The above assessments
suggest that the active muscle volume and the rate of
muscle force generation were not the reason for the improved
running economy, but rather the increase in soleus muscle
operating enthalpy efficiency.
Previous studies provided evidence that the cost of force
to support the body mass and the time course of force appli-
cation to the ground are the major determinants of the
energetic cost of running [6,41]. According to the ‘cost of
generating force hypothesis’ [6], the rate of metabolic
energy consumption is directly related to the body mass
and the time available to generate force, which results in a
constant cost coefficient (i.e. energy required per unit force).
However, modifications in the muscle effective mechanical
advantage (i.e. ratio of the muscle moment arm to the
moment arm of the ground reaction force [42]) within the
lower extremities can influence the cost coefficient of loco-
motion [43,44]. In our study, the metabolic energy cost of
running was reduced after the training without any changes
in the contact time and body mass, indicating a decrease of
the cost coefficient. The similar strike index and lower leg kin-
ematics before and after the intervention suggest unchanged
effective mechanical advantages within the lower extremities;
therefore, this would not be the reason for the reduced cost
coefficient. Instead, our findings show that an adjusted time
course of the soleus shortening velocity during the stance
phase following the training can influence the cost coefficient
as a result of increased enthalpy efficiency of the soleus and,
thus, complement the earlier studies on the mechanical
advantage
and
cost
coefficient
interaction
[41,42].
The
observed continuous soleus fascicle shortening during the
stance phase is in agreement with other experiments using
the
ultrasound
methodology
and
comparable
running
speeds [19,45]. The importance of the energy production by
the plantar flexor muscles for the propulsion phase (i.e. short-
ening of the MTU) during running is well accepted [19,46],
because the mechanical power produced at the ankle joint
in this phase is highest and determines running performance
[47]. Our current results regarding the enthalpy efficiency of
muscular energy generation and running economy show for
the first time that also the phase of the MTU lengthening is
crucial for the overall metabolic energy consumption during
running. Recently, findings of our group [9] but also others
[48,49] provided evidence that soleus muscle dynamics may
improve the economy of locomotion by a modulation of the
force–length–velocity potential, thus decreasing the active
muscle volume. In the present study, the soleus force–
length–velocity potential throughout stance was not signifi-
cantly changed following the intervention, while in the same
time the adjusted time course of the shortening velocity
increased the efficiency of muscle work production. Thus,
the present study expands the importance of the soleus fas-
cicle dynamics towards the efficiency–velocity dependency
as a further factor for improvements of locomotor economy.
The findings of the current study provide further evi-
dence [15,16] that strength training of the plantar flexors
has the potential to enhance running economy. We used a
specific high-intensity muscle–tendon training programme
[24,28], targeting an adaptation of both AT stiffness and plan-
tar flexor muscle strength [14,15], to maintain the functional
integrity of the contractile and series elastic element. Strength
increases without concomitant stiffening of the AT after a
period of training can increase levels of operating and maxi-
mum
strain
[24],
which
have
been
associated
with
pathologies [50], and also possible functional decline [51].
On the other hand, increased stiffness without higher
muscle strength may also limit function by reducing relevant
operating tendon strains [51]. In our study, the maximum AT
strain during the MVCs was not affected by the training (pre
6.2 ± 1.6%, post 6.0 ± 1.2%, p = 0.501) despite an increase in
the plantar flexor muscle strength, indicating a balanced
adaptation of muscle and tendon. Therefore, a specific
1.0
0.8
0.6
0.4
0.2
force–length potential
MTU
lengthening
MTU
shortening
stance
phase
MTU
lengthening
MTU
shortening
stance
phase
MTU
lengthening
MTU
shortening
stance
phase
1.0
0.8
0.6
0.4
0.2
EMGnorm
*
1.0
0.8
0.6
0.4
0.2
force–velocity potential
pre
post
*
*
*
(a)
(b)
(c)
Figure 4. (a) Soleus fascicle force–length potential, (b) force–velocity poten-
tial and (c) EMG activity (normalized to a maximum voluntary isometric
contraction, averaged over the phase of MTU lengthening, MTU shortening
and the entire stance phase of running before and after the training inter-
vention (n = 13). *Significant difference between pre and post ( p < 0.05).
royalsocietypublishing.org/journal/rspb
Proc. R. Soc. B 288: 20202784
7
muscle-tendon training [24,28] can be recommended to
improve running economy.
To assess the enthalpy efficiency–shortening velocity
relationship, we used a biologically founded value of Vmax
(i.e. 6.77 L0 s−1). However, during submaximal running, the
lower activation level and selective slow fibre-type recruit-
ment
may
affect
the
actual
relationship.
Furthermore,
differences in fibre-type distribution may also affect the
shape of the enthalpy efficiency–shortening velocity curve
[22]. We evaluated the effect of (i) decreasing Vmax by 10%
intervals and (ii) replacing the underlying efficiency values
measured at the frog sartorius at 0°C from Hill [20] by the
data presented by Barclay et al. [22] for the predominantly
slow fibre-type soleus mouse muscle at 21°C, comparable
with the human soleus muscle. The significant pre- to post-
enthalpy efficiency increase for the MTU lengthening phase
and the entire stance phase persisted for values till Vmax−30%
both using the data of Hill or Barclay et al. ( p < 0.05), which
confirms and strengthens the observed intervention effect
(for descriptive values and p-values see electronic supplemen-
tary material, S2). Furthermore, since we calculated the
efficiency as a function of the soleus muscle shortening vel-
ocity (adjusted for physiological temperature) and only
discussed our findings in terms of percentage change, any
uncertainties about the magnitude of the enthalpy efficiency
would not affect our results. The soleus fascicle dynamics
were not assessed in the control group because alterations
were not expected with continued training habits as pre-
viously evidenced [45]. Furthermore, the controls did not
show alterations in any of the assessed parameters, giving
strong support for an unchanged fascicle behaviour after the
intervention period.
5. Conclusion
In conclusion, the current study gives new insights into the
soleus muscle mechanics and metabolic energetics during
human running. In support of our earlier study, an exercise-
induced increase of plantar flexor muscle strength and AT
stiffness reduced the metabolic energy cost of running. The
proposed reason for this improvement is an alteration in the
soleus fascicle velocity profile throughout the stance phase,
which led to a significantly higher enthalpy efficiency of the
operating soleus muscle. The enthalpy efficiency was particu-
larly increased in the phase of MTU lengthening, where the
activation is high and the soleus generates an important part
of the mechanical energy required for running.
Ethics. The ethics committee of the Humboldt-Universität zu Berlin
approved the study and the participants gave written informed
consent in accordance with the Declaration of Helsinki.
Data accessibility. The processed datasets generated and analysed
during the current study are available as part of the electronic
supplementary material..
Authors’ contributions. S.B., F.M., A.S. and A.A. designed research. S.B.,
F.M. and A.S. performed research. S.B. analysed data. S.B. and A.A.
drafted the manuscript. F.M. and A.S. made important intellectual
contributions during revision.
Competing interests. We declare we have no competing interests.
0.2
0.4
0.6
0.8
1.0
Vnorm (V/Vmax)
Vnorm (V/Vmax)
0
0.1
0.2
0.3
0.4
enthalpy efficiency
pre
post
0
20
40
60
80
100
stance phase (%)
–0.2
–0.1
0
0.1
0.2
0.3
0.4
pre
post
}*
MTU lengthening
MTU shortening
(a)
(b)
Figure 5. (a) Soleus muscle fascicle operating velocity over the stance phase of running before and after the intervention (mean ± s.e.m.) and velocity of maximum
enthalpy efficiency (i.e. 0.18 Vmax, horizontal dashed line). Following the intervention, the fascicle shortening velocity was closer to the velocity optimal for maxi-
mum enthalpy efficiency during most of the MTU lengthening phase. (b) Enthalpy efficiency–fascicle velocity relationship with average values of the phase of MTU
lengthening, showing that the fascicles operated at a significantly higher enthalpy efficiency following the intervention (*p < 0.05). Circles indicate that the single
participant values before (white) and after (black) the intervention and squares show the respective mean with standard error bars (n = 13). The vertical dotted line
shows the velocity of maximum efficiency.
enthalpy efficiency
0.50
0.45
0.40
0.35
0.30
0.25
pre
post
MTU
lengthening
MTU
shortening
stance
phase
0
*
*
Figure 6. Soleus muscle fascicle enthalpy efficiency averaged over the phase
of MTU lengthening, MTU shortening and the entire stance phase of running
before and after the training intervention (n = 13). *Significant difference
between pre and post ( p < 0.05).
royalsocietypublishing.org/journal/rspb
Proc. R. Soc. B 288: 20202784
8
Funding. Funding for this research was supplied by the German
Federal Institute of Sport Science (grant no. ZMVI14-070604/
17-18).
Acknowledgements. We acknowledge the support of Antonis Ekizos,
Arno Schroll, Leon Brüll and Victor Munoz-Martel for data recording
and analysis.
References
1.
Pontzer H. 2017 Economy and endurance in human
evolution. Curr. Biol. 27, R613–R621. (doi:10.1016/
j.cub.2017.05.031)
2.
Bramble DM, Lieberman DE. 2004 Endurance
running and the evolution of Homo. Nature 432,
345–352. (doi:10.1038/nature03052)
3.
Fletcher JR, MacIntosh BR. 2017 Running economy
from a muscle energetics perspective. Front. Physiol.
8, 433. (doi:10.3389/fphys.2017.00433)
4.
Kipp S, Byrnes WC, Kram R. 2018 Calculating
metabolic energy expenditure across a wide range
of exercise intensities: the equation matters. Appl.
Physiol. Nutr. Metab. Physiol. Appl. Nutr. Metab. 43,
639–642. (doi:10.1139/apnm-2017-0781)
5.
Joyner MJ. 1991 Modeling: optimal marathon
performance on the basis of physiological factors.
J. Appl. Physiol. 70, 683–687. (doi:10.1152/jappl.
1991.70.2.683)
6.
Kram R, Taylor CR. 1990 Energetics of running: a
new perspective. Nature 346, 265–267. (doi:10.
1038/346265a0)
7.
Gordon AM, Huxley AF, Julian FJ. 1966 The variation
in isometric tension with sarcomere length in
vertebrate muscle fibres. J. Physiol. 184, 170–192.
8.
Hill AV. 1938 The heat of shortening and the dynamic
constants of muscle. Proc. R. Soc. Lond. B Biol. Sci.
126, 136–195. (doi:10.1098/rspb.1938.0050)
9.
Bohm S, Mersmann F, Santuz A, Arampatzis A. 2019
The force–length–velocity potential of the human
soleus muscle is related to the energetic cost of
running. Proc. R. Soc. B Biol. Sci. 286, 20192560.
(doi:10.1098/rspb.2019.2560)
10. Bohm S, Marzilger R, Mersmann F, Santuz A,
Arampatzis A. 2018 Operating length and velocity of
human vastus lateralis muscle during walking and
running. Sci. Rep. 8, 5066. (doi:10.1038/s41598-
018-23376-5)
11. Roberts TJ, Marsh RL, Weyand PG, Taylor CR. 1997
Muscular force in running turkeys: the economy of
minimizing work. Science 275, 1113–1115. (doi:10.
1126/science.275.5303.1113)
12. Hamner SR, Delp SL. 2013 Muscle contributions to
fore-aft and vertical body mass center accelerations
over a range of running speeds. J. Biomech. 46,
780–787. (doi:10.1016/j.jbiomech.2012.11.024)
13. Fletcher JR, MacIntosh BR. 2015 Achilles tendon
strain energy in distance running: consider the
muscle energy cost. J. Appl. Physiol. 118, 193–199.
(doi:10.1152/japplphysiol.00732.2014)
14. Arampatzis A, De Monte G, Karamanidis K, Morey-
Klapsing G, Stafilidis S, Brueggemann G-P. 2006
Influence of the muscle-tendon unit’s mechanical
and morphological properties on running economy.
J. Exp. Biol. 209, 3345–3357. (doi:10.1242/jeb.
02340)
15. Albracht K, Arampatzis A. 2013 Exercise-induced
changes in triceps surae tendon stiffness and
muscle strength affect running economy in humans.
Eur. J. Appl. Physiol. 113, 1605–1615. (doi:10.1007/
s00421-012-2585-4)
16. Fletcher JR, Esau SP, MacIntosh BR. 2010 Changes in
tendon stiffness and running economy in highly
trained distance runners. Eur. J. Appl. Physiol. 110,
1037–1046. (doi:10.1007/s00421-010-1582-8)
17. Rogers SA, Whatman CS, Pearson SN, Kilding AE.
2017 Assessments of mechanical stiffness and
relationships to performance determinants in
middle-distance runners. Int. J. Sports Physiol.
Perform. 12, 1329–1334. (doi:10.1123/ijspp.2016-
0594)
18. Albracht K, Arampatzis A, Baltzopoulos V. 2008
Assessment of muscle volume and physiological
cross-sectional area of the human triceps surae
muscle in vivo. J. Biomech. 41, 2211–2218. (doi:10.
1016/j.jbiomech.2008.04.020)
19. Lai A, Lichtwark GA, Schache AG, Lin Y-C, Brown
NAT, Pandy MG. 2015 In vivo behavior of the
human soleus muscle with increasing walking and
running speeds. J. Appl. Physiol. 118, 1266–1275.
(doi:10.1152/japplphysiol.00128.2015)
20. Hill AV. 1964 The efficiency of mechanical power
development during muscular shortening and its
relation to load. Proc. R. Soc. Lond. B Biol. Sci. 159,
319–324. (doi:10.1098/rspb.1964.0005)
21. Barclay CJ. 2015 Energetics of contraction. Compr.
Physiol. 5, 961–995. (doi:10.1002/cphy.c140038)
22. Barclay CJ, Constable JK, Gibbs CL. 1993 Energetics
of fast- and slow-twitch muscles of the mouse.
J. Physiol. 472, 61–80. (doi:10.1113/jphysiol.1993.
sp019937)
23. Folland DJP, Williams AG. 2007 Morphological and
neurological contributions to increased strength.
Sports Med. 37, 145–168. (doi:10.2165/00007256-
200737020-00004)
24. Arampatzis A, Karamanidis K, Albracht K. 2007
Adaptational responses of the human Achilles
tendon by modulation of the applied cyclic strain
magnitude. J. Exp. Biol. 210, 2743–2753. (doi:10.
1242/jeb.003814)
25. Santuz A, Ekizos A, Arampatzis A. 2016 A pressure
plate-based method for the automatic assessment
of foot strike patterns during running. Ann. Biomed.
Eng. 44, 1646–1655. (doi:10.1007/s10439-015-
1484-3)
26. Cavanagh PR, Lafortune MA. 1980 Ground reaction
forces in distance running. J. Biomech. 13, 397–406.
(doi:10.1016/0021-9290(80)90033-0)
27. Nikolaidou ME, Marzilger R, Bohm S, Mersmann F,
Arampatzis A. 2017 Operating length and velocity of
human M. vastus lateralis fascicles during vertical
jumping. R. Soc. Open Sci. 4, 170185. (doi:10.1098/
rsos.170185)
28. Bohm S, Mersmann F, Tettke M, Kraft M, Arampatzis
A. 2014 Human Achilles tendon plasticity in response
to cyclic strain: effect of rate and duration. J. Exp.
Biol. 217, 4010–4017. (doi:10.1242/jeb.112268)
29. Péronnet F, Massicotte D. 1991 Table of nonprotein
respiratory quotient: an update. Can. J. Sport Sci.
16, 23–29.
30. Fellin RE, Rose WC, Royer TD, Davis IS. 2010
Comparison of methods for kinematic identification
of footstrike and toe-off during overground and
treadmill running. J. Sci. Med. Sport 13, 646–650.
(doi:10.1016/j.jsams.2010.03.006)
31. Lutz GJ, Rome LC. 1996 Muscle function during
jumping in frogs. I. Sarcomere length change, EMG
pattern, and jumping performance. Am. J. Physiol.
271, C563–C570. (doi:10.1152/ajpcell.1996.271.2.
c563)
32. Hawkins D, Hull ML. 1990 A method for
determining lower extremity muscle-tendon lengths
during flexion/extension movements. J. Biomech.
23, 487–494. (doi:10.1016/0021-9290(90)90304-L)
33. Marzilger R, Legerlotz K, Panteli C, Bohm S,
Arampatzis A. 2018 Reliability of a semi-automated
algorithm for the vastus lateralis muscle
architecture measurement based on ultrasound
images. Eur. J. Appl. Physiol. 118, 291–301. (doi:10.
1007/s00421-017-3769-8)
34. Hof AL, van den Berg Jw. 1977 Linearity between
the weighted sum of the EMGs of the human
triceps surae and the total torque. J. Biomech. 10,
529–539. (doi:10.1016/0021-9290(77)90033-1)
35. Luden N, Minchev K, Hayes E, Louis E, Trappe T,
Trappe S. 2008 Human vastus lateralis and soleus
muscles display divergent cellular contractile
properties. Am. J. Physiol. Regul. Integr. Comp.
Physiol. 295, R1593–R1598. (doi:10.1152/ajpregu.
90564.2008)
36. Ranatunga KW. 1984 The force-velocity relation of
rat fast- and slow-twitch muscles examined at
different temperatures. J. Physiol. 351, 517–529.
(doi:10.1113/jphysiol.1984.sp015260)
37. Umberger BR, Gerritsen KG, Martin PE. 2003 A
model of human muscle energy expenditure.
Comput. Methods Biomech. Biomed. Eng. 6, 99–111.
(doi:10.1080/1025584031000091678)
38. Blagrove RC, Howatson G, Hayes PR. 2017 Test–
retest reliability of physiological parameters in elite
junior distance runners following allometric scaling.
Eur. J. Sport Sci. 17, 1231–1240. (doi:10.1080/
17461391.2017.1364301)
39. Hoogkamer W, Kipp S, Spiering BA, Kram R. 2016
Altered running economy directly translates to
altered distance-running performance. Med. Sci.
royalsocietypublishing.org/journal/rspb
Proc. R. Soc. B 288: 20202784
9
Sports Exerc. 48, 2175–2180. (doi:10.1249/MSS.
0000000000001012)
40. Farris DJ, Sawicki GS. 2012 Human medial
gastrocnemius force–velocity behavior shifts with
locomotion speed and gait. Proc. Natl. Acad. Sci.
USA 109, 977–982. (doi:10.1073/pnas.1107972109)
41. Roberts TJ, Kram R, Weyand PG, Taylor CR. 1998
Energetics of bipedal running. I. Metabolic cost of
generating force. J. Exp. Biol. 201, 2745–2751.
42. Biewener AA. 1998 Muscle function in vivo: a
comparison of muscles used for elastic energy
savings versus muscles used to generate mechanical
power. Integr. Comp. Biol. 38, 703–717. (doi:10.
1093/icb/38.4.703)
43. Ekizos A, Santuz A, Arampatzis A. 2018 Short- and
long-term effects of altered point of ground
reaction force application on human running
energetics. J. Exp. Biol. 221, jeb.176719. (doi:10.
1242/jeb.176719)
44. Biewener AA, Farley CT, Roberts TJ, Temaner M.
2004 Muscle mechanical advantage of human
walking and running: implications for energy cost.
J. Appl. Physiol. 97, 2266–2274. (doi:10.1152/
japplphysiol.00003.2004)
45. Werkhausen A, Cronin NJ, Albracht K, Paulsen G,
Larsen AV, Bojsen-Møller J, Seynnes OR. 2019
Training-induced increase in Achilles tendon
stiffness affects tendon strain pattern during
running. PeerJ 7, e6764. (doi:10.7717/peerj.6764)
46. Stefanyshyn DJ, Nigg BM. 1998 Contribution of the
lower extremity joints to mechanical energy in
running vertical jumps and running long jumps.
J. Sports Sci. 16, 177–186. (doi:10.1080/
026404198366885)
47. Buczek FL, Cavanagh PR. 1990 Stance phase knee
and ankle kinematics and kinetics during level and
downhill running. Med. Sci. Sports Exerc. 22,
669–677.
48. Beck ON, Punith LK, Nuckols RW, Sawicki GS. 2019
Exoskeletons improve locomotion economy by
reducing active muscle volume. Exerc. Sport Sci. Rev.
47, 237–245. (doi:10.1249/JES.0000000000000204)
49. Nuckols RW, Dick TJM, Beck ON, Sawicki GS. 2020
Ultrasound imaging links soleus muscle
neuromechanics and energetics during human
walking with elastic ankle exoskeletons. Sci. Rep.
10, 1–15. (doi:10.1038/s41598-020-60360-4)
50. Obst SJ, Heales LJ, Schrader BL, Davis SA, Dodd KA,
Holzberger CJ, Beavis LB, Barrett RS. 2018 Are the
mechanical or material properties of the Achilles
and patellar tendons altered in tendinopathy? A
systematic review with meta-analysis. Sports Med.
48, 2179–2198. (doi:10.1007/s40279-018-0956-7)
51. Lichtwark GA, Wilson AM. 2007 Is Achilles tendon
compliance optimised for maximum muscle
efficiency during locomotion? J. Biomech. 40,
1768–1775. (doi:10.1016/j.jbiomech.2006.07.025)
royalsocietypublishing.org/journal/rspb
Proc. R. Soc. B 288: 20202784
10
| Enthalpy efficiency of the soleus muscle contributes to improvements in running economy. | 01-27-2021 | Bohm, Sebastian,Mersmann, Falk,Santuz, Alessandro,Arampatzis, Adamantios | eng |
PMC6021049 | RESEARCH ARTICLE
On the apparent decrease in Olympic sprinter
reaction times
Payam Mirshams Shahshahani1*, David B. Lipps2, Andrzej T. Galecki3,4, James
A. Ashton-Miller1,2,3
1 Department of Mechanical Engineering, University of Michigan, Ann Arbor, Michigan, United States of
America, 2 School of Kinesiology, University of Michigan, Ann Arbor, Michigan, United States of America,
3 Institute of Gerontology, University of Michigan, Ann Arbor, Michigan, United States of America,
4 Department of Biostatistics, University of Michigan, Ann Arbor, Michigan, United States of America
* [email protected]
Abstract
Reaction times of Olympic sprinters provide insights into the most rapid of human response
times. To determine whether minimum reaction times have changed as athlete training has
become ever more specialized, we analyzed the results from the Olympic Games between
2004 and 2016. The results for the 100 m and 110 m hurdle events show that minimum reac-
tion times have systematically decreased between 2004 and 2016 for both sexes, with
women showing a marked decrease since 2008 that eliminated the sex difference in 2012.
Because overall race times have not systematically decreased between 2004 and 2016, the
most likely explanation for the apparent decrease in reaction times is a reduction in the pro-
prietary force thresholds used to calculate the reaction times based on force sensors in
starting blocks—and not the result of more specialized or effective training.
Introduction
The ability to rapidly respond to an external auditory stimulus is important when encounter-
ing emergent situations in daily activities such as driving or when operating machinery. Reac-
tion times of sprinters at the Olympic Games offer insights into the fastest human reaction
times because there is little question as to their states of arousal, motivation, learning or train-
ing [1]. (To avoid confusion, since this paper employs International Association of Athletics
Federation (IAAF) data on ‘reaction time’, that term is used throughout this paper rather than
the more usual scientific term ‘response time’.) It is unknown if the reaction times of sprinters
at the Olympic Games remain stable from year to year, or whether changes to the focused
training that athletes undergo in preparation for each Olympics can improve their reaction
times. Understanding the effect of this focused training on the fastest human reaction times
could provide insights into the trainability of athletes and other individuals for time-critical
situations.
Investigating potential sex differences in the fastest human reaction times in elite Olympic
sprinters has important implications on the design of human-machine interfaces for handling
time-critical behaviors. Such interfaces, as in the braking system of an automobile, often do
PLOS ONE | https://doi.org/10.1371/journal.pone.0198633
June 27, 2018
1 / 7
a1111111111
a1111111111
a1111111111
a1111111111
a1111111111
OPEN ACCESS
Citation: Mirshams Shahshahani P, Lipps DB,
Galecki AT, Ashton-Miller JA (2018) On the
apparent decrease in Olympic sprinter reaction
times. PLoS ONE 13(6): e0198633. https://doi.org/
10.1371/journal.pone.0198633
Editor: Maria Francesca Piacentini, University of
Rome, ITALY
Received: February 13, 2018
Accepted: May 22, 2018
Published: June 27, 2018
Copyright: © 2018 Mirshams Shahshahani et al.
This is an open access article distributed under the
terms of the Creative Commons Attribution
License, which permits unrestricted use,
distribution, and reproduction in any medium,
provided the original author and source are
credited.
Data Availability Statement: All the raw data are
already available at the International Association of
Athletics Federations webpage at https://www.iaaf.
org/results. We also uploaded the dataset that we
used for this manuscript on the Deep Blue
webpage at https://deepblue.lib.umich.edu/data/
concern/generic_works/cr56n184r.
Funding: This work was supported by the Claude
Pepper Center Grant AG024824 from the National
Institute of Aging, (JAAM and ATG), https://www.
nia.nih.gov/. The funders had no role in study
not consider potential sex differences in reaction times despite women having faster auditory
latencies [2,3], shorter neural pathways [4] but less muscle strength [5]. A significant sex differ-
ence was found in the reported reaction times of sprinters at the 2008 Beijing Olympics
Games, but this may have been an artifact of the algorithm used to calculate the reaction times
[6]. One goal of this paper was to determine whether sex differences in reported reaction times
have occurred at other Olympic Games. A second goal was to determine whether reaction
times have decreased over Olympic years.
To measure reaction times on the Olympic Track Swiss Timing, a subsidiary of Omega SA,
uses their ASC3 (Automatic Start Control) false start detection system which includes instru-
mented starting blocks to measure the time course of the force applied by the sprinter to the
blocks with a precision of 1 ms following the starting gun. If this force, which their datasheet
shows can reach ~2 kN, exceeds a given (unpublished) force threshold before 100 ms has
elapsed from the onset of the start gun signal, a false start is registered (Rule 162, IAAF Com-
petition Rules). A reaction time is reported in the published Olympic results if the force crosses
the designated threshold after 100 ms, but no reaction time is reported in the event of a false
start (< 100 ms) or other reasons for disqualification.
Any longitudinal comparisons of reaction time of sprinters competing at the Olympic
Games between 2004 and 2016 is confounded by an IAAF rule change in 2010 (Rule 162.7,
IAAF Competition Rules) that disqualified any athlete who false started, rather than permit-
ting a second chance as the prior rule allowed. This change apparently led runners to adopt
slightly more conservative strategies in their reaction times in order to avoid disqualification
[7]. However, since there was no rule change between 2004 and 2008 or between 2012 and
2016, the results from those years should permit a direct comparison of athlete reaction times
independent of the rule change.
Methods
Official reaction times from every heat in the 100, 110, 200, 400 and 440 m track events in
2004, 2008, 2012 and 2016 were downloaded from the official IAAF web site. All names were
stripped from the record to blind the analyses and University of Michigan Institutional Review
Board Approval was received (Exempt—Not Regulated Research, HUM00135664, dated 9/8/
2017). Runners who were disqualified in a heat were excluded, as were those who did not start.
The reaction time (RT) data were positively skewed, so a power transformation (RT-1.5) was
used to obtain a normal distribution [6].
Since we are interested in the minimum human auditory reaction times, we focused the
analysis on the races with the shortest reaction times: in an initial analysis these proved to be
the 100 m and 100 and 110 m hurdles races (S2 Fig and S1 Table). Since one reaction time was
an outlier, exceeding 300 ms, we excluded it for being non-competitive; whether or not it was
included in the analyses would prove not to affect the results.
All data analyses were performed in R (a language and environment for statistical comput-
ing) version 3.4.2. Linear mixed-effect models (LMM) with a random intercept for each athlete
were fit to the data using the ‘nlme’ version 3.1–132 package. We allowed for different residual
variations for each Olympic year and sex. Likelihood Ratio and t-tests were used to examine
the roles of sex and year on an individual’s reaction time for the years 2004–2016. We only
considered the minimum reaction time for each athlete for a given Olympic year; Those results
were then back transformed to find the mean value as well as the -3SD minimum reaction
time values (S1 Text) by sex and year [6].
Official reaction time results for the athletes who competed in the 2004, 2008, 2012 and
2016 Olympic sprinting competitions were included in our analyses (Fig 1A). To account for
On the apparent decrease in Olympic sprinter reaction times
PLOS ONE | https://doi.org/10.1371/journal.pone.0198633
June 27, 2018
2 / 7
design, data collection and analysis, decision to
publish, or preparation of the manuscript.
Competing interests: The authors have declared
that no competing interests exist.
Fig 1. Sprinter minimum reaction times by sex and year in the 2004–2016 Olympics. (A) Scatter plot of reaction
times by sex and year. The solid black circles and bars represent the mean and ±3SD reaction times after back-
transformation. The hatched area designates reaction times deemed by IAAF rule to be a false start. The number of
On the apparent decrease in Olympic sprinter reaction times
PLOS ONE | https://doi.org/10.1371/journal.pone.0198633
June 27, 2018
3 / 7
repeated measures a mixed-effect model with Olympic year and sex (4- and 2-level factors,
respectively) was fit to the set of transformed minimum reaction times. A Likelihood Ratio
(LR) test revealed a significant interaction between Olympic year and sex [LR = 22.82,
p < 0.0001]. After considering the plots of the transformed minimum reaction times (Fig 1B),
the mixed effect model was simplified based on the results of likelihood ratio tests for nested
models. In particular the model was simplified by combining the 2004 and 2008 years into one
category both in main effect of year as well as in year by sex interaction terms [LR = 1.20,
p = 0.54].
Results
The coefficients for the resulting fixed effects can be seen in Table 1 and, when considered
with the results in Fig 1B, they suggest that reaction times decreased significantly in the more
recent Olympics. When only the 2012 and 2016 data were considered as the fixed effects in the
model, there remained significant year and sex differences, but the interaction was no longer
significant [LR = 2.54, p = 0.111]. This suggests that the large sex differences in 2004 and 2008,
which disappeared in 2012, drove the significant interaction term.
Discussion
Our starting point for this study was the finding by Lipps et al. [6] of a significant sex differ-
ence in the reaction times of sprinters at the 2008 Beijing Olympics. Natural questions were
whether this difference would persist at the 2012 London Olympics and whether ever more
focused training leads to slight decreases in reaction times for both men and women. The pres-
ent results showing the abrupt nature of the decrease for only women at the London Olympics
suggests something other than training may have been responsible. One explanation for the
apparent decrease in minimum reaction times from 2012 to 2016 is that both sexes became
more comfortable competing under the threat of disqualification imposed by the 2010 IAAF
rule change, and thereby became less conservative. If so, the results do not support this
false starts reported for the 100 m sprints, 100 m hurdles, and 110 m hurdles were: 1 false start in 2004 and 2008 each, 3
in 2012, and 2 in 2016. (B) Mean (±2SE) transformed minimum reaction times (s-1.5) by sex and year. The 100 ms false
start threshold when transformed becomes 31.6 s-1.5. The most parsimonious Linear Mixed-effect Model found to fit
the data used a random intercept for each athlete. The fixed effects consisted of Olympic year as a factor with 3 levels
(2004 and 2008 together, 2012, and 2016), Sex as a factor with 2 levels, with the interaction of Olympic year by sex. As a
check, the predicted mean values from the LMM were found (the black dots and dashed lines) and agreed well with the
original data (red and blue lines). Minimum reaction times decreased significantly by year (See Results section). Note
that the 2010 IAAF rule change decreed that any runner who false started would be disqualified from the race, and that
applied to the 2012 and 2016 Olympics.
https://doi.org/10.1371/journal.pone.0198633.g001
Table 1. Linear mixed-effect model for transformed minimum reaction times (s-1.5).
Fixed effect
Parameter estimate (s-1.5)
SE
p
Olympic year (2004, 2008)
13.28
0.22
<0.001
Olympic year (2012)
15.37
0.28
<0.001
Olympic year (2016)
17.68
0.3
<0.001
Olympic year (2004, 2008):Men
3.05
0.31
<0.001
Olympic year (2012):Men
0.51
0.35
0.16
Olympic year (2016):Men
1.3
0.39
0.001
The analysis shows that women’s reaction times decreased significantly each year from 2008 to 2012, and 2012 to
2016. However, men’s reaction times only decreased significantly from 2012 to 2016 (Fig 1B).
https://doi.org/10.1371/journal.pone.0198633.t001
On the apparent decrease in Olympic sprinter reaction times
PLOS ONE | https://doi.org/10.1371/journal.pone.0198633
June 27, 2018
4 / 7
explanation because there is no systematic decrease in the race finish times between 2012 and
2016 (S1 Fig). A simpler explanation is that the decreasing reaction times between 2012 and
2016 might be due to a decrease in the force thresholds employed by Swiss Timing to calculate
the reaction times of the men and women [8]. It was suggested by Lipps et al. that a 22%
decrease in the force threshold could have eliminated the sex difference in the reported 2008
reaction time data [6]. The present data for the 2012 Olympics suggest that the force threshold
was indeed reduced for the women in 2012, but not the men; there was no significant sex dif-
ference in their reported reaction times, and the reaction times of the men did not change
between 2008 and 2012 (Table 1, Fig 1B). There is no physiologic reason why women should
be slower than men in their central auditory processing time [2]; put simply, their shorter
limbs mean the signal reaches the leg muscles more quickly than in men, but their smaller leg
muscles mean than it takes longer to develop a significant plantarflexion force against the start-
ing block [6]. This means that it may be fruitful for companies to re-examine how they detect a
false start with an automatic starting system. Our results suggest that the back-transformed
Mean - 3SD value should be set to 100 ms (S1 Text, Fig 1A).
The method for calculating reaction time can vary with the company awarded the timing
contract [9], so such a reduction is within the purview of Swiss Timing, who regard the force
threshold used as proprietary information [6]. The choice of this force threshold is an impor-
tant compromise for the quality of the athletic competitions. If the force threshold is set too
low, the slightest twitch could result in a false start being recorded by the IAAF-certified Start
Information System (SIS). This would not be practical because too many sprinters would be
disqualified and that would spoil the competition. If the force threshold is set too high, sprint-
ers would exhibit unreasonably long reaction times. Of course, in the most recent IAAF rules,
it is the starter that makes the final disqualification decision based on data from the SIS system
as well as whether the athlete initiated his/her starting action before the starter pulled the trig-
ger. Any motion that is not part of the continuous starting movement would simply result in a
caution the first time (Rule 162.7-Note (i), IAAF Competition Rules). The IAAF has been
examining SIS methods for detecting false starts. (see for example [10])
A limitation of this study is the absence of data for reaction times less than 100 ms because
the reaction times are not reported for false-starts. This, and not having access to the starting
block force-time curves, prevents an accurate calculation of the minimum human reaction
time for lower force thresholds than were used; in those cases reaction times could become less
than 100 ms.
We conclude that the apparent decrease in sprinter reaction times between 2004 and 2016
is caused by decreases in the force thresholds used to calculate the reaction time. The decrease
in both men’s and women’s reaction times after 2012 appears to reflect fine tuning of those
force thresholds by Swiss Timing rather than a decrease in the acoustic neuromuscular reac-
tion time of the athletes due to specialized training.
In terms of applicability of the results to other situations, a rapid acoustic reaction time can
be important, for example, when a driver needs to brake an automobile in an emergency after
hearing a warning horn. Our results suggest that the designer of mechanical or electronic
equipment to which humans are to be mechanically coupled should employ the lowest practi-
cal force threshold that does not disadvantage women.
Conclusions
We conclude that the apparent decrease in reaction times is due to a reduction in the proprie-
tary force thresholds used to calculate the reaction times, based on measurements from force
sensors in the starting blocks, not the result of more specialized or effective training.
On the apparent decrease in Olympic sprinter reaction times
PLOS ONE | https://doi.org/10.1371/journal.pone.0198633
June 27, 2018
5 / 7
Supporting information
S1 Text. Why the Mean– 3SD value is a good estimate of minimum auditory reaction time
for Olympic false start detection in sprinting.
(DOCX)
S1 Fig. Distribution of overall ‘mark’ times by sex, year, and race type. IAAF terminology
designates the overall race time as the ‘mark’ time. The boxplot lines represent the median,
and the first and third quartiles. The vertical lines extend up to 1.5 times the interquartile dis-
tance from the top and bottom boxplot lines. The graph shows no systematic change in overall
race times with Olympic year for men or women.
(TIF)
S2 Fig. Transformed minimum reaction times for all the sprints in 2016. Distribution of
transformed minimum (min) reaction time in 2016 for 196, 46, 63, 148, 138, and 94 athletes
who competed in 100 m, 100 m hurdles, 110 m hurdles, 200 m, 400 m, and 400 m hurdles
respectively. The error bars show ±2SE. The shorter sprints (100 m, 100 m hurdles, and 110 m
hurdles) have significantly faster reaction times than the longer sprints (p = 0.0001). To view
the linear mixed effect model for this analysis, please refer to S1 Table.
(TIF)
S1 Table. Linear mixed-effect model results for transformed minimum reaction times for
all sprints in 2016. Transformed minimum reaction time results for the 100 m sprints was
chosen as the reference group. We can see that 100 m hurdles, and 110 m hurdles were not sig-
nificantly different than the reference. However, 200 m, 400 m, and 400 m hurdles were signif-
icantly different than the 100 m sprints.
(PDF)
Acknowledgments
All Olympic Game reaction times are publicly available and downloaded from the official
International Association of Athletic Federation webpage. We are grateful for the developers
of the R Core Team, and ‘nlme’, ‘ggplot2’ and ‘stargazer’ packages who have made their prod-
uct publicly available for free.
Author Contributions
Conceptualization: Payam Mirshams Shahshahani, David B. Lipps, Andrzej T. Galecki, James
A. Ashton-Miller.
Data curation: Payam Mirshams Shahshahani.
Formal analysis: Payam Mirshams Shahshahani, Andrzej T. Galecki.
Funding acquisition: Andrzej T. Galecki, James A. Ashton-Miller.
Investigation: Payam Mirshams Shahshahani.
Methodology: Payam Mirshams Shahshahani, Andrzej T. Galecki.
Project administration: Payam Mirshams Shahshahani.
Resources: James A. Ashton-Miller.
Software: Payam Mirshams Shahshahani.
Supervision: Andrzej T. Galecki, James A. Ashton-Miller.
On the apparent decrease in Olympic sprinter reaction times
PLOS ONE | https://doi.org/10.1371/journal.pone.0198633
June 27, 2018
6 / 7
Validation: Payam Mirshams Shahshahani, Andrzej T. Galecki.
Visualization: Payam Mirshams Shahshahani.
Writing – original draft: Payam Mirshams Shahshahani, James A. Ashton-Miller.
Writing – review & editing: Payam Mirshams Shahshahani, David B. Lipps, Andrzej T.
Galecki, James A. Ashton-Miller.
References
1.
Bell DG, Jacobs I. Electro-mechanical response times and rate of force development in males and
females. Med Sci Sports Exerc. 1986 Feb; 18(1):31–36. PMID: 3959861
2.
Don M, Ponton CW, Eggermont JJ, Masuda A. Gender differences in cochlear response time: An expla-
nation for gender amplitude differences in the unmasked auditory brain-stem response. J Acoust Soc
Am. 1993; 94(4):2135–48. PMID: 8227753
3.
Trune DR, Mitchell C, Phillips DS. The relative importance of head size, gender and age on the auditory
brainstem response. Hear Res. 1988; 32(2–3):165–74. PMID: 3360676
4.
Uth N. Anthropometric comparison of world-class sprinters and normal populations. J Sport Sci Med.
2005; 4(4):608–16.
5.
Thelen DG, Schultz AB, Alexander NB, Ashton-Miller JA. Effects of age on rapid ankle torque develop-
ment. J Gerontol. 1996; 51(5):M226–32.
6.
Lipps DB, Galecki AT, Ashton-Miller JA. On the implications of a sex difference in the reaction times of
sprinters at the Beijing Olympics. PLoS One. 2011; 6(10):4–8.
7.
Brosnan KC, Hayes K, Harrison AJ. Effects of false-start disqualification rules on response-times of
elite-standard sprinters. J Sports Sci. 2017; 35(10):929–35. https://doi.org/10.1080/02640414.2016.
1201213 PMID: 27351870
8.
ASC3 -FALSE START DETECTION SYSTEM [Internet]. [cited 2017 Nov 18]. Available from: https://
www.swisstiming.com/fileadmin/Resources/Data/Datasheets/DOCM_AT_ASC3_
FalseStartDetectionSystem_0715_EN.pdf
9.
Pain MTG, Hibbs A. Sprint starts and the minimum auditory reaction time. J Sports Sci. 2007; 25(1):79–
86. https://doi.org/10.1080/02640410600718004 PMID: 17127583
10.
Willwacher S, Feldker MK, Zohren S, Herrmann V, Bru¨ggemann GP. A novel method for the evaluation
and certification of false start apparatus in sprint running. Procedia Eng [Internet]. 2013; 60:124–9.
Available from: http://dx.doi.org/10.1016/j.proeng.2013.07.073
On the apparent decrease in Olympic sprinter reaction times
PLOS ONE | https://doi.org/10.1371/journal.pone.0198633
June 27, 2018
7 / 7
| On the apparent decrease in Olympic sprinter reaction times. | 06-27-2018 | Mirshams Shahshahani, Payam,Lipps, David B,Galecki, Andrzej T,Ashton-Miller, James A | eng |
PMC10664904 | RESEARCH ARTICLE
Partial-body cryostimulation procured
performance and perceptual improvements
in amateur middle-distance runners
Massimo De Nardi1,2☯, Luca FilipasID3,4☯, Carlo Facheris1,3, Stefano RighettiID5,6,
Marco Tengattini5, Emanuela Faelli2,7, Ambra BisioID2,7, Gabriele Gallo2,7, Antonio La
Torre3,5,8, Piero Ruggeri2,7‡, Roberto CodellaID3,4‡*
1 Krioplanet Ltd, Treviglio, Bergamo, Italy, 2 Department of Experimental Medicine, Università degli Studi di
Genova, Genoa, Italy, 3 Department of Biomedical Sciences for Health, Università degli Studi di Milano,
Milan, Italy, 4 Department of Endocrinology, Nutrition and Metabolic Diseases, IRCCS MultiMedica, Milano,
Italy, 5 Italian Athletics Federation, Rome, Italy, 6 Interventional Cardiology Department, San Gerardo
Hospital, Monza, Italy, 7 Centro Polifunzionale di Scienze Motorie, Università degli Studi di Genova, Genoa,
Italy, 8 IRCCS Istituto Ortopedico Galeazzi, Milan, Italy
☯ These authors contributed equally to this work.
‡ PR and RC also contributed equally to this work.
* [email protected]
Abstract
The purpose of this study was to investigate the effects of partial-body cryostimulation on
middle-distance runners before two 3000-m tests at the speed of the first and second venti-
latory threshold, and before a time to exhaustion test at 110% of the maximal aerobic speed.
Twelve amateur runners (age: 46 ± 9 years; VO2max: 51.7 ± 4.9 mlkg-1min-1) completed six
running testing sessions in a randomized counterbalanced cross-over fashion: three of
them were preceded by a partial-body cryostimulation and the other three by a control condi-
tion. The testing sessions consisted of: 1) a 3000-m continuous running test at the speed of
the first ventilatory threshold; 2) a 3000-m continuous running test at the speed of the sec-
ond ventilatory threshold; 3) a time to exhaustion test at 110% of the maximal aerobic
speed. Heart rate, ratings of perceived exertion and visual analogue scale relative to muscle
pain were recorded throughout the tests. Total quality recovery was evaluated 24–48 h after
the end of each test. Distance to exhaustion was higher after partial-body cryostimulation
than control condition (p = 0.018; partial-body cryostimulation: 988 ± 332 m, control: 893 ±
311 m). There were differences in the ratings of perceived exertion during each split of the
3000-m continuous running test at the speed of the second ventilatory threshold (p = 0.001).
Partial-body cryostimulation can be positively considered to enhance middle-distance run-
ning performance and reduce perception of effort in amateur runners.
Introduction
In high-performance sport, coaches and sports scientists strive to identify novel interventions
to enhance performance maximizing the physical, psychological and behavioral state before a
PLOS ONE
PLOS ONE | https://doi.org/10.1371/journal.pone.0288700
November 22, 2023
1 / 12
a1111111111
a1111111111
a1111111111
a1111111111
a1111111111
OPEN ACCESS
Citation: De Nardi M, Filipas L, Facheris C, Righetti
S, Tengattini M, Faelli E, et al. (2023) Partial-body
cryostimulation procured performance and
perceptual improvements in amateur middle-
distance runners. PLoS ONE 18(11): e0288700.
https://doi.org/10.1371/journal.pone.0288700
Editor: Nejka Potocnik, University of Ljubljana,
Medical faculty, SLOVENIA
Received: December 10, 2021
Accepted: July 3, 2023
Published: November 22, 2023
Copyright: © 2023 De Nardi et al. This is an open
access article distributed under the terms of the
Creative Commons Attribution License, which
permits unrestricted use, distribution, and
reproduction in any medium, provided the original
author and source are credited.
Data Availability Statement: All relevant data are
within the paper and its Supporting Information
files.
Funding: The authors received no specific funding
for this work.
Competing interests: The authors have declared
that no competing interests exist.
competition [1, 2]. Prior to performance, several routines have been pursued with the aim of
enhancing oxygen uptake, cardiac output, blood flow to skeletal muscle, neuromuscular activa-
tion and mental readiness [3]. Running economy is certainly another typical target.
Pre-training typically involves a combination of passive and active elements which can
enhance exercise performance [4]. Active warm-ups are used to increase body temperature
leading to a higher muscle metabolism, enhanced oxygen uptake and subsequent cardiac out-
put [5], while passive warm up has the same goal but is mainly focused on various psychologi-
cal techniques or cooling/heating interventions [6]. Preparation through passive modalities is
becoming increasingly common for athlete management and performance enhancement [7].
Although every team, staff, strength and conditioning coach, have individualized methods and
preferences for their athletes/players, cutting-edge methodologies and techniques are evolving
and being available at large scale [8].
Cryotherapy is a relatively new clinical intervention used in different medical treatments to
alleviate pain derived from inflammatory conditions [9]. Given the extremely cold air temper-
atures of –110˚C or below, reached in short (1–4 min) exposures, cryostimulation can elicit
strong physiological responses in users [10], ranging from muscular soreness relief up to a
modest immunomodulatory effect [11].
The utilization of cryostimulation techniques before physical efforts is also known as pre-
cooling [12]. Usually, precooling aims to a rapid removal of heat of the body prior to exercising
in warm environments and to prevent the side effects of heat-stress-induced fatigue [13].
Reducing body temperature before commencing an effort has been found to be successful in
improving endurance performance both in hot and thermoneutral environments [14, 15].
Higher precooling effects were also observed in prolonged efforts [16], ambient temperature
and for aerobic capacity [17]. Improvements in endurance performance following precooling
interventions were achieved both with external (cold air exposure, water immersion, exposure
to ice, iced garments, iced towels, etc.), internal modalities (beverage ingestion, ice ingestion)
or combining two or more practical precooling methods [12, 17]. Cooling strategies are logisti-
cally challenging, therefore it is important to analyze the effectiveness of practical precooling
in competition or field setting [12].
In particular, the new transportable models of cryo-saunas have increased the diffusion of
partial-body cryostimulation (PBC) treatments before races and training [18], which have
been demonstrated to lower core and skin temperature for several hours [19]. Exposures to
very-low temperatures (approximately at -120˚C) could have positive impacts on core body
temperature, cardiovascular and autonomic functions [20], and perception of effort [21]. Nev-
ertheless, the disparities in the literature among various precooling methods, exercise efforts
and thermometry protocols remain substantial. This breath of knowledge could be scoped by
athletes, coaches and their extended entourage to tailor the maximizing-performance routine.
In this study, the effects of PBC on running performance were evaluated with two 3000-m
tests at the speed of the first and second ventilatory threshold, and before a time to exhaustion
test at 110% of the maximal aerobic speed (MAS). We aimed to test if PBC could be applied as
a prior to competition performance enhancement methodology in middle-distance events.
Our hypothesis was that running performance would have been improved by cryo-exposures,
along with lower value of perceptual fatigue.
Materials & methods
Experimental approach to the problem
A randomized counterbalanced cross-over design was used for the experimental component
of the present study. The order of the experimental treatments (PBC; control) and testing
PLOS ONE
Precooling in endurance runners
PLOS ONE | https://doi.org/10.1371/journal.pone.0288700
November 22, 2023
2 / 12
sessions were randomly allocated based on balanced permutations generated by a computer
program (www.randomization.com). A flow-chart of the protocol is offered in Fig 1.
Subjects
To determine an a priori sample-size (software package, G * Power 3.1.9.2), the following
input parameters were selected as per an F test for ANOVA-repeated measures-within factors
analysis: a statistical power (1-β) of 0.8, a significance α level of 0.05, an effect size f of 0.38
(which corresponds to a η2
p = 0.13), 1 group, 6 measurements, 0.5 as correlation among
repeated measurements. As output parameters, an actual power of 0.83, a critical F of 2.45
Fig 1. Flow-chart of the study.
https://doi.org/10.1371/journal.pone.0288700.g001
PLOS ONE
Precooling in endurance runners
PLOS ONE | https://doi.org/10.1371/journal.pone.0288700
November 22, 2023
3 / 12
were obtained. Therefore, eight subjects would have been sufficient to assess the sought effects.
However, to face a possible drop-out of one third of the subjects, thirteen male amateur run-
ners were recruited for this study. Participants were evaluated by a medical doctor who ascer-
tained their competitive suitability and excluded any contraindication to systemic
cryostimulation. Eligibility criteria were as follows: being free from any known medical dis-
eases, medication, injuries, color vision deficiencies. The study design and procedures were
approved by the local Ethics Committee and followed the ethical principles for medical
research involving human subjects set by the World Medical Association Declaration of Hel-
sinki. After ethical approval, written informed consent and medical declaration were obtained
from the participants in line with the procedures set by the local Institution’s Research Ethics
Committee. Subjects were informed of the procedures and potential risks involved, although
these were minimal as already described [22]. Participants were also informed that they could
withdraw from the study at any time, for any reason.
Procedures
Participants performed eight testing sessions on eight different occasions, in a period no lon-
ger than three weeks between the first and last visit (with at least 48 h between two visits). Vis-
its were carried in a private laboratory. All the experimental procedures were performed in an
isolated and air-conditioned room, at the constant temperature of 21 ± 1˚C and at a relative
humidity of 40–50%. Before each visit, participants were instructed to sleep for at least 8 h,
refrain from the consumption of alcohol and caffeine, and avoid any vigorous exercise for the
24-h preceding the testing sessions. Each participant carried out the visits individually and at
the same time of day (within 1 h period, between 7:30 and 10:30).
During visit 1, participants’ body weight and height were measured. Afterwards, they familiar-
ized with the procedures employed for the testing sessions, i.e. running on the treadmill used for
the study (Excite Treadmill, Technogym, Cesena, Italy) for at least 20 min wearing a portable gas
exchange system in breath-by-breath mode (Cosmed K5, Cosmed, Rome, Italy). Treadmill speed
was validated for each stage using an odometer (Stanley, Milano, Italy) to confirm that it was the
one declared by the manufacturer. During visit 2, participants completed an incremental exercise
test to determine their maximal oxygen uptake (VO2max). The test started with a standardized
warm-up consisting in 10 min of running at a constant speed of 10 kmh-1, followed by 5 min of
mobility drills. At the end of the mobility exercises, participants completed 10 min of passive
recovery and, after wearing the mask, 3 min in a standing position on the treadmill for the acquisi-
tion of basal values. After the warm-up, the incremental test started at a speed equal to 10 kmh-1,
with an increase of 0.1 kmh-1 every 12 s until exhaustion. Alveolar gas exchanges were measured
breath by breath in the mouth, eliminating any values outside the pre-established range, using a
specific software (OMNIA, Cosmed, Rome, Italy). Maximum oxygen uptake was calculated as the
30 s mean oxygen uptake once the plateau was reached. Runner’s VO2max was reached when at
least three of the following criteria were fulfilled: i) a steady state of VO2 (change in VO2 at
VO2max 150 mLmin-1), ii) final respiratory-exchange ratio (RER) exceeded 1.1, iii) visible
exhaustion, iv) a HR at the end of exercise within the 10 bpm of the predicted maximum, and v) a
lactate concentration at the end of exercise higher than 8 mmolL-1 [23]. Blood samples were
obtained immediately after exhaustion, in a single measurement, through the ear lobe, and were
analyzed for whole blood lactate using a portable lactate analyzer (Lactate Pro, Arcray Inc, Kyoto,
Japan), reported to have good reliability and accuracy [24]. First and second ventilatory thresholds
were detected through the ventilatory equivalents method [25] and the speeds associated were cal-
culated. In addition, researchers calculated the maximal aerobic speed (MAS) as the lowest run-
ning speed at which VO2max occurred.
PLOS ONE
Precooling in endurance runners
PLOS ONE | https://doi.org/10.1371/journal.pone.0288700
November 22, 2023
4 / 12
After the first two preliminary visits, participants carried out randomly the three testing ses-
sions, preceded either by a PBC or a control condition (six sessions in total, all randomized,
Fig 1). The testing sessions consisted of a 3000-m continuous running test at the speed of the
first ventilatory threshold, a 3000-m continuous running test at the speed of the second venti-
latory threshold and a time to exhaustion test at 110% of the MAS. The order of the tests and
the interventions were randomly counterbalanced. The tests were preceded by the same
warm-up routine described for the incremental ramp test. The warm-up started within two
minutes by the end of the PBC or control condition. The tests at the first and second ventila-
tory thresholds were performed over a distance of 3000 m to give enough time to the physio-
logical parameters to reach a steady-state.
Experimental treatments
Before each testing session, participants underwent either a PBC session or a control session.
During the PBC duty, participants completed the cryo-session (150 s) in a cryo-cabin (Space
Cabin, Criomed Ltd, Kherson, Ukraine). Temperature was set between -130 and -170˚C as
recommended [26]. Participants were instructed to turn around continuously (standing rota-
tions) in the cabin for the 150-second session. The control condition required to perform simi-
lar movement (standing rotations) for the same duration in a thermo neutral environment
(21 ± 1˚C). Immediately after the cryo-exposure or control task, the running tests were per-
formed. Due to the nature of the cryostimulation, participants were not blinded to their treat-
ments. However, the research team was blinded to the treatment because specific personnel
oversaw the treatments in a separated room.
Physiological and psychological measures
Heart rate (HR) was recorded, and averaged for the duration of the tests, using a HR monitor
fitted with a chest strap. VO2 was recorded using portable gas exchange system in breath-by-
breath mode (Cosmed K5, Cosmed, Rome, Italy) and data were averaged using the specific
software (OMNIA, Cosmed, Rome, Italy). During the 3000-m continuous running tests and
the 110% of the MAS test, VO2 were reported as the average of the last 30 s of the tests. Rating
of perceived exertion (RPE) was registered during the final 10 s of 1000- and 2000-m splits, at
the end of the 3000-m continuous running tests, and at the end of the time to exhaustion test.
RPE was measured with the 11-point CR10 scale developed by Borg [27]. Participants were
familiar with the scale as it had been employed during their daily training sessions for at least
six months ahead the tests. Participants recorded their subjective sensation of muscle pain by
using a visual analogue scale (VAS), before and after each test. Participants marked their
response on a 100-mm line anchored by 0 (no muscle pain at all) and 100 (maximum muscle
pain) [28]. The total quality recovery (TQR) scale [29] was used to monitor recovery. After 24
and 48 hours from the end of each test, participants answered the question “How do you feel
about your recovery?” using the TQR scale, in which answers are rated from 6 to 20.
Statistical analyses
All data are presented as mean ± standard deviation. Assumptions of statistical tests such as
normal distribution (Shapiro-Wilk test with visual inspection) and sphericity of data
(Mauchly’s test) were checked as appropriate. Greenhouse-Geisser correction to the degrees of
freedom was applied when violation to sphericity was present. Two-way repeated measured
ANOVA was used to determine the treatment factor (2 levels, PBC and Control), time factor
(3 levels, at 1000 m, 2000 m, 3000 m splits), and interaction on RPE during the 3000-m tests.
Significant main effects and interactions were interpreted through pairwise comparisons with
PLOS ONE
Precooling in endurance runners
PLOS ONE | https://doi.org/10.1371/journal.pone.0288700
November 22, 2023
5 / 12
Bonferroni correction. Paired sample t tests were used to compare the other variables between
the PBC and control conditions. Significance was set at 0.05 (two-tailed) for all analyses. Effect
sizes for repeated measure ANOVA are reported as partial eta squared (η2
p), using the small
(< 0.13), medium (0.13–0.25) and large (> 0.25) interpretation for effect size [30], while effect
sizes for pairwise comparison were calculated using Cohen’s d and considered to be either triv-
ial (effect size: < 0.20), small (0.21–0.60), moderate (0.61–1.20), large (1.21–2.00), or very large
(> 2.00) [31]. Data analysis was conducted using the Statistical Package for the Social Sciences,
version 25 (SPSS Inc., Chicago, IL, USA).
Results
Participants’ baseline characteristics and data derived from the incremental ramp test are
reported in Table 1. One participant was classified as outlier based on his maximum oxygen
uptake (VO2max) (more than two standard deviation from the mean of the sample) and
excluded from the analysis. Therefore, twelve subjects were included in the study procedures
(age: 46 ± 9 years; height: 1.75 ± 0.05 m; mass: 72 ± 8 kg).
Performance outcomes
Table 2 shows the performance outcomes derived from the 3000-m continuous running test at
the speed of the first and second ventilatory threshold and the time to exhaustion test at 110%
of the MAS. Running distance (p = 0.018, d = 0.30) and time (p = 0.020, d = 0.31) during the
time to exhaustion test were higher after PBC than control condition. No differences were
found among the other parameters.
Psychological outcomes
Table 3 shows the pairwise comparison of the psychological outcomes derived during the
3000-m continuous running test at the speed of the first and second ventilatory threshold and
the time to exhaustion test at 110% of the MAS. There was a significant condition x time inter-
action in the RPE during the 3000-m continuous running test at the speed of the second venti-
latory threshold (F (2,18) = 15.716, p = 0.001, η2
p = 0.43). Pairwise comparison revealed
significant differences for RPE in 1000-m (p = 0.010, d = 0.54), 2000-m (p = 0.030, d = 0.65)
Table 1. Participants’ characteristics at baseline (mean ± SD).
VO2max (mLkg-1min-1)
51.7 ± 4.9
HRmax, incremental test (bpm)
181 ± 16
vVT1 (kmh-1)
11.8 ± 0.9
vVT2 (kmh-1)
13.4 ± 0.9
MAS (kmh-1)
14.7 ± 1.1
VO2VT1 (mLkg-1min-1)
45.2 ± 5.3
VO2VT2 (mLkg-1min-1)
49.8 ± 5.3
HRVT1 (bpm)
156 ± 18
HRVT2 (bpm)
167 ± 17
VO2max: maximum oxygen uptake; HRmax: maximum heart rate; vVT1: velocity at first ventilatory threshold; vVT2:
velocity at second ventilatory threshold; MAS: maximum aerobic speed; VO2VT1: oxygen uptake at first ventilatory
threshold; VO2VT2: oxygen uptake at second ventilatory threshold; HRVT1: heart rate at first ventilatory threshold;
HRVT2: heart rate at second ventilatory threshold.
https://doi.org/10.1371/journal.pone.0288700.t001
PLOS ONE
Precooling in endurance runners
PLOS ONE | https://doi.org/10.1371/journal.pone.0288700
November 22, 2023
6 / 12
Table 2. Performance outcomes from the three running tests in partial-body cryotherapy (PBC) and control conditions (mean ± SD).
3000-m test at VT1
PBC
Control
p
Cohen’s d
Time (s)
910 ± 77
919 ± 70
0.237
0.13
VO2 (mLkg-1min-1)
46.4 ± 5.5
47.1 ± 6.0
0.731
0.12
HR (bpm)
159 ± 17
159 ± 18
0.478
0.04
RER (VCO2VO2
-1)
0.84 ± 0.06
0.85 ± 0.05
0.750
0.13
3000-m test at VT2
Time (s)
693 ± 217
711 ± 152
0.491
0.09
VO2 (mLkg-1min-1)
48.8 ± 5.8
48.1 ± 5.2
0.502
0.13
HR (bpm)
168 ± 15
167 ± 17
0.625
0.05
RER (VCO2VO2
-1)
0.92 ± 0.08
0.92 ± 0.09
0.817
0.06
TTE at 110% of MAS
Distance (m)
988 ± 332
893 ± 311
0.018 *
0.30
Time (s)
222 ± 73
201 ± 67
0.020 *
0.31
VO2 (mLkg-1min-1)
49.8 ± 6.1
46.7 ± 6.2
0.233
0.49
HR (bpm)
170 ± 13
167 ± 13
0.227
0.21
RER (VCO2VO2
-1)
1.13 ± 0.11
1.13 ± 0.12
0.781
0.07
VT1: first ventilatory threshold; VT2: second ventilatory threshold; TTE: time to exhaustion; MAS: maximum aerobic speed; VO2: oxygen uptake; HR: heart rate; RER:
respiratory exchange ratio.
* Significant difference between the conditions (p < 0.05).
https://doi.org/10.1371/journal.pone.0288700.t002
Table 3. Psychological outcomes from the three running tests in partial-body cryotherapy (PBC) and control conditions (mean ± SD).
3000-m test at VT1
PBC
Control
p
Cohen’s d
VAS pre (mm)
11 ± 14
18 ± 16
0.151
0.44
VAS post (mm)
15 ± 18
17 ± 17
0.638
0.10
1000-m RPE
2.6 ± 1.2
2.3 ± 0.9
0.410
0.24
2000-m RPE
3.2 ± 1.5
3.0 ± 1.2
0.305
0.13
3000-m RPE
3.3 ± 1.9
3.3 ± 1.9
1.000
0.00
24-h TQR
17.5 ± 2.3
17.5 ± 2.6
0.915
0.02
48-h TQR
18.6 ± 2.1
18.1 ± 2.3
0.053
0.35
3000-m test at VT2
VAS pre (mm)
13 ± 11
11 ± 12
0.504
0.15
VAS post (mm)
13 ± 11
18 ± 17
0.141
0.41
1000-m RPE
3.2 ± 1.3
3.9 ± 1.5
0.010 *
0.54
2000-m RPE
4.3 ± 1.7
5.5 ± 2.1
0.030 *
0.65
3000-m RPE
5.3 ± 1.8
6.6 ± 2.7
0.008 *
0.56
24-h TQR
17.9 ± 2.1
16.3 ± 3.3
0.091
0.57
48-h TQR
18.2 ± 2.4
17.2 ± 2.4
0.090
0.44
TTE at 110% of MAS
VAS pre (mm)
8 ± 11
10 ± 14
0.748
0.14
VAS post (mm)
15 ± 14
23 ± 28
0.279
0.36
RPE at exhaustion
8.2 ± 2.0
8.3 ± 2.0
0.928
0.02
24-h TQR
17.2 ± 2.0
17.1 ± 2.1
0.903
0.04
48-h TQR
17.8 ± 1.8
17.3 ± 2.7
0.324
0.22
VT1: first ventilatory threshold; VT2: second ventilatory threshold; TTE: time to exhaustion; MAS: maximum aerobic speed; VAS: visual analogue scale; RPE: rating of
perceived exertion; TQR: total quality recovery.
* Significant difference between the conditions (p < 0.05).
https://doi.org/10.1371/journal.pone.0288700.t003
PLOS ONE
Precooling in endurance runners
PLOS ONE | https://doi.org/10.1371/journal.pone.0288700
November 22, 2023
7 / 12
and 3000-m (p = 0.008, d = 0.56) splits. No differences were found among the other
parameters.
Discussion
The main finding of this study was that a PBC session increased running time to exhaustion,
showing an improved performance at 110% of MAS after a single PBC session compared to a
control condition. Further, RPE was significantly reduced after the PBC in each split of the
3000-m continuous running test at the second ventilatory threshold, suggesting that both cen-
tral and peripheral mechanisms could be affected by PBC. Partially unexpected, no other dif-
ferences were found in the two different conditions for other performance or psychological
measures.
This study showed a potential implication in adopting this relatively new technique before
running performance in a moderate-temperature condition. To date, an abundance of evi-
dence demonstrated the favorable use of PBC as a recovery modality after high intensity exer-
cise for athletes. Researchers have identified how PBC can improve recovery post-exercise by
maximizing anti-inflammatory and decreasing pro-inflammatory actions [9]. Conversely, to
the best of our knowledge, the effects of PBC before an exercise are mostly uncertain. Prior
investigations have documented the effectiveness of precooling using PBC in ameliorating
flexibility without losing the trunk position sense proprioception [32]. In addition, numerous
studies showed the positive influence of several precooling strategies in hot environments on
endurance performance, while no studies have evaluated the effects of PBC in thermoneutral
conditions [14]. Our results indicated PBC as a method capable to induce improvements in
middle-distance running performance, probably mediated by a lower RPE for the same exter-
nal workload. This is only a hypothesis as in the present study the improvement in perfor-
mance and the reduction of RPE occurred not in combination but during different tests. This
result is possibly due to the timing of the RPE evaluation, i.e. at the end of the time to exhaus-
tion test, and at each 1000-m split in the 3000-m time trials. Therefore, the ceiling effect of the
RPE in an exercise-to-exhaustion probably determined the lack of significance in the TTE at
110% of the MAS [27]. This result is not an unicum in the body of literature as a similar one
was found in a submaximal exercise in elite synchronized swimmers [33]. Klimek and col-
leagues have also shown an improvement in anaerobic capacity after 10 whole-body cryosti-
mulation sessions, principally explained by metabolic changes (i.e. increased activity of
anaerobic glycolytic enzymes) and a greater tolerance to pain, highlighted by an increase in
blood lactate concentration [34].
Reduction in pain could be easily linked to a decrease in perception of effort, a master regu-
lator of performance in endurance exercises [35]. Of note, perception of effort is determinant
in a TTE test, as it has been shown that maximum values of perception of effort lead to an
early stop to exercise, despite athletes could continue both muscularly and metabolically [36].
Physiologically, the reduction of perception of effort could be explained by the modification
of peripheral afferent signals generated by PBC, that could play a crucial role in changing the
perception of effort during exercise at low- and high-intensity [37]. In fact, together with neu-
ral drive from the motor cortex area, previous studies showed that RPE also is influenced by
afferent feedback from the periphery for the prediction of the endpoint of the exercise bout
[37, 38].
No other changes were observed in the other physiological and psychological parameters.
Our results are in line with previous studies that have shown no change in performance, recov-
ery, soreness perception parameters [39, 40], and with strategies that altered the perception of
effort [41, 42]. Moreover, the reduction of RPE at the same timepoints in the 3000-m tests
PLOS ONE
Precooling in endurance runners
PLOS ONE | https://doi.org/10.1371/journal.pone.0288700
November 22, 2023
8 / 12
implies that the RPE:VO2 and RPE:HR ratios are reduced in both tests in the PBC condition
compared to the control one. This reduction denotes an indirect change in the VO2 and HR
parameters, that would have been higher for the same RPE. This influence of PBC on RPE
indicates a strong impact of cryostimulation on fatigue, also shown by the lack of significance
shown by the VAS relative to muscle pain.
We must certainly highlight how the great variability of the measures is a factor to consider
in the interpretation of the results. It is known that the coefficient of variation of the TTE is
large. Despite constant-work test has a much lower coefficient of variation, with amateur ath-
letes we must also consider the variability generated by pacing behavior. Hinckson and Hop-
kins [43], using the relationship between exercise duration and power output showed that the
reliability of the equivalent mean power calculated from the constant-power test (0.6%) is even
better than from the constant-work test (1.0%). In addition, Laursen and colleagues [44]
reported that a small random variation in performance results in large variation in time to
exhaustion, which causes large value of coefficient of variation. On the other hand, the small
change after an intervention also leads to a large change in the time to exhaustion, therefore
the signal-to-noise ratio of the constant-power test should be comparable to the constant-work
test. In addition, participants were familiar with the experimental procedures, but especially in
the TTE, the nature of the performance was far from the performance habits of the
participants.
An important limitation of the present study is the lack of cardiovascular and physiological
measures that compared the effects of the exposure to the PBC itself. Defining cardiovascular
changes will be crucial for the assessment of the efficiency of PBC in sport.
Since the implementation of cryostimulation prior to competition is relatively recent [45],
there is a lack of comparability among methodologies, in terms of cryostimulation modalities,
exposure parameters, and types of exercise. Besides, future investigations should evaluate
whether the exercise-induced inflammatory response is preserved in warmer conditions, and
whether precooling effects are likewise effective with higher temperatures. Moreover, since we
included amateur master athletes, future studies should aim to verify if the results of the pres-
ent study could be extended to other population of athletes (e.g. young, elite, etc.). In addition,
the 48-h TQR values could indicate an inadequate recovery for participants who underwent
the subsequent visit after 48 h. However, most of them completed each visit with at least 72 h
of recovery in between.
Practical applications
This study results advocate for the use of cryostimulation before anerobic activities. Beyond
the recovery benefits reported in the literature, cryostimulation may boost performance in the
pre-execution phase. However, it is essential to evaluate the best timeframe to maximize the
potential enhancing-performance effects. Cryostimulation may be also reducing the percep-
tion of effort and this effect could be crucial for coaches to modulate the training prescription
of their athletes. Considering this result, athletes could potentially increase their external train-
ing load given the lower internal/perceptual load for the same external load. In fact, reducing
RPE for the same external load could lead to a higher number of repetitions during an interval
training session or a higher external intensity for each repetition.
Conclusions
A single PBC session may represent a favorable set-up routine before running, improving mid-
dle-distance running performance and reducing RPE for the same external effort also in mod-
erate-temperature conditions. Future studies should investigate the optimal integration of
PLOS ONE
Precooling in endurance runners
PLOS ONE | https://doi.org/10.1371/journal.pone.0288700
November 22, 2023
9 / 12
PBC with the traditional elements of active of pre-exercise routines, so to maximize middle-
distance performance.
Supporting information
S1 Data.
(XLSX)
Acknowledgments
The authors acknowledge support from the University of Milan through the APC initiative.
Author Contributions
Conceptualization: Massimo De Nardi, Luca Filipas, Carlo Facheris, Stefano Righetti, Marco
Tengattini, Emanuela Faelli, Ambra Bisio, Antonio La Torre, Piero Ruggeri, Roberto
Codella.
Data curation: Massimo De Nardi, Carlo Facheris, Stefano Righetti, Marco Tengattini.
Formal analysis: Massimo De Nardi, Carlo Facheris, Marco Tengattini.
Investigation: Massimo De Nardi, Carlo Facheris, Stefano Righetti, Marco Tengattini.
Methodology: Massimo De Nardi, Luca Filipas, Ambra Bisio, Roberto Codella.
Project administration: Massimo De Nardi, Luca Filipas, Emanuela Faelli, Gabriele Gallo,
Piero Ruggeri, Roberto Codella.
Supervision: Luca Filipas, Emanuela Faelli, Ambra Bisio, Antonio La Torre, Piero Ruggeri,
Roberto Codella.
Validation: Massimo De Nardi, Luca Filipas, Stefano Righetti, Gabriele Gallo, Antonio La
Torre, Piero Ruggeri, Roberto Codella.
Visualization: Massimo De Nardi, Luca Filipas, Marco Tengattini, Ambra Bisio, Gabriele
Gallo.
Writing – original draft: Massimo De Nardi, Luca Filipas, Carlo Facheris, Roberto Codella.
Writing – review & editing: Luca Filipas, Stefano Righetti, Emanuela Faelli, Ambra Bisio,
Gabriele Gallo, Antonio La Torre, Piero Ruggeri, Roberto Codella.
References
1.
Filipas L, Bonato M, Gallo G, Codella R. Effects of 16 weeks of pyramidal and polarized training intensity
distributions in well-trained endurance runners. Scand J Med Sci Sports. 2022; 32: 498–511. https://doi.
org/10.1111/sms.14101 PMID: 34792817
2.
Sands WA, Kavanaugh AA, Murray SR, McNeal JR, Jemni M. Modern techniques and technologies
applied to training and performance monitoring. International Journal of Sports Physiology and Perfor-
mance. 2017. https://doi.org/10.1123/ijspp.2016-0405 PMID: 27918664
3.
McGowan CJ, Pyne DB, Thompson KG, Rattray B. Warm-Up Strategies for Sport and Exercise: Mech-
anisms and Applications. Sports Medicine. 2015. https://doi.org/10.1007/s40279-015-0376-x PMID:
26400696
4.
McGowan CJ, Pyne DB, Thompson KG, Raglin JS, Osborne M, Rattray B. Elite sprint swimming perfor-
mance is enhanced by completion of additional warm-up activities. J Sports Sci. 2017. https://doi.org/
10.1080/02640414.2016.1223329 PMID: 27631544
5.
Kilduff LP, Finn C V., Baker JS, Cook CJ, West DJ Preconditioning strategies to enhance physical per-
formance on the day of competition. Int J Sports Physiol Perform. 2013. https://doi.org/10.1123/ijspp.8.
6.677 PMID: 23689163
PLOS ONE
Precooling in endurance runners
PLOS ONE | https://doi.org/10.1371/journal.pone.0288700
November 22, 2023
10 / 12
6.
Cook CJ, Crewther BT, Kilduff LP. Are free testosterone and cortisol concentrations associated with
training motivation in elite male athletes? Psychology of Sport and Exercise. 2013. https://doi.org/10.
1016/j.psychsport.2013.08.001
7.
Bishop D. Warm up I: Potential mechanisms and the effects of passive warm up on exercise perfor-
mance. Sports Medicine. 2003. https://doi.org/10.2165/00007256-200333060-00005 PMID: 12744717
8.
Faelli E, Bisio A, Codella R, Ferrando V, Perasso L, Panascı` M, et al. Acute and chronic catabolic
responses to crossfit® and resistance training in young males. Int J Environ Res Public Health. 2020.
https://doi.org/10.3390/ijerph17197172 PMID: 33007966
9.
Banfi G, Lombardi G, Colombini A, Melegati G. Whole-body cryotherapy in athletes. Sports Medicine.
2010. https://doi.org/10.2165/11531940-000000000-00000 PMID: 20524715
10.
Ferreira-Junior JB, Bottaro M, Vieira A, Siqueira AF, Vieira CA, Durigan JLQ, et al. One session of par-
tial-body cryotherapy (-110˚C) improves muscle damage recovery. Scand J Med Sci Sport. 2015.
https://doi.org/10.1111/sms.12353
11.
Hohenauer E, Taeymans J, Baeyens JP, Clarys P, Clijsen R. The effect of post-exercise cryotherapy
on recovery characteristics: A systematic review and meta-analysis. PLoS One. 2015. https://doi.org/
10.1371/journal.pone.0139028 PMID: 26413718
12.
Ross M, Abbiss C, Laursen P, Martin D, Burke L. Precooling methods and their effects on athletic per-
formance: A systematic review and practical applications. Sports Medicine. 2013. https://doi.org/10.
1007/s40279-012-0014-9 PMID: 23329610
13.
Gonzàlez-Alonso J, Teller C, Andersen SL, Jensen FB, Hyldig T, Nielsen B. Influence of body tempera-
ture on the development of fatigue during prolonged exercise in the heat. J Appl Physiol. 1999. https://
doi.org/10.1152/jappl.1999.86.3.1032 PMID: 10066720
14.
Tyler CJ, Sunderland C, Cheung SS. The effect of cooling prior to and during Exercise on Exercise per-
formance and capacity in the heat: A meta-analysis. British Journal of Sports Medicine. 2015. https://
doi.org/10.1136/bjsports-2012-091739 PMID: 23945034
15.
Olschewski H, Bruck K. Thermoregulatory, cardiovascular, and muscular factors related to exercise
after precooling. J Appl Physiol. 1988. https://doi.org/10.1152/jappl.1988.64.2.803 PMID: 3372438
16.
Marino FE. Methods, advantages, and limitations of body cooling for exercise performance. British Jour-
nal of Sports Medicine. 2002. https://doi.org/10.1136/bjsm.36.2.89 PMID: 11916888
17.
Wegmann M, Faude O, Poppendieck W, Hecksteden A, Fro¨hlich M, Meyer T. Pre-Cooling and sports
performance: A meta-analytical review. Sports Medicine. 2012. https://doi.org/10.2165/11630550-
000000000-00000 PMID: 22642829
18.
Bouzigon R, Grappe F, Ravier G, Dugue B. Whole- and partial-body cryostimulation/cryotherapy: Cur-
rent technologies and practical applications. Journal of Thermal Biology. 2016. pp. 67–81. https://doi.
org/10.1016/j.jtherbio.2016.08.009 PMID: 27712663
19.
Hoshikawa M, Dohi M, Nakamura M. Effects of evening partial body cryostimulation on the skin and
core temperatures. J Therm Biol. 2019. https://doi.org/10.1016/j.jtherbio.2018.12.016 PMID: 30612674
20.
Zalewski P, Bitner A, Słomko J, Szrajda J, Klawe JJ, Tafil-Klawe M, et al. Whole-body cryostimulation
increases parasympathetic outflow and decreases core body temperature. J Therm Biol. 2014. https://
doi.org/10.1016/j.jtherbio.2014.08.001 PMID: 25436954
21.
Duffield R, Coutts A, McCall A, Burgess D. Pre-cooling for football training and competition in hot and
humid conditions. Eur J Sport Sci. 2013. 13:1, 58–67. https://doi.org/10.1080/17461391.2011.589474
22.
Lombardi G, Ziemann E, Banfi G. Whole-body cryotherapy in athletes: From therapy to stimulation. An
updated review of the literature. Frontiers in Physiology. 2017. https://doi.org/10.3389/fphys.2017.
00258 PMID: 28512432
23.
Thevenet D, Tardieu M, Zouhal H, Jacob C, Abderrahman BA, Prioux J. Influence of exercise intensity
on time spent at high percentage of maximal oxygen uptake during an intermittent session in young
endurance-trained athletes. Eur J Appl Physiol. 2007. https://doi.org/10.1007/s00421-007-0540-6
PMID: 17851682
24.
Tanner RK, Fuller KL, Ross MLR. Evaluation of three portable blood lactate analysers: Lactate Pro,
Lactate Scout and Lactate Plus. Eur J Appl Physiol. 2010. https://doi.org/10.1007/s00421-010-1379-9
PMID: 20145946
25.
Binder RK, Wonisch M, Corra U, Cohen-Solal A, Vanhees L, Saner H, et al. Methodological approach
to the first and second lactate threshold in incremental cardiopulmonary exercise testing. European
Journal of Preventive Cardiology. 2008. https://doi.org/10.1097/HJR.0b013e328304fed4 PMID:
19050438
26.
Fonda B, De Nardi M, Sarabon N. Effects of whole-body cryotherapy duration on thermal and cardio-
vascular response. J Therm Biol. 2014. https://doi.org/10.1016/j.jtherbio.2014.04.001 PMID: 24802149
27.
Borg G. Borg’s rating of perceived exertion and pain scales. Hum Kinet. 1998.
PLOS ONE
Precooling in endurance runners
PLOS ONE | https://doi.org/10.1371/journal.pone.0288700
November 22, 2023
11 / 12
28.
Lee KA, Hicks G, Nino-Murcia G. Validity and reliability of a scale to assess fatigue. Psychiatry Res.
1991. https://doi.org/10.1016/0165-1781(91)90027-m PMID: 2062970
29.
Kentta¨ G, Hassme´n P. Overtraining and recovery. A conceptual model. Sports Medicine. 1998. https://
doi.org/10.2165/00007256-199826010-00001 PMID: 9739537
30.
Bakeman R. Recommended effect size statistics for repeated measures designs. Behav Res Methods.
2005. https://doi.org/10.3758/bf03192707 PMID: 16405133
31.
Hopkins WG, Marshall SW, Batterham AM, Hanin J. Progressive statistics for studies in sports medicine
and exercise science. Medicine and Science in Sports and Exercise. 2009. https://doi.org/10.1249/
MSS.0b013e31818cb278 PMID: 19092709
32.
De Nardi M, Facheris C, Ruggeri P, La Torre A, Codella R. High-impact Routines to Ameliorate Trunk
and Lower Limbs Flexibility in Women. Int J Sports Med. 2020. https://doi.org/10.1055/a-1119-7902
PMID: 32668475
33.
Schaal K, Le Meur Y, Louis J, Filliard JR, Hellard P, Casazza G, et al. Whole-body cryostimulation limits
overreaching in elite synchronized swimmers. Med Sci Sports Exerc. 2015. https://doi.org/10.1249/
MSS.0000000000000546 PMID: 25314578
34.
Klimek A, Lubkowska A, Szyguła Z, Chudecka M, Frączek B. Influence of the ten sessionsof the whole
body cryostimulationon aerobic and anaerobic capacity. Int J Occup Med Environ Health. 2010. https://
doi.org/10.2478/v10001-010-0019-2 PMID: 20682489
35.
Filipas L, Martin K, Northey JM, La Torre A, Keegan R, Rattray B. A 4-week endurance training program
improves tolerance to mental exertion in untrained individuals. J Sci Med Sport. 2020. https://doi.org/10.
1016/j.jsams.2020.04.020 PMID: 32456979
36.
Marcora S. Perception of effort during exercise is independent of afferent feedback from skeletal mus-
cles, heart, and lungs. Journal of Applied Physiology. 2009. https://doi.org/10.1152/japplphysiol.90378.
2008 PMID: 18483166
37.
Pires FO, Lima-Silva AE, Bertuzzi R, Casarini DH, Kiss MAPDM, Lambert MI, et al. The influence of
peripheral afferent signals on the rating of perceived exertion and time to exhaustion during exercise at
different intensities. Psychophysiology. 2011. https://doi.org/10.1111/j.1469-8986.2011.01187.x PMID:
21375538
38.
Noakes TD, Tucker R. Do we really need a central governor to explain brain regulation of exercise per-
formance? A response to the letter of Dr. Marcora. European Journal of Applied Physiology. 2008.
https://doi.org/10.1007/s00421-008-0842-3
39.
Costello JT, Algar LA, Donnelly AE. Effects of whole-body cryotherapy (-110˚C) on proprioception and
indices of muscle damage. Scand J Med Sci Sport. 2012. https://doi.org/10.1111/j.1600-0838.2011.
01292.x PMID: 21477164
40.
Russell M, Birch J, Love T, Cook CJ, Bracken RM, Taylor T, et al. The effects of a single whole-body
cryotherapy exposure on physiological, performance, and perceptual responses of professional acad-
emy soccer players after repeated sprint exercise. J Strength Cond Res. 2017. https://doi.org/10.1519/
JSC.0000000000001505 PMID: 27227791
41.
Filipas L, Gallo G, Pollastri L, La Torre AMental fatigue impairs time trial performance in sub-elite under
23 cyclists. PLoS One. 2019. https://doi.org/10.1371/journal.pone.0218405 PMID: 31206523
42.
Filipas L, Mottola F, Tagliabue G, La Torre A. The effect of mentally demanding cognitive tasks on row-
ing performance in young athletes. Psychol Sport Exerc. 2018. https://doi.org/10.1016/j.psychsport.
2018.08.002
43.
Hinckson EA, Hopkins WG. Reliability of time to exhaustion analyzed with critical-power and log-log
modeling. Med Sci Sports Exerc. 2005; 37(4):696–701. https://doi.org/10.1249/01.mss.0000159023.
06934.53 PMID: 15809572
44.
Laursen PB, Francis GT, Abbiss CR, Newton MJ, Nosaka K. Reliability of time-to-exhaustion versus
time-trial running tests in runners. Med Sci Sports Exerc. 2007; 39(8):1374–9. https://doi.org/10.1249/
mss.0b013e31806010f5 PMID: 17762371
45.
Partridge EM, Cooke J, McKune A, Pyne DB. Whole-body cryotherapy: Potential to enhance athlete
preparation for competition? Front Physiol. 2019. https://doi.org/10.3389/fphys.2019.01007 PMID:
31447697
PLOS ONE
Precooling in endurance runners
PLOS ONE | https://doi.org/10.1371/journal.pone.0288700
November 22, 2023
12 / 12
| Partial-body cryostimulation procured performance and perceptual improvements in amateur middle-distance runners. | 11-22-2023 | De Nardi, Massimo,Filipas, Luca,Facheris, Carlo,Righetti, Stefano,Tengattini, Marco,Faelli, Emanuela,Bisio, Ambra,Gallo, Gabriele,La Torre, Antonio,Ruggeri, Piero,Codella, Roberto | eng |
PMC8914642 |
Citation: Chahal, A.K.; Lim, J.Z.; Pan,
J.-W.; Kong, P.W. Inter-Unit
Consistency and Validity of 10-Hz
GNSS Units in Straight-Line Sprint
Running. Sensors 2022, 22, 1888.
https://doi.org/10.3390/s22051888
Academic Editor: Mario
Munoz-Organero
Received: 19 January 2022
Accepted: 25 February 2022
Published: 28 February 2022
Publisher’s Note: MDPI stays neutral
with regard to jurisdictional claims in
published maps and institutional affil-
iations.
Copyright:
© 2022 by the authors.
Licensee MDPI, Basel, Switzerland.
This article is an open access article
distributed
under
the
terms
and
conditions of the Creative Commons
Attribution (CC BY) license (https://
creativecommons.org/licenses/by/
4.0/).
sensors
Article
Inter-Unit Consistency and Validity of 10-Hz GNSS Units in
Straight-Line Sprint Running
Amandeep Kaur Chahal, Jolene Ziyuan Lim, Jing-Wen Pan
and Pui Wah Kong *
Physical Education and Sports Science Academic Group, National Institute of Education,
Nanyang Technological University, Singapore 637616, Singapore; [email protected] (A.K.C.);
[email protected] (J.Z.L.); [email protected] (J.-W.P.)
* Correspondence: [email protected]
Abstract: The present study aimed to investigate the inter-unit consistency and validity of multiple
10-Hz Catapult Global Navigation Satellite System (GNSS) units in measuring straight-line sprint
distances and speeds. A total of 13 participants performed one 45.72-m linear sprint at maximum
effort while wearing all eight GNSS units at once. Total run distance and peak speed recorded using
GNSS units during the sprint duration were extracted for analysis. Sprint time and peak speed were
also obtained from video recordings as reference values. Inter-unit consistency was assessed using
intraclass correlation coefficients (ICC) and standard errors of measurements (SEM). For a validity
test, one-sample t-tests were performed to compare each GNSS unit’s distance with the known
distance. Additionally, Wilcoxon signed-rank tests were performed to compare each unit’s peak
speed with the reference peak speed measured using video analysis. Results showed poor inter-unit
consistency for both distance (ICC = 0.131; SEM = 8.8 m) and speed (ICC = 0.323; SEM 1.3 m/s)
measurements. For validity, most units recorded a total distance (44.50 m to 52.69 m) greater than the
known distance of 45.72 m and a lower peak speed (7.25 (0.51) m/s) than the video-based reference
values (7.78 (0.90) m/s). The present findings demonstrate that there exist variations in distance and
speed measurements among different units of the same GNSS system during straight-line sprint
running. Practitioners should be aware of the window of errors associated with GNSS measurements
and interpret the results with caution. When making comparisons over a season, players should wear
the same unit every time if logistically possible.
Keywords: Global Navigation Satellite System; reliability; distance; speed; video; movement analysis
1. Introduction
In sports, the movement characteristics of players during competitions and training
are of interest for in-game analyses. Traditionally, player activity data were manually
collected on pen and paper, which was extremely labor-intensive and time-consuming [1,2].
With technological advancements in time-motion analysis, more convenient methods, such
as video analysis, have been regularly used to track player movements during competitions
and training. However, video analysis can be troublesome to set up and requires extensive
manual analysis after data collection. The development of Global Navigation Satellite
System (GNSS) and Global Positioning System (GPS) units, which are light, small, and
portable, allows for simultaneous movement patterns analyses of multiple players [3,4].
Since then, the use of GNSS/GPS for athlete tracking has become widespread in various
sports, such as soccer, rugby, and field hockey [3,5–7] due to the ease of data collection and
quality of analysis provided by these systems [8–11]. Additionally, GNSS/GPS units are
the conventional technology used for the assessment of external training load variables
in team sports due to their ability to give real-time feedback. This is essential, given the
limited amount of time to process data and carry out post-session analysis [3,12,13].
GPS is a navigation system based on connections to satellites that allows locations
of users to be triangulated through signals sent out by the satellites and received by the
Sensors 2022, 22, 1888. https://doi.org/10.3390/s22051888
https://www.mdpi.com/journal/sensors
Sensors 2022, 22, 1888
2 of 10
units [14]. The accuracy of data is dependent on the configuration of the satellites in relation
to the receiver and how evenly spaced they are, known as the dilution of precision (DOP).
Triangulation of position is the most accurate when one satellite is directly overhead the
receiver while the rest of the satellites are evenly spaced around the horizon (DOP = 1).
It has generally been suggested that GPS units require at least 4 satellites for data to be
considered accurate. In addition, satellites that are more evenly spaced are considered
stronger than when satellites are close together [14]. GPS units are usually combined with
microsensors such as accelerometers that are capable of recording movements in three
planes, allowing the intensity of body load (also known as player load in some systems) to
be measured. In addition, the inclusion of gyroscopes and magnetometers in these units
allows for directional orientation and rotational velocity to also be measured [15]. Sampling
rates of GPS units may range from 1 to 15 Hz, indicating the multiple speeds at which the
GPS units collect data. Existing studies have shown that higher sampling rates increase
the accuracy of performance indicators [16–18] recorded by the GPS units. For instance,
10-Hz GPS units are more accurate than those of lower sampling rates in measuring total
distance covered during both linear activities and sport-specific circuits and measuring
peak speed [19]. No additional accuracy has been found between 10-Hz and 15-Hz GPS
units [19], indicating that a sampling rate of 10 Hz could be sufficient.
GNSS/GPS units provide a multitude of movement variables including distance,
speed, acceleration/deceleration, and metabolic power [3,20]. These movements can be
purposefully analyzed (external training load) to comprehend the positional demands
in sports, allowing practitioners to design programs that accurately emulate and equip
athletes for their specific sport [3]. Furthermore, the GNSS/GPS data have also been proven
useful in aiding practitioners to understand physiological and technical demands of their
players through the extraction of various external training load measures such as volume,
intensity, and frequency [21]. Such information can inform and guide coaches and sport
scientists to develop appropriate conditioning and recovery plans [22,23].
While GNSS/GPS units provide practical and useful feedback, environmental objects
such as surrounding tall buildings [24], atmospheric pressure [25], as well as the level of
satellite giving out the signals (with signals from lower satellites having to go through
more atmosphere) can result in obstruction of signals, leading to lower signal-to-noise
ratio and lower accuracy in measurements. Hence, it is important to establish the validity
and reliability of these units before applying them in sports [4,16,26]. Testing validity
provides an understanding of the differences between the measures recorded by the units
and standard measures. Reliability testing, on the other hand, tests reproducibility of values
when the same test is repeated by another unit. While studies have generally agreed that
GNSS/GPS devices can be reliable in straight-line running, there is a sizable inconsistency
in accuracy among the models of GNSS/GPS manufacturers [12,18,27]. Imparting the
validation of one system to another can be imprecise even if it is introduced by the same
manufacturer [28]. While Johnston and co-workers [29] used a different software to collect
and analyze GPS data collected by other brands of GPS units, the authors cautioned that
the mismatching in GPS models may have influenced the movement demand data. Hence,
it is vital to carry out an independent and thorough trial for each new GNSS/GPS device
(hardware) and its analysis tool (software).
Within the same system, high consistency between different GNSS/GPS units is critical
especially in team sports whereby each player wears an independent unit. Previous studies
examining the inter-unit reliability of GPS units have placed multiple units on solid objects
such as a golf cart and motorcycle [26], plastic sled [30], and a trundle wheel [31]. It should
be noted that the movement trajectories of solid objects may differ from those of the human
players who can freely move individual body segments in different directions at various
magnitudes. There are very few studies placing GPS units on human participants and these
studies typically compared among only two to four units each time [27,32]. To the best of
the authors’ knowledge, only one study has tested the inter-unit reliability of eight GPS
units on an individual [33]. Although their study found inter- and intra-receiver reliability
Sensors 2022, 22, 1888
3 of 10
to be acceptable, the GPS units were sampled at 1 Hz which is far below the recommended
frequency of 10 Hz for accurate measurements [19]. Thus, there is a need to examine the
inter-unit consistency and validity of multiple GNSS/GPS units sampled at sufficiently
high frequency (i.e., at least 10 Hz) with the units placed on human participants and not
solid objects.
This study, therefore, aimed to investigate the inter-unit consistency and validity of
10-Hz Catapult GNSS (S5 OptimEye, Catapult Innovations, Melbourne, Australia) units
during straight-line sprint running. Eight GNSS units were analyzed using the Sprint
software developed by Catapult. It was hypothesized that all GNSS units, when placed on
human participants, would be consistent and accurate in measuring distances and peak
speeds during sprint running [18].
2. Materials and Methods
2.1. Participants
This study was approved by the Nanyang Technological University Institutional
Review Board (IRB-2020-09-033). Thirteen active participants (4 males, 9 females) were
recruited via convenient sampling [age 21.6 (1.6) years, height 170.6 (7.7) cm, body mass 63.1
(10.1) kg]. To be eligible for this study, participants must have been training with a sports
team at least twice a week and had minimally a year of experience in the specified sport.
Additionally, they were required to be injury-free and pain-free at the time of the study.
2.2. Equipment
The Catapult S5 OptimEye GNSS system was used in the present study. Accessing
GPS and Global Navigation Satellite System (GLONASS) satellite constellations, this
GNSS system ensures high-quality data even in challenging performance environments
(https://www.catapultsports.com (accessed on 1 October 2020)). Eight 10-Hz GNSS units
were worn at once on each participant during the test (Figure 1). The eight GNSS units
were placed near the mid-back area using a custom-made strap, with slightly different
positions. While the tightness of strap was adjusted to fit individual body sizes, the relative
positions of the GNSS units on the strap remained consistent across all participants.
p y
y
y
g
different directions at various magnitudes. There are very few studies placing GPS units
on human participants and these studies typically compared among only two to four units
each time [27,32]. To the best of the authors’ knowledge, only one study has tested the
inter-unit reliability of eight GPS units on an individual [33]. Although their study found
inter- and intra-receiver reliability to be acceptable, the GPS units were sampled at 1 Hz
which is far below the recommended frequency of 10 Hz for accurate measurements [19].
Thus, there is a need to examine the inter-unit consistency and validity of multiple
GNSS/GPS units sampled at sufficiently high frequency (i.e., at least 10 Hz) with the units
placed on human participants and not solid objects.
This study, therefore, aimed to investigate the inter-unit consistency and validity of
10-Hz Catapult GNSS (S5 OptimEye, Catapult Innovations, Melbourne, Australia) units
during straight-line sprint running. Eight GNSS units were analyzed using the Sprint
software developed by Catapult. It was hypothesized that all GNSS units, when placed
on human participants, would be consistent and accurate in measuring distances and peak
speeds during sprint running [18].
2. Materials and Methods
2.1. Participants
This study was approved by the Nanyang Technological University Institutional
Review Board (IRB-2020-09-033). Thirteen active participants (4 males, 9 females) were
recruited via convenient sampling [age 21.6 (1.6) years, height 170.6 (7.7) cm, body mass
63.1 (10.1) kg]. To be eligible for this study, participants must have been training with a
sports team at least twice a week and had minimally a year of experience in the specified
sport. Additionally, they were required to be injury-free and pain-free at the time of the
study.
2.2. Equipment
The Catapult S5 OptimEye GNSS system was used in the present study. Accessing
GPS and Global Navigation Satellite System (GLONASS) satellite constellations, this
GNSS system ensures high-quality data even in challenging performance environments
(https://www.catapultsports.com (accessed on 1 October 2020)). Eight 10-Hz GNSS units
were worn at once on each participant during the test (Figure 1). The eight GNSS units
were placed near the mid-back area using a custom-made strap, with slightly different
positions. While the tightness of strap was adjusted to fit individual body sizes, the
relative positions of the GNSS units on the strap remained consistent across all
participants.
Figure 1. Each participant wore eight GNSS units at once in the maximal sprint test.
Figure 1. Each participant wore eight GNSS units at once in the maximal sprint test.
2.3. Experimental Protocol
This experiment involved one single visit to the field hockey pitch at the National
Institute of Education, Nanyang Technological University, Singapore. GNSS data were
collected outdoors without high surrounding buildings to enhance satellite reception [34].
After sufficient warm-up and putting on the strap with 8 GNSS units, participants were
asked to perform one straight-line sprint across the hockey pitch with maximum effort
(Figure 2). A sprint distance of 45.72 m was chosen as it is exactly half of the hockey pitch
Sensors 2022, 22, 1888
4 of 10
with a clear marked line. This distance also allowed sufficient time for participants to reach
their peak speed. To help participants discern the start and end points, the sprint area was
marked out using colored cones. Participants were asked to run pass the 45.72 m end-line
before they could slow down. The sprint test was recorded using two video cameras at
60 Hz for subsequent determination of sprint times and peak speeds. To minimize the effect
of camera lens distortion, we used two smartphone cameras to cover the entire 45.72 m
range (Figure 2). The midline distance of 22.86 m was used to calibrate each camera (0 to
22.86 m, 22.86 to 45.72 m) as the midline can be clearly seen from both camera views.
This experiment involved one single visit to the field hockey pitch at the National
Institute of Education, Nanyang Technological University, Singapore. GNSS data were
collected outdoors without high surrounding buildings to enhance satellite reception [34].
After sufficient warm-up and putting on the strap with 8 GNSS units, participants were
asked to perform one straight-line sprint across the hockey pitch with maximum effort
(Figure 2). A sprint distance of 45.72 m was chosen as it is exactly half of the hockey pitch
with a clear marked line. This distance also allowed sufficient time for participants to
reach their peak speed. To help participants discern the start and end points, the sprint
area was marked out using colored cones. Participants were asked to run pass the 45.72
m end-line before they could slow down. The sprint test was recorded using two video
cameras at 60 Hz for subsequent determination of sprint times and peak speeds. To
minimize the effect of camera lens distortion, we used two smartphone cameras to cover
the entire 45.72 m range (Figure 2). The midline distance of 22.86 m was used to calibrate
each camera (0 to 22.86 m, 22.86 to 45.72 m) as the midline can be clearly seen from both
camera views.
Figure 2. Experimental set-up of the sprint test over half a field hockey pitch (45.72 m) with two
smartphone cameras recording the performances (Camera 1: 0 to 22.86 m, Camera 2: 22.86 m to 45.72
m).
GNSS units were switched on at least 5 min before the units were strapped on the
participants. After strapping on all units, participants were verbally briefed and then
asked to familiarize themselves with the task. The GNSS units were switched on for more
than 15 min to receive the complete almanac before the commencement of the test.
Participants were also instructed to stay still for 30 s, before the start of the sprint. This
was to enable subsequent determination of the start time for each trial when the speed
increased sharply from zero.
2.4. Data Processing
The GNSS movement data were downloaded using the manufacturer’s software
(Catapult Sprint Version 5.1.7, Melbourne, Australia) at the default ‘GPS rate’ of 10 Hz.
Customized MATLAB codes were written to extract the relevant distance and speed time-
series data using MATLAB (R2021a, MathWorks, Natick, MA, USA). The start of the
sprint was identified from a sharp and continuous increase in speed above a threshold of
0.5 m/s. The duration each participant took to complete the 45.72 m distance was obtained
based on the video recordings of the sprint. This sprint duration was then used to
determine the end time of the sprint in the GNSS data. From the start to the end of the
sprint, total distance traveled, and peak speeds were obtained from each of the 8 GNSS
Figure 2. Experimental set-up of the sprint test over half a field hockey pitch (45.72 m) with two
smartphone cameras recording the performances (Camera 1: 0 to 22.86 m, Camera 2: 22.86 m to 45.72 m).
GNSS units were switched on at least 5 min before the units were strapped on the
participants. After strapping on all units, participants were verbally briefed and then asked
to familiarize themselves with the task. The GNSS units were switched on for more than
15 min to receive the complete almanac before the commencement of the test. Participants
were also instructed to stay still for 30 s, before the start of the sprint. This was to enable
subsequent determination of the start time for each trial when the speed increased sharply
from zero.
2.4. Data Processing
The GNSS movement data were downloaded using the manufacturer’s software
(Catapult Sprint Version 5.1.7, Melbourne, Australia) at the default ‘GPS rate’ of 10 Hz.
Customized MATLAB codes were written to extract the relevant distance and speed time-
series data using MATLAB (R2021a, MathWorks, Natick, MA, USA). The start of the sprint
was identified from a sharp and continuous increase in speed above a threshold of 0.5 m/s.
The duration each participant took to complete the 45.72 m distance was obtained based on
the video recordings of the sprint. This sprint duration was then used to determine the end
time of the sprint in the GNSS data. From the start to the end of the sprint, total distance
traveled, and peak speeds were obtained from each of the 8 GNSS units. Raw GNSS data
were used without further down sampling, filtering, or smoothing procedures. Due to
transmission and technical errors, it was not possible to obtain complete data sets from all
8 GNSS units throughout all trials. Among the 13 participants, 7 had complete data set and
6 had missing data from either 1 or 2 GNSS units.
For validity analysis, a reference value of the gold standard was needed. In the present
study, the total distance was 45.72 m, which was the known size of half of a standard
field hockey pitch. This distance was also confirmed by experimental measurement using
a trundle wheel. To calculate the speed from position data, manual digitization of the
player’s center of the head was performed through the sprint duration using the software
Kinovea (version 0.9.3, Kinovea, Bordeaux, France, available for download at: http://
Sensors 2022, 22, 1888
5 of 10
www.kinovea.org (accessed on 15 April 2021)). The present study used video analysis
as the gold standard for kinematics, which is aligned with previous work evaluating the
accuracy of 10 Hz GPS system [12]. Kinovea has been demonstrated as a reliable and
accurate tool for video-based angular and linear measurements via digitization of x- and
y-axis coordinates [35]. While an optimal angle of 90◦ was recommended, an accepted level
accuracy was also established when the camera was placed within an angle range of 45◦ to
90◦ [35].
Figure 3 illustrates examples of the speed-time data measured using one GNSS unit and
video analysis. The raw speed data from videos were low-passed filter at 10 Hz to remove
the noise associated with manual digitization. The peak value of the filtered speed data
during the entire sprint duration was then identified. This video-based peak speed was used
as a reference value in the subsequent validity analysis of GNSS units. The mean (SD) of the
raw and filtered peak speeds were 7.82 (0.81) m/s and 7.78 (0.90) m/s, respectively.
present study, the total distance was 45.72 m, which was the known size of half of a
standard field hockey pitch. This distance was also confirmed by experimental
measurement using a trundle wheel. To calculate the speed from position data, manual
digitization of the player’s center of the head was performed through the sprint duration
using the software Kinovea (version 0.9.3, Kinovea, Bordeaux, France, available for
download at: http://www.kinovea.org (accessed on 15 April 2021)). The present study
used video analysis as the gold standard for kinematics, which is aligned with previous
work evaluating the accuracy of 10 Hz GPS system [12]. Kinovea has been demonstrated
as a reliable and accurate tool for video-based angular and linear measurements via
digitization of x- and y-axis coordinates [35]. While an optimal angle of 90° was
recommended, an accepted level accuracy was also established when the camera was
placed within an angle range of 45° to 90° [35].
Figure 3 illustrates examples of the speed-time data measured using one GNSS unit
and video analysis. The raw speed data from videos were low-passed filter at 10 Hz to
remove the noise associated with manual digitization. The peak value of the filtered speed
data during the entire sprint duration was then identified. This video-based peak speed
was used as a reference value in the subsequent validity analysis of GNSS units. The mean
(SD) of the raw and filtered peak speeds were 7.82 (0.81) m/s and 7.78 (0.90) m/s,
respectively.
Figure 3. Representative raw speed-time from one participant measured using Kinovea video
analysis and one GNSS unit.
2.5. Statistical Analyses
Statistical analyses were carried out on JASP (version 0.14.1, JASP Team 2020) and SPSS
(version 26.0, IBM Corp., Armonk, NY, USA). Data are expressed as mean (standard
deviation). An alpha level of p < 0.05 was set as the level of significance. Inter-unit
consistency was assessed using intraclass correlation coefficients (ICC). ICC was
interpreted as slight (<0.20), fair (0.21–0.40), moderate (0.41–60), substantial (0.61–0.80), or
Figure 3. Representative raw speed-time from one participant measured using Kinovea video analysis
and one GNSS unit.
2.5. Statistical Analyses
Statistical analyses were carried out on JASP (version 0.14.1, JASP Team 2020) and
SPSS (version 26.0, IBM Corp., Armonk, NY, USA). Data are expressed as mean (standard
deviation). An alpha level of p < 0.05 was set as the level of significance. Inter-unit
consistency was assessed using intraclass correlation coefficients (ICC). ICC was interpreted
as slight (<0.20), fair (0.21–0.40), moderate (0.41–60), substantial (0.61–0.80), or almost perfect
reliability (>0.80) [36,37]. Standard error of measurement (SEM) was calculated from the
ICC results using the formula: SEM = SD × √(1 − ICC).
For the validity assessment, one-sample t-tests were performed to compare the distance
measured using each GNSS unit with the known distance of 45.72 m. Effect sizes were
indicated by Cohen’s d and interpreted as small (0.2 ≤ d < 0.5), medium (0.5 ≤ d < 0.8),
or large (d ≥ 0.8). Since the speed data were not normally distributed, non-parametric
statistical tests were employed. Specially, Wilcoxon signed-rank tests were used to compare
each GNSS unit’s peak speed with the reference speed measured using video analysis.
Effect size (r) for the Wilcoxon signed-rank tests was calculated from the Z-value and
interpreted as small (0.1 ≤ |r| < 0.3), medium (0.3 ≤ |r| < 0.5), or large (|r| ≥ 0.5).
Sensors 2022, 22, 1888
6 of 10
3. Results
3.1. Inter-Unit Consistency
The results of ICC analysis showed slight reliability for the total sprint distance and
fair reliability for peak speed (Table 1). These results indicate that the 8 tested GNSS units
are not sufficiently consistent among themselves.
Table 1. Reliability statistical outputs to assess inter-unit consistency.
GNSS Variables
ICC
95% Confidence Intervals
SEM
Total distance
0.131
[−0.024, 0.556]
8.8 m
Peak speed
0.323
[0.101, 0.736]
1.3 m/s
Note. ICC denotes intraclass correlation coefficients; SEM denotes standard error of measurement.
3.2. Validity
Most GNSS units recorded a total distance greater than the known distance of 45.72 m
(Table 2). While statistical significance was only found in two units, the effect sizes of the
differences were large across all units. These results indicate that GNSS units, although
belonging to the same system, do not always measure distance with the same degree
of accuracy.
Table 2. Validity of GNSS distance measurements against known distance of 45.72 m.
GNSS Units
Mean (SD)
p-Value
Effect Size (d)
Unit 1 (n = 13)
49.77 (5.92)
0.030 *
8.41
Large
Unit 2 (n = 13)
46.69 (10.62)
0.747
4.49
Large
Unit 3 (n = 10)
44.50 (8.55)
0.663
5.29
Large
Unit 4 (n = 13)
52.23 (10.11)
0.039 *
5.17
Large
Unit 5 (n = 11)
52.00 (10.13)
0.067
5.13
Large
Unit 6 (n = 12)
50.83 (8.57)
0.063
5.93
Large
Unit 7 (n = 12)
47.50 (8.06)
0.460
5.89
Large
Unit 8 (n = 13)
52.69 (12.18)
0.061
4.33
Large
Note. * Significant differences detected using one-sample t-tests (p < 0.05).
Compared with the reference speed data obtained from video analysis, Unit 4 mea-
sured significantly higher peak speed (p = 0.010, large effect size, Table 3). No significant
differences were identified between other GNSS units and video analysis, with data of
4 units approaching statistical significance (Units 1, 3, 7, 8). In general, most GNSS units
measured a lower peak speed (7.25 (0.51) m/s) than the video-based value (7.78 (0.90) m/s)
and the effect sizes of the differences were medium to large.
Table 3. Validity of GNSS peak speed measurements against video analysis.
GNSS Units
Mean (SD)
p-Value
Effect Size (r)
Unit 1 (n = 13)
7.04 (1.15)
0.057
0.604
Large
Unit 2 (n = 13)
6.89 (1.98)
0.127
0.495
Medium
Unit 3 (n = 10)
6.92 (1.03)
0.064
0.673
Large
Unit 4 (n = 13)
7.11 (0.89)
0.010 *
0.780
Large
Unit 5 (n = 11)
7.37 (1.55)
0.416
0.303
Medium
Unit 6 (n = 12)
8.40 (2.53)
0.970
0.026
Negligible
Unit 7 (n = 12)
6.86 (1.34)
0.064
0.615
Large
Unit 8 (n = 13)
7.37 (1.05)
0.057
0.604
Large
Note. * Significant differences detected using Wilcoxon signed-rank tests (p < 0.05). Group mean (SD) of video-
based peak speed was 7.78 (0.90) m/s.
4. Discussion
The aim of the study was to investigate the inter-unit reliability and validity of multiple
10-Hz Catapult GNSS units during straight-line sprint running. Inter-unit consistency was
Sensors 2022, 22, 1888
7 of 10
assessed among eight GNSS units worn on each participant, and validity was tested by
comparing total distance and peak speed against criterion-referenced values. The most
prevailing outcomes were that despite all GNSS units belonging to the same system, low
inter-unit reliability and varied accuracies in distance and speed measurements were found
during fast speed running.
4.1. Distance
We originally expect that all GNSS units, when placed on the participant, would
be consistent and accurate in measuring total distance traveled during 45.72-m sprint.
However, there was only slight reliability for inter-unit consistency among the eight GNSS
units and two out of eight units (Units 1 and 4, Table 2) had significantly different values
from the criterion distance. In addition, seven out of the eight GNSS units overestimated the
values during the straight-line sprint. These results in the present study are somewhat in
congruence with previous research which reported moderate errors when measuring total
distance over very high-speed running (>5.56 m/s) [17]. Additionally, overestimation of the
total distance measured using GNSS units has also been found when the sprinting distances
were set as 15 m and 30 m [34,35]. The reliability and accuracy may also be affected by rapid
changes in speed during the acceleration phase of the sprint. A previous study revealed
that distance measures over the post-acceleration phase of 20–40 m were more accurate
than the acceleration phase of 0–20 m in a 40-m linear acceleration run [16], suggesting that
smaller variations in speed may facilitate more accurate measures in distance. In the present
study, participants started from a stationary position and were asked to sprint as fast as
they could using maximal effort. Hence, phases with great variations in speed could have
resulted in inconsistent and less accurate total distances measurement across different units.
It is also possible that some participants did not sprint in a perfectly straight line hence
covering a longer distance than the reference value of 45.72 m. Although the deviation
from a straight line can be expected to be quite small, this could partly explain why seven
out of eight GNSS units recorded a longer total distance than the reference value based on
the distance between standard marked lines on the field. Finally, the GNSS units could
miss data owing to the poor satellite connection [19]. This may have caused measurement
errors in certain GNSS units, leading to inconsistency among the different units.
4.2. Peak Speed
This study hypothesized that multiple GNSS units of the same system would be
consistent and accurate in measuring peak speed during a maximal effort sprint. The
results demonstrated fair reliability among the eight GNSS units and that seven out of the
eight units generally measured lower peak speeds than that video-based reference values
(Table 3). The results are not in line with previous findings which suggested confidence in
10-Hz GNSS units being able to accurately measure consistent speeds and velocities [18].
The discrepancy in the peak speeds measured can be attributed to the compromises when
measuring instantaneous velocities during great decelerations [38] and accelerations [28].
Hence, rapid changes in speed during the acceleration phase of the 45.72-m sprint in the
present study could affect the accuracy of the GNSS units when measuring peak speeds.
Higher accuracy and inter-unit reliability may be expected if GNSS units are applied to
measure speed during a stable phase with small decelerations or accelerations.
Compared with the video analysis which was used as the golden standard for the
speed measurement, only one GNSS unit displayed statistically significant result (Unit 4,
Table 3). It is worth noting that the effect sizes of the differences were medium to large
across all units regardless of statistical significance. As the GNSS units tend to register
lower peak speeds (7.25 (0.51) m/s) than video-based reference values (7.78 (0.90) m/s),
such differences cannot be disregarded. Sport practitioners should keep in mind that GNSS
readings may slightly underestimate peak speeds during high-speed running and interpret
the results with consideration of the error window (SEM = 1.3 m/s).
Sensors 2022, 22, 1888
8 of 10
4.3. Limitations
There were a few limitations to the current study. Firstly, six participants had miss-
ing data due to either faulty units or the poor connection to the satellites. The current
sample size of 13 participants was smaller than expected since the experiment was halted
prematurely due to the COVID-19 pandemic. A larger sample size may have brought
about more reliable results, which was unfortunately not possible due to time constraints.
Secondly, environmental factors (e.g., presence of clouds) during the experiment may have
occurred and affected the results. Thirdly, we acknowledge that the use of smartphone
cameras can reduce the accuracy of data collected due to optical effects, such as lens dis-
tortion and parallax error. For fast sprint movements, the relatively low frame rate of
60 Hz could have also compromised accuracy of speed and time data collected. Lastly,
the current study investigated only two variables of linear sprints (total distance and peak
speed). In the future, researchers should expand to other variables and movement types
concerning the utilizations of GNSS units in sports such as change in direction, acceleration,
and deceleration.
5. Conclusions
In team sports, high consistency between different GNSS units is critical as coaches
compare the movement characteristics across players in a game or training. This study
revealed that there exist variations in distance and speed measurements among eight
GNSS units worn by participants at the same time. In general, GNSS units may lead to
an overestimation of total distance and underestimation of peak speed during high-speed
sprint running. Practitioners should be aware of the window of errors associated with
GNSS measurements and interpret the results with caution. This is especially important for
data collected during sport competitions or training which involve movement demands
at high speeds. When making comparisons over a season, players should wear the same
GNSS unit every time if logistically possible.
Despite some limitations, the use of GNSS/GPS technology is still widespread, and
it offers practical insights to players’ movements characteristics and playing demands.
In view of the rapid advancement in technology, it may be possible to improve current
GNSS/GPS systems so as to enhance their inter-unit consistency and measurement accura-
cies across different movement types including high-speed sprinting.
Author Contributions: Conceptualization, J.Z.L. and P.W.K.; methodology, J.Z.L., A.K.C. and P.W.K.;
validation, A.K.C.; formal analysis, A.K.C., J.-W.P. and P.W.K.; writing—original draft preparation,
A.K.C., J.Z.L. and J.-W.P.; writing—review and editing, A.K.C., J.Z.L., J.-W.P. and P.W.K.; supervision,
P.W.K. All authors have read and agreed to the published version of the manuscript.
Funding: We wish to acknowledge the funding support for this project from Nanyang Technological
University under the URECA Undergraduate Research Programme.
Institutional Review Board Statement: The study was conducted according to the guidelines of the
Declaration of Helsinki, and approved by the Nanyang Technological University Institutional Review
Board (IRB-2020-09-033).
Informed Consent Statement: Informed consent was obtained from all subjects involved in the study.
Acknowledgments: The authors would like to acknowledge all participants for taking their time to
participate in this study.
Conflicts of Interest: The authors declare no conflict of interest.
References
1.
Roberts, S.; Trewartha, G.; Stokes, K. A Comparison of Time–Motion Analysis Methods for Field-Based Sports. Int. J. Sports
Physiol. Perform. 2006, 1, 388–399. [CrossRef]
2.
Carling, C.; Bloomfield, J.; Nelsen, L.; Reilly, T. The Role of Motion Analysis in Elite Soccer. Sports Med. 2008, 38, 839–862.
[CrossRef]
Sensors 2022, 22, 1888
9 of 10
3.
Cummins, C.; Orr, R.; O’Connor, H.; West, C. Global Positioning Systems (GPS) and Microtechnology Sensors in Team Sports: A
Systematic Review. Sports Med. 2013, 43, 1025–1042. [CrossRef]
4.
Shergill, A.S.; Twist, C.; Highton, J. Importance of GNSS data quality assessment with novel control criteria in professional soccer
match-play. Int. J. Perform. Anal. Sport 2021, 21, 820–830. [CrossRef]
5.
Duthie, G.; Pyne, D.; Hooper, S. Time motion analysis of 2001 and 2002 super 12 rugby. J. Sports Sci. 2005, 23, 523–530. [CrossRef]
6.
Sirotic, A.C.; Coutts, A.J.; Knowles, H.; Catterick, C. A comparison of match demands between elite and semi-elite rugby league
competition. J. Sports Sci. 2009, 27, 203–211. [CrossRef]
7.
Spencer, M.; Lawrence, S.; Rechichi, C.; Bishop, D.; Dawson, B.; Goodman, C. Time–motion analysis of elite field hockey, with
special reference to repeated-sprint activity. J. Sports Sci. 2004, 22, 843–850. [CrossRef]
8.
Aughey, R.J.; Falloon, C. Real-time versus post-game GPS data in team sports. J. Sci. Med. Sport 2010, 13, 348–349. [CrossRef]
9.
Gabbett, T.J.; Jenkins, D.G.; Abernethy, B. Physical demands of professional rugby league training and competition using
microtechnology. J. Sci. Med. Sport 2012, 15, 80–86. [CrossRef]
10.
Hartwig, T.B.; Naughton, G.; Searl, J. Motion Analyses of Adolescent Rugby Union Players: A Comparison of Training and Game
Demands. J. Strength Cond. Res. 2011, 25, 966–972. [CrossRef]
11.
Johnston, R.J.; Watsford, M.L.; Pine, M.J.; Spurrs, R.W.; Murphy, A.; Pruyn, E.C. Movement Demands and Match Performance in
Professional Australian Football. Laryngo-Rhino-Otologie 2012, 33, 89–93. [CrossRef]
12.
Beato, M.; Bartolini, D.; Ghia, G.; Zamparo, P. Accuracy of a 10 Hz GPS Unit in Measuring Shuttle Velocity Performed at Different
Speeds and Distances (5–20 M). J. Hum. Kinet. 2016, 54, 15–22. [CrossRef] [PubMed]
13.
Buchheit, M.; Simpson, B.M. Player-Tracking Technology: Half-Full or Half-Empty Glass? Int. J. Sports Physiol. Perform. 2017, 12,
35–41. [CrossRef] [PubMed]
14.
Witte, T.; Wilson, A. Accuracy of non-differential GPS for the determination of speed over ground. J. Biomech. 2004, 37, 1891–1898.
[CrossRef] [PubMed]
15.
Lutz, J.; Memmert, D.; Raabe, D.; Dornberger, R.; Donath, L. Wearables for Integrative Performance and Tactic Analyses:
Opportunities, Challenges, and Future Directions. Int. J. Environ. Res. Public Health 2019, 17, 59. [CrossRef] [PubMed]
16.
Jennings, D.; Cormack, S.; Coutts, A.J.; Boyd, L.; Aughey, R.J. The Validity and Reliability of GPS Units for Measuring Distance in
Team Sport Specific Running Patterns. Int. J. Sports Physiol. Perform. 2010, 5, 328–341. [CrossRef]
17.
Rampinini, E.; Alberti, G.; Fiorenza, M.; Riggio, M.; Sassi, R.; Borges, T.O.; Coutts, A.J. Accuracy of GPS Devices for Measuring
High-intensity Running in Field-based Team Sports. Laryngo-Rhino-Otologie 2014, 36, 49–53. [CrossRef]
18.
Scott, M.T.U.; Scott, T.J.; Kelly, V.G. The Validity and Reliability of Global Positioning Systems in Team Sport. J. Strength Cond. Res.
2016, 30, 1470–1490. [CrossRef]
19.
Johnston, R.J.; Watsford, M.L.; Pine, M.J.; Spurrs, R.W.; Murphy, A.J.; Pruyn, E.C. The Validity and Reliability of 5-hZ Global
Positioning System Units to Measure Team Sport Movement Demands. J. Strength Cond. Res. 2012, 26, 758–765. [CrossRef]
20.
Osgnach, C.; Poser, S.; Bernardini, R.; Rinaldo, R.; DI Prampero, P.E. Energy Cost and Metabolic Power in Elite Soccer. Med. Sci.
Sports Exerc. 2010, 42, 170–178. [CrossRef] [PubMed]
21.
Akenhead, R.; Harley, J.A.; Tweddle, S.P. Examining the External Training Load of an English Premier League Football Team with
Special Reference to Acceleration. J. Strength Cond. Res. 2016, 30, 2424–2432. [CrossRef] [PubMed]
22.
Dawson, B.; Gow, S.; Modra, S.; Bishop, D.; Stewart, G. Effects of immediate post-game recovery procedures on muscle soreness,
power and flexiblity levels over the next 48 hours. J. Sci. Med. Sport 2005, 8, 210–221. [CrossRef]
23.
Dawson, B.; Hopkinson, R.; Appleby, B.; Stewart, G.; Roberts, C. Comparison of training activities and game demands in the
Australian Football League. J. Sci. Med. Sport 2004, 7, 292–301. [CrossRef]
24.
Larsson, P. Global positioning system and sport-specific testing. Sports Med. 2003, 33, 1093–1101. [CrossRef]
25.
Pireaux, S.; Defraigne, P.; Wauters, L.; Bergeot, N.; Baire, Q.; Bruyninx, C. Influence of ionospheric perturbations in GPS time and
frequency transfer. Adv. Space Res. 2010, 45, 1101–1112. [CrossRef]
26.
López, A.M.; Granero-Gil, P.; Ortega, J.P.; De Hoyo, M. The validity and reliability of a 5-hz GPS device for quantifying athletes’
sprints and movement demands specific to team sports. J. Hum. Sport Exerc. 2017, 12. [CrossRef]
27.
Coutts, A.J.; Duffield, R. Validity and reliability of GPS devices for measuring movement demands of team sports. J. Sci. Med.
Sport 2010, 13, 133–135. [CrossRef]
28.
Akenhead, R.; French, D.; Thompson, K.G.; Hayes, P.R. The acceleration dependent validity and reliability of 10Hz GPS. J. Sci.
Med. Sport 2014, 17, 562–566. [CrossRef]
29.
Johnston, R.J.; Watsford, M.L.; Kelly, S.J.; Pine, M.J.; Spurrs, R.W. Validity and Interunit Reliability of 10 Hz and 15 Hz GPS Units
for Assessing Athlete Movement Demands. J. Strength Cond. Res. 2014, 28, 1649–1655. [CrossRef]
30.
Buchheit, M.; Al Haddad, H.; Simpson, B.M.; Palazzi, D.; Bourdon, P.C.; Di Salvo, V.; Mendez-Villanueva, A. Monitoring
Accelerations With GPS in Football: Time to Slow Down? Int. J. Sports Physiol. Perform. 2014, 9, 442–445. [CrossRef]
31.
Tessaro, E.; Williams, J.H. Validity and reliability of a 15 Hz GPS device for court-based sports movements. Sport Perform. Sci. Rep.
2018, 1, 29.
32.
Duffield, R.; Reid, M.; Baker, J.; Spratford, W. Accuracy and reliability of GPS devices for measurement of movement patterns in
confined spaces for court-based sports. J. Sci. Med. Sport 2010, 13, 523–525. [CrossRef]
33.
Gray, A.; Jenkins, D.; Andrews, M.H.; Taaffe, D.; Glover, M.L. Validity and reliability of GPS for measuring distance travelled in
field-based team sports. J. Sports Sci. 2010, 28, 1319–1325. [CrossRef]
Sensors 2022, 22, 1888
10 of 10
34.
Williams, M.; Morgan, S. Horizontal positioning error derived from stationary GPS units: A function of time and proximity to
building infrastructure. Int. J. Perform. Anal. Sport 2009, 9, 275–280. [CrossRef]
35.
Puig-Diví, A.; Escalona-Marfil, C.; Padullés-Riu, J.M.; Busquets, A.; Padullés-Chando, X.; Marcos-Ruiz, D. Validity and reliability
of the Kinovea program in obtaining angles and distances using coordinates in 4 perspectives. PLoS ONE 2019, 14, e0216448.
[CrossRef]
36.
Altman, D.G. Practical Statistics for Medical Research; Chapman and Hall: London, UK, 1991; pp. 403–405.
37.
Heng, M.L.; Chua, Y.K.; Pek, H.K.; Krishnasamy, P.; Kong, P.W. A novel method of measuring passive quasi-stiffness in the first
metatarsophalangeal joint. J. Foot Ankle Res. 2016, 9, 41. [CrossRef]
38.
Varley, M.; Fairweather, I.H.; Aughey, R.J. Validity and reliability of GPS for measuring instantaneous velocity during acceleration,
deceleration, and constant motion. J. Sports Sci. 2012, 30, 121–127. [CrossRef] [PubMed]
| Inter-Unit Consistency and Validity of 10-Hz GNSS Units in Straight-Line Sprint Running. | 02-28-2022 | Chahal, Amandeep Kaur,Lim, Jolene Ziyuan,Pan, Jing-Wen,Kong, Pui Wah | eng |
PMC7557486 | Vol.:(0123456789)
1 3
European Journal of Applied Physiology (2020) 120:2507–2515
https://doi.org/10.1007/s00421-020-04474-7
ORIGINAL ARTICLE
Biomechanical and metabolic aspects of backward (and forward)
running on uphill gradients: another clue towards an almost inelastic
rebound
L. Rasica1 · S. Porcelli1,2 · A. E. Minetti3 · G. Pavei3
Received: 23 June 2020 / Accepted: 10 August 2020 / Published online: 25 August 2020
© The Author(s) 2020
Abstract
Purpose On level, the metabolic cost (C) of backward running is higher than forward running probably due to a lower elastic
energy recoil. On positive gradient, the ability to store and release elastic energy is impaired in forward running. We studied
running on level and on gradient to test the hypothesis that the higher metabolic cost and lower efficiency in backward than
forward running was due to the impairment in the elastic energy utilisation.
Methods Eight subjects ran forward and backward on a treadmill on level and on gradient (from 0 to + 25%, with 5% step).
The mechanical work, computed from kinematic data, C and efficiency (the ratio between total mechanical work and C)
were calculated in each condition.
Results Backward running C was higher than forward running at each condition (on average + 35%) and increased linearly
with gradient. Total mechanical work was higher in forward running only at the steepest gradients, thus efficiency was lower
in backward running at each gradient.
Conclusion Efficiency decreased by increasing gradient in both running modalities highlighting the impairment in the elastic
contribution on positive gradient. The lower efficiency values calculated in backward running in all conditions pointed out
that backward running was performed with an almost inelastic rebound; thus, muscles performed most of the mechanical
work with a high metabolic cost. These new backward running C data permit, by applying the recently introduced ‘equiva-
lent slope’ concept for running acceleration, to obtain the predictive equation of metabolic power during level backward
running acceleration.
Keywords Backward acceleration · Efficiency · Mechanical work · Metabolic cost · Metabolic power
Abbreviations
C
Metabolic cost
CBA
Metabolic cost of backward running acceleration
BCoM Body centre of mass
PE
Potential energy of BCoM
KE
Kinetic energy of BCoM
TE
Total energy of BCoM
WEXT
Positive external work
WINT
Positive internal work
WTOT
Total work
WEXT
−
Negative external work
Introduction
Backward running is commonly used in rehabilitation and
as an injury prevention strategy (e.g. Soligard et al. 2008;
Gilchrist et al. 2008; Heiderscheit et al. 2010; Rössler et al.
2016), thanks to the reduced knee joint forces and lower
vertical peak of the ground reaction force compared with
forward running (Flynn and Soutas-Little 1995; Sussman
et al. 2000; Roos et al. 2012). Moreover, the reverse direc-
tion of the movement gives the possibility to involve and
Communicated by Jean-René Lacour.
* G. Pavei
[email protected]
1
Institute of Biomedical Technologies, National Research
Council, Segrate, Italy
2
Department of Molecular Medicine, University of Pavia,
Pavia, Italy
3
Laboratory of Physiomechanics of Locomotion, Department
of Pathophysiology and Transplantation, Physiology
Division, University of Milan, Via Mangiagalli 32,
20133 Milan, Italy
2508
European Journal of Applied Physiology (2020) 120:2507–2515
1 3
train different muscles groups (DeVita and Stribling 1991;
Flynn and Soutas-Little 1995; Sterzing et al. 2016); for a
comprehensive review on backward running see Uthoff et al.
(2018). An increasing number of backward running competi-
tions have also been organised all over the world (also the
RetroRunning world championship), with athletes training
specifically backward for improving their performance.
On level, the metabolic demand of backward running
is higher than forward running (Reilly and Bowen 1984;
Flynn et al. 1994; Wright and Weyand 2001) probably due
to a higher muscle activation (Flynn and Soutas-Little 1993,
1995; Wright and Weyand 2001; Sterzing et al. 2016) and/
or a reduced elastic energy utilisation (Cavagna et al. 2011,
2012). This lower elastic contribution could be caused by
the inverse approach of the foot on the ground that does not
allow to store and recoil the energy from Achilles tendon or
foot arch. Up to now, on level, no studies have analysed the
mechanical work and metabolic cost of backward running
concurrently so that conclusions about efficiency and elastic
energy were inferred only indirectly.
When moving on positive gradient, the energy saving
mechanism of forward running is impaired (Minetti et al.
1994). When running uphill the downward trajectory of the
body centre of mass is reduced and less energy can be stored
in the elastic elements of the lower limbs, which decreases
the overall running efficiency (Minetti et al. 1994). There
are no studies on the metabolic aspects (or efficiency) of
backward running on gradient yet. However, it has been
shown that the difference in metabolic cost between forward
and backward walking was 100% on level, and decreased to
5–8% at gradients steeper than + 15% (Minetti and Ardigò
2001) and this decrement was addressed to the impairment
in the pendulum like motion while walking uphill.
Based on this general knowledge, the analysis of mechan-
ical and metabolic aspects of backward running on gradi-
ent would test the hypothesis of the higher metabolic cost
and the possible decreased efficiency in backward than for-
ward running due to the impairment in the elastic energy
utilisation.
Materials and methods
Subjects
Eight male endurance runners (age: 25.6 ± 3.2 year, height:
1.76 ± 0.07 m, mass: 68.4 ± 6.6 kg, ̇V O2max: 65.7 ± 6.2 mlO2
kg−1 min−1; mean ± SD) took part in the study. Each sub-
ject was fully informed about the aims, methods, and risks
associated with participation and gave his written informed
consent before the start of the study. All procedures were
in accordance with the Declaration of Helsinki and the
study was approved by the local ethics committee. Subjects
undertook three familiarisation sessions with backward run-
ning at all speeds and gradients to get used with balance and
proprioception while moving backward. After familiarisa-
tion, subjects came to the laboratory six times to complete
the entire protocol.
Experimental protocol
Subjects visited the laboratory on six different not-con-
secutive days. This protocol was designed to avoid any
fatigue effect due to the high metabolic and neuromuscu-
lar demand of each acquisition; the comparison between
forward and backward running on the same subject was
performed to avoid any mechanical or metabolic con-
founding factors; a number of speeds were tested to check
the metabolic cost behaviour. On day 1, subjects ran for-
ward on level at 2.78 m s−1, on gradient + 5% at 2.5 m s−1
and + 10% at 2.22 m s−1, with 15 min of recovery among
trials. On day 2, subjects ran forward on gradient + 15% at
1.94 m s−1 and + 20% at 1.67 m s−1, with 15 min of recovery
between trials. On day 3, subjects ran backward on level at
1.67 m s−1, on gradient + 5% at 1.53 m s−1 and + 20% at
1.11 m s−1, with 15 min of recovery among trials. On day
4, subjects ran backward on gradient + 10% at 1.11 m s−1,
1.39 m s−1 and 1.67 m s−1, with 15 min of recovery among
trials. On day 5, subjects ran backward on gradient + 15%
at 1.11 m s−1, 1.25 m s−1, 1.39 m s−1 and 1.67 m s−1, with
15 min of recovery among trials. All acquisitions lasted
5 min. On day 6, kinematics data for all conditions were
recorded (see below). The mechanical parameters (and
efficiency) were compared between backward and forward
running at each slope pairwise at these speeds: 1.67, 1.53,
1.39, 1.25, 1.11, 0.97 m s−1 for backward running and 2.78,
2.50, 2.22, 1.94, 1.67, 1.39 m s−1 for forward running at
0, + 5, + 10, + 15, + 20 and + 25% gradient, respectively.
Metabolic measurements
Each experimental session was preceded by an 8-min
stand resting oxygen consumption ( ̇VO2, mlO2 kg−1 min−1)
assessment after which subjects started running on the
treadmill. Data acquisition lasted 5 min in order to reach
a steady state ̇VO2. Pulmonary ventilation, oxygen con-
sumption and carbon dioxide production were analysed
breath by breath by a metabolic cart (Vmax229, Sensor-
Medics, The Netherlands). The metabolic cost of running
(C, J kg−1 m−1, Margaria et al. 1963) was calculated from
the data collected during the last minute of exercise by
dividing the measured net ̇VO2 (total – resting ̇VO2) by the
running speed. The unit conversion from mlO2 to meta-
bolic J was achieved by considering the mean respiratory
exchange ratio ( ̇VCO2 ̇VO2
−1) for each acquisition. At
rest and during recovery (3rd and 5th minute) 20 μL of
2509
European Journal of Applied Physiology (2020) 120:2507–2515
1 3
capillary blood was obtained from a preheated earlobe for
the determination of blood lactate concentration ([La−]b)
by an enzymatic method (Biosen 5030, EKF, Germany).
Kinematics
Three-dimensional (3D) body motion was collected by
an 8-camera system (6 Vicon MX 1.3, 2 T20-S, Oxford
Metrics, UK), by sampling at 100 Hz the spatial coordi-
nates of 18 reflective markers located on the main joint
centres (Minetti et al. 1993; Pavei et al. 2017), while the
subject was running on a treadmill (Ergo LG Woodway,
Germany). Marker positions were filtered through a ‘zero-
lag’ second-order Butterworth low pass filter with a cutoff
frequency detected by a residual analysis on each marker
coordinate (Winter 1979). Each acquisition lasted 1 min
and the time course of the 3D body centre of mass (BCoM)
position was computed from an 11-segment model
(Minetti et al. 1993; Pavei et al. 2017) based on Dempster
inertial parameters of body segments (Winter 1979). From
the BCoM 3D trajectory, the time course of potential (PE)
and kinetic (KE) energies was computed to obtain the total
mechanical energy (TE = PE + KE). The summation of all
increases in TE time course constitutes the positive exter-
nal work (WEXT, J kg−1 m−1), the work done to accelerate
and lift the BCoM (Cavagna et al. 1963; Cavagna et al.
1976). The work necessary to rotate and accelerate limbs
with respect to BCoM (WINT, J kg−1 m−1) (Cavagna and
Kaneko 1977; Willems et al. 1995) was also calculated
(according to Minetti et al. 1993) and summed to WEXT to
obtain the total mechanical work (WTOT, J kg−1 m−1). The
frictional component of WINT (Minetti et al 2020) was not
included in the present calculation. The negative exter-
nal work (WEXT
−, J kg−1 m−1), the decreases in TE time
course, was analysed as percentage of ‘comprehensive’
external mechanical work (= (WEXT) + (WEXT
−)) in gradi-
ent locomotion, as suggested by Minetti et al. (1994). The
ratio between WTOT and C was used to estimate locomotion
efficiency. Elastic energy contribution was estimated at
each step as the difference between the mechanical equiva-
lent of C and WTOT. C was converted into WTOT by multi-
plying by an efficiency value of the positive work of 0.28
(Woledge et al. (1985) reported a range of 0.25–0.30 for
positive work muscle efficiency), then the measured WTOT
was subtracted from it. The result, multiplied by the pro-
gression speed and divided by step frequency, provides an
estimate of the elastic energy stored in a step. The elastic
energy value of forward running on level was set to 1, and
all the other conditions are reported as (sub)multiples. All
data were analysed with custom-written Labview programs
(release 10, National Instruments, USA).
Statistics
Data were presented as mean ± SD and compared between
running conditions using paired t test; difference among
speeds were compared using one-way ANOVA for repeated
measures and Bonferroni post hoc test; significance level
was set at p < 0.05. Statistical analyses were performed with
SPSS version 20 (IBM).
Results
Metabolic cost
Forward running C increased with slope and present data are
comparable with Minetti et al. (2002) values (Fig. 1). Back-
ward running C was significantly higher than forward run-
ning at each slope (P < 0.01, Fig. 1) and speed independent
at the analysed gradients. Backward running C (J kg−1 m−1)
can be computed as a function of gradient (with same units
as in Fig. 1) with the equation: C = 0.31*gradient + 4.9
(R2 = 0.99). The difference between forward and backward
running was almost constant among gradients 35 ± 7%.
Biomechanical parameters
The mechanical WEXT, WINT, and WTOT of backward run-
ning in all gradient conditions are plotted as a function of
speed in Fig. 2. WEXT was the major determinant of WTOT
and decreased with speed, but increased with gradient.
WINT was almost gradient independent due to the decrease
Fig. 1 Metabolic cost (J kg−1 m−1) as a function of gradient (%).
Black circles represent backward running, and white circles repre-
sent forward running. The superimposed dotted line represents the
Minetti et al. 2002 equation of metabolic cost on gradient and well fit
the experimental data. Backward running cost is always higher than
forward running (*p < 0.01) on average of 35%. Data are mean ± SD
2510
European Journal of Applied Physiology (2020) 120:2507–2515
1 3
of speed. In Fig. 3, the mechanical parameters of back-
ward and forward running are shown at each slope. Data
were collected and presented at these identical gradients
(0, + 5, + 10, + 15, + 20 and + 25%), however, at different
speeds: 1.67, 1.53, 1.39, 1.25, 1.11, 0.97 m s−1 for back-
ward running and 2.78, 2.50, 2.22, 1.94, 1.67, 1.39 m s−1
for forward running. WEXT was greater in backward run-
ning from 0 to 10%, whereas WINT was significantly lower
in backward running at all gradients (p < 0.01) and WTOT
turned to be greater in forward running only at maximal
gradients (20–25%, p < 0.05) (Fig. 3). Stride frequency
(SF, Hz, Fig. 4) was statistically higher in backward than
forward running at all slopes (p < 0.01).
Locomotion efficiency (Fig. 5) was greater in forward
than backward running (p < 0.001) and decreased with gradi-
ent. Backward running reached values close to the muscular
efficiency (0.25–0.30) at the steepest gradient where both
metabolic and mechanical variable were measured.
Estimated elastic energy contribution (Fig. 6) was higher
in forward than backward running in all gradient conditions
(p < 0.001) and decreased with gradient. Backward running
approached no elastic energy contribution at the steepest
gradient.
Discussion
The metabolic cost of backward running was higher than
forward running in all the investigated gradients, whereas
the total mechanical work was similar in the two gaits at all
gradients. Thus, the lower locomotion efficiency of back-
ward than forward running (also on gradient) seems to be
explained by the lower elastic energy contribution that does
Fig. 2 The mechanical external (WEXT), internal (WINT) and total (WTOT) work (J kg−1 m−1) as a function of speed (m s−1) in backward running is
represented at the different investigated gradients. Data are mean ± SD
Fig. 3 The mechanical external (WEXT), internal (WINT) and total
(WTOT) work (J kg−1 m−1) as a function of gradient (%) is represented
in backward (black circles) and forward (white circles) running. Sta-
tistical difference between backward and forward running: #p < 0.05;
*p < 0.01. Data are mean ± SD
2511
European Journal of Applied Physiology (2020) 120:2507–2515
1 3
not assist muscles in performing mechanical work, which is
carried out with a higher metabolic cost.
The metabolic cost of backward running was already
shown to be higher than forward running on level over a
range of speeds (Flynn et al. 1994; Wright and Weyand
2001) and the percentage difference is close to that reported
in the present study. The novelty of this work consists in
extending the previous knowledge also to gradients, where
we found that the difference in metabolic cost was almost
constant between the two running modalities at the different
slopes, with a similar increase among gradients (Fig. 1). This
behaviour differs from walking, since Minetti and Ardigò
(2001) reported a decrease in delta cost between forward
and backward walking on gradient, down to a + 5–8% dif-
ference at gradients steeper than 15%. They ascribed this
decrease in delta cost to the impairment of the pendulum-
like energy-saving mechanism of forward walking (energy
recovery decreased in parallel with the metabolic cost) on
gradient. Running does not rely on this mechanism, there-
fore a direct comparison cannot be performed; we will
discuss later the energy-saving mechanism of running and
its implication on the metabolic cost. The high metabolic
power required for running backward forced us to test dif-
ferent speeds in the two running modalities, and to decrease
speed (in both modalities) by increasing the gradient. The
metabolic cost of forward running is speed independent on
level and on gradient (Margaria 1938; Margaria et al. 1963;
Minetti et al. 2002). Backward running C showed the speed
independency on level (Wright and Weyand 2001), and here
we extended this speed independency also on gradient [in
Fig. 4 Stride frequency (Hz) as a function of gradient (%). Black cir-
cles represent backward running, and white circles represent forward
running. Statistical difference between backward and forward run-
ning: #p < 0.05; *p < 0.01; §p < 0.001. Data are mean ± SD
Fig. 5 Running efficiency, calculated as the ratio between total
mechanical work (WTOT, J kg−1 m−1) and metabolic cost (C, J
kg−1 m−1), as a function of gradient (%) is represented in backward
(black circles) and forward (white circles) running. Statistical differ-
ence between backward and forward running: *p < 0.01; §p < 0.001.
Data are mean ± SD
Fig. 6 Estimated elastic energy contribution is represented as a func-
tion of gradient (%) in backward (black circles) and forward (white
circles) running. The mean elastic energy of forward running on level
is considered as 1 (see Material and methods for details), and all the
other conditions are represented as submultiple. Statistical differ-
ence between backward and forward running: §p < 0.001. Data are
mean ± SD
2512
European Journal of Applied Physiology (2020) 120:2507–2515
1 3
the tested range of speeds (1.11–1.67 m s−1) and gradients
(+ 10%, + 15%)]; thus, this speed difference between run-
ning modalities should not affect our metabolic conclusion.
The mechanical work values of backward running on
level of present investigation showed the same pattern as in
Cavagna et al. (2011) values (Fig. 2), whereas no data have
been previously reported for backward running on gradi-
ent. WEXT decreased with running speed, WINT increased by
increasing speed, but its contribution was small, and then
WTOT decreased in the investigated range of speeds at all
gradients. The mechanical work data for forward running
(Fig. 3) revealed similar trend compared with Minetti et al.
(1994) values up to + 15%, which was the steepest gradient
analysed in that study, whereas data on steeper slopes are
not reported in the literature. At the two steepest gradients
(+ 20 and + 25%), forward running WEXT increased with the
same trend as the previous gradients (Fig. 3). However, WINT
that was gradient independent until + 15% (present data and
Minetti et al. 1994) showed a tendency to increase probably
due to an increased duty factor and more extended limbs that
increased the inertia during the swing (thus the compound
factor q of the predictive equation for WINT (Minetti 1998)
is increased). This WINT tendency to increase at the steep-
est gradients is similar to data reported by Nardello et al.
(2011). A similar behaviour in the increase of WINT and q
factor has been reported at the beginning of the acceleration
phase in sprint running (Pavei et al. 2019) and reinforces the
idea that the mechanics of constant speed uphill running can
be assimilated to running acceleration (di Prampero et al.
2005; Minetti and Pavei 2018). When comparing forward
and backward running, albeit not at the same speed, on the
different slopes the same trend in WEXT and WTOT was found,
with WEXT that increased linearly with gradient (on level
WEXT is higher in backward running, as reported by Cavagna
et al. (2011)) and was the main determinant of WTOT. WINT
was slope independent in backward running, but showed a
tendency to increase in forward running, which caused a
higher WTOT in forward than backward running at the steep-
est gradients. Stride frequency was higher in backward than
forward running at all slopes (Fig. 4). On level, a higher
stride frequency in backward compared with forward run-
ning at paired speed was already reported (Threlkeld et al.
1989; Flynn et al. 1994; Wright and Weyand 2001; Cav-
agna et al. 2011, 2012). Our results on level showed that
speed (1.39–2.22 m s−1 range) was increased with a constant
stride frequency and an increased stride length, similar to the
results of Cavagna et al. (2012). The higher stride frequency
would increase WINT, but we found higher values in forward
than backward running. Other kinematics parameters concur
in the computation of WINT: duty factor, defined as the frac-
tion of foot contact within the stride duration, mean velocity
and a compound q factor that accounts for the limb mass and
spatial configuration during the stride (Minetti 1998). When
analysing the differences of each WINT component between
backward and forward running on gradients, we found the
already mentioned increase in stride frequency (+ 9%), an
increase in duty factor (+ 27%), together with a decrease in
velocity (− 36%) and q (− 32%), which led to a decreased
WINT (− 35%) in backward running.
Running has been classically represented as a bouncing
ball (Cavagna et al. 1964) or a spring mass model (Blick-
han 1989), where the lowering trajectory of the BCoM dur-
ing the first half of the contact time compresses the spring
(or deforms the ball) that can store elastic energy, which is
then released to assist muscles while lifting and accelerat-
ing BCoM for the next step. Thanks to this elastic recoil of
the muscle–tendon structures, running efficiency values are
higher than the muscle efficiency (25–30%) and it is also
termed ‘apparent efficiency’. In the present study, forward
running apparent efficiency on level was ~ 60%, in line with
the literature (Cavagna and Kaneko 1977), and decreased
with increasing gradient, ~ 40% at + 20%, losing most of
the ‘apparent’ part (Fig. 5). This is in accordance with and
expand the results of Minetti et al. (1994). Apparent effi-
ciency of backward running decreased similarly to forward
running, but with about − 20% value in the slope range from
level to + 20% (Fig. 5). These results showed that the energy-
saving mechanism of running (the storage and release of
elastic energy) is impaired on gradient. One explanation
can be found by looking at the trajectory of the BCoM and
the fraction of positive (WEXT) and negative external work
(WEXT
−) (Fig. 7). On level, positive (WEXT) and negative
(WEXT
−) external work equally contributes to the ‘compre-
hensive’ external mechanical work (= (WEXT) + (WEXT
−)).
By moving uphill, WEXT
− reduced its contribution as the
BCoM trajectory became more ascending (as an effect of the
slope) than descending (Minetti et al. 1994). Since the spring
is compressed, and elastic energy is stored, during the lower-
ing part of the trajectory, and this part is smaller by increas-
ing gradient, less elastic energy can be stored. The muscles
had then to perform the positive work to lift the BCoM,
which increases with slopes, with less assistance from ten-
dons; this required more metabolic energy that increased C
(which is the denominator of the efficiency equation) and the
efficiency decreased (Fig. 5). Since muscles are required to
perform more work, a higher sEMG activity can be expected
in backward than forward running; we did not assess sEMG,
but higher activity was found when running backward on
level (Flynn and Soutas-Little 1993, 1995; Sterzing et al.
2016). The partitioning between positive and negative exter-
nal work was similar between the two running modalities
(Fig. 7), highlighting the same behaviour of the BCoM tra-
jectory on gradient. The estimated elastic energy contribu-
tion showed the same decreasing tendency with gradient
of efficiency (Fig. 6), reinforcing the aforementioned idea
that the energy-saving mechanism is impaired. Backward
2513
European Journal of Applied Physiology (2020) 120:2507–2515
1 3
running values were always lower than forward running,
and while at the steepest gradient forward running main-
tained some kind of elastic contributions, backward running
relied only on muscle capability to perform work and power
(Fig. 7). The mechanical inefficacy of backward running was
already described by Cavagna et al. (2011, 2012) with the
reversed landing take-off asymmetry, which resulted in a
greater muscle activation during positive work and a lower
ability to store and release elastic energy. These mechanical
premises for inefficiency were tested here (since Cavagna
et al. did not measure metabolic cost), confirmed in their
original theory (elastic energy) and extended to the gradient,
where we already knew that forward running energy saving
was impaired (Minetti et al. 1994). Backward running with
a reversed use of the lever system of the limbs that already
impaired the efficiency on level showed the same impair-
ment of forward running on gradient. However, starting from
a lower level of ‘apparent efficiency’, at the steepest gradi-
ent backward running reached values of the ‘pure’ muscular
efficiency, very likely with no elastic component.
Backward running is performed also in various sport
activities, e.g. in soccer it has been reported to be as frequent
as high-speed running (Mohr et al. 2003). However, up to
now, backward running bouts are only counted (frequency
of occurrence) and/or considered for their duration. The
‘Equivalent Slope’ concept has been an ingenious idea to
infer the metabolic cost of running acceleration (di Pramp-
ero et al. 2005) from the metabolic cost of the steady-state
uphill running (Minetti et al. 2002). With the present meta-
bolic cost data of backward running on gradient (Fig. 1), we
can calculate the metabolic cost of backward running over a
range of 0–2 m s−2 acceleration (CBA, J kg−1 m−1). However,
since the metabolic cost increased linearly with gradient in
backward running (as occurred in forward running), we can
expect that the proposed equation can be used over a wider
range of accelerations. Rearranging the Minetti and Pavei
(2018) equation for the metabolic cost in forward running
acceleration with present data of backward running C on
gradient, the cost of backward running acceleration can be
computed as:
where ab is the absolute backward acceleration (a positive
value, e.g. + 1.5 m s−2, even if it is performed backward,
because the negative value is usually given to deceleration).
With this new equation, the metabolic power (= instanta-
neous CBA × instantaneous speed) of backward acceleration
can be computed, with the acceleration and speed values
obtained from any GPS system, and added to the metabolic
power for forward running acceleration and deceleration
(Minetti and Pavei 2018) to obtain a more precise estimate
of the metabolic power during different types of sports and
activities.
Conclusions
The metabolic cost of backward running on level and uphill
gradient is higher than for forward running, with a similar
difference between the two running modalities. This higher
cost was not determined by an increased mechanical work;
thus, the locomotion efficiency was lower in backward than
forward running. When analysing the trajectory of the body
centre of mass, the two running modalities showed a similar
impairment in the spring mass model behaviour; however,
backward running relied less on the elastic energy. With less
elastic contribution, the muscles have to perform ‘alone’
the work to lift and accelerate BCoM with a higher meta-
bolic demand. With the metabolic cost of backward running
on gradient, and the concept of equivalent slope, the new
equation for the metabolic cost of backward running accel-
eration was computed. The metabolic power of backward
acceleration can be now calculated and integrated with the
well-known equations for forward running acceleration and
deceleration to obtain a more precise estimate of the meta-
bolic demand of the sport activities.
CBA = (a2
b + 96.2)0.5 × (3.14ab + 4.9)
,
Fig. 7 Negative external work (WEXT
−) as a percentage of ‘compre-
hensive’ external mechanical work (= (WEXT) + (WEXT
−)) is repre-
sented as a function of gradient (%). Black circles represent back-
ward running, White circles represent forward running. Data are
mean ± SD
2514
European Journal of Applied Physiology (2020) 120:2507–2515
1 3
Author contributions GP, SP, and AEM conceived and designed the
study. GP and LR conducted the experiments. GP and LR analysed
the data. GP and AEM interpreted the results of the experiments. GP
wrote the manuscript. All authors read and approved the manuscript.
Funding Open access funding provided by Universitá degli Studi di
Milano within the CRUI-CARE Agreement.
Compliance with ethical standards
Conflict of interest The authors report no conflict of interest.
Ethical approval All procedures were performed in accordance with
the ethical standards of the institutional research committee and with
the 1964 Helsinki Declaration and its later amendments or comparable
ethical standards.
Open Access This article is licensed under a Creative Commons Attri-
bution 4.0 International License, which permits use, sharing, adapta-
tion, distribution and reproduction in any medium or format, as long
as you give appropriate credit to the original author(s) and the source,
provide a link to the Creative Commons licence, and indicate if changes
were made. The images or other third party material in this article are
included in the article’s Creative Commons licence, unless indicated
otherwise in a credit line to the material. If material is not included in
the article’s Creative Commons licence and your intended use is not
permitted by statutory regulation or exceeds the permitted use, you will
need to obtain permission directly from the copyright holder. To view a
copy of this licence, visit http://creat iveco mmons .org/licen ses/by/4.0/.
References
Blickhan R (1989) The spring-mass model for running and hopping.
J Biomech 22(11–12):1217–1227. https ://doi.org/10.1016/0021-
9290(89)90224 -8
Cavagna GA, Kaneko M (1977) Mechanical work and efficiency in
level walking and running. J Physiol 268:467–481
Cavagna GA, Saibene FP, Margaria R (1963) External work in walk-
ing. J Appl Physiol 18:1–9
Cavagna GA, Saibene FP, Margaria R (1964) Mechanical work in
running. J Appl Physiol 19:249–256
Cavagna GA, Thys H, Zamboni A (1976) The sources of external work
in level walking and running. J Physiol 262:639–657
Cavagna GA, Legramandi MA, La Torre A (2011) Running back-
wards: soft landing-hard takeoff, a less efficient rebound. Proc
Biol Sci 278(1704):339–346
Cavagna GA, Legramandi MA, La Torre A (2012) An analysis of the
rebound of the body in backward human running. J Exp Biol
215(Pt 1):75–84
DeVita P, Stribling J (1991) Lower extremity joint kinetics and
energetics during backward running. Med Sci Sports Exerc
23(5):602–610
di Prampero PE, Fusi S, Sepulcri L, Morin JB, Belli A, Antonutto
G (2005) Sprint running: a new energetic approach. J Exp Biol
208:2809–2816
Flynn TW, Soutas-Little RW (1993) Mechanical power and muscle
action during forward and backward running. J Orthop Sports
Phys Ther 17:108–112
Flynn TW, Soutas-Little RW (1995) Patellofemoral joint compres-
sive forces in forward and backward running. J Orthop Sports
Phys Ther 21(5):277–282
Flynn TW, Connerty SM, Smutok MA, Zeballos RJ, Weisman I
(1994) Comparison of cardiopulmonary responses to forward
and backward walking and running. Med Sci Sports Exerc
26:89–94
Gilchrist J, Mandelbaum BR, Melancon H, Ryan GW, Silvers HJ,
Griffin LY, Watanabe DS, Dick RW, Dvorak J (2008) A rand-
omized controlled trial to prevent noncontact anterior cruciate
ligament injury in female collegiate soccer players. Am J Sports
Med 36:1476–1483
Heiderscheit BC, Sherry MA, Silder A, Chumanov ES, Thelen DG
(2010) Hamstring strain injuries: recommendations for diagno-
sis, rehabilitation, and injury prevention. J Orthop Sports Phys
Ther 40(2):67–81
Margaria R (1938) Sulla fisiologia e specialmente sul consumo ener-
getico della marcia e della corsa a varia velocità ed inclinazione
del terreno. Atti Acc Naz Lincei 6:299–368
Margaria R, Cerretelli P, Aghemo P, Sassi G (1963) Energy cost of
running. J Appl Physiol 18:367–370. https ://doi.org/10.1152/
jappl .1963.18.2.367
Minetti AE (1998) A model equation for the prediction of mechanical
internal work of terrestrial locomotion. J Biomech 31:463–468
Minetti AE, Ardigò LP (2001) The transmission efficiency of
backward walking at different gradients. Pflugers Arch
442(4):542–546
Minetti AE, Pavei G (2018) Update and extension of the ’equivalent
slope’ of speed-changing level locomotion in humans: a com-
putational model for shuttle running. J Exp Biol. https ://doi.
org/10.1242/jeb.18230 3
Minetti AE, Ardigò LP, Saibene F (1993) Mechanical determinants
of gradient walking energetics in man. J Physiol 472:725–735
Minetti AE, Ardigò LP, Saibene F (1994) Mechanical determinants
of the minimum energy cost of gradient running in humans. J
Exp Biol 195:211–225
Minetti AE, Moia C, Roi GS, Susta D, Ferretti G (2002) Energy cost
of walking and running at extreme uphill and downhill slopes.
J Appl Physiol 93:1039–1046
Minetti AE, Moorhead PA, Pavei G (2020) Frictional internal work
of damped limbs oscillation in human locomotion. Proc R Soc
B 287:20201410
Mohr M, Krustrup P, Bangsbo J (2003) Match performance of high-
standard soccer players with special reference to development
of fatigue. J Sport Sci 21:519–528
Nardello F, Ardigò LP, Minetti AE (2011) Measured and Predicted
mechanical internal work in human locomotion. Hum Mov Sci
30:90–104
Pavei G, Seminati E, Cazzola D, Minetti AE (2017) On the estima-
tion accuracy of the 3D body center of mass trajectory during
human locomotion: Inverse vs forward dynamics. Front Physiol
8:129. https ://doi.org/10.3389/fphys .2017.00129
Pavei G, Zamparo P, Fujii N, Otsu T, Numazu N, Minetti AE, Monte
A (2019) Comprehensive mechanical power analysis in sprint
running acceleration. Scand J Med Sci Sports 29:1892–1900.
https ://doi.org/10.1111/sms.13520
Reilly T, Bowen T (1984) Exertional cost of changes in directional
modes of running. Percept Mot Skills 58:149–150
Roos PE, Barton N, van Deursen RWM (2012) Patellofemoral joint
compression forces in backward and forward running. J Bio-
mech 45:1656–1660
Rössler R, Donath L, Bizzini M, Faude O (2016) A new injury pre-
vention programme for children’s football–FIFA 11+ Kids–can
improve motor performance: a cluster-randomised controlled
trial. J Sports Sci 34(6):549–556
Soligard T, Myklebust G, Steffen K, Holme I, Silvers H, Bizzini M,
Junge A, Dvorak J, Bahr R, Andersen TE (2008) Comprehen-
sive warm-up programme to prevent injuries in young female
footballers: cluster randomised controlled trial. BMJ 337:a2469
2515
European Journal of Applied Physiology (2020) 120:2507–2515
1 3
Sterzing T, Frommhold C, Rosenbaum D (2016) In-shoe plantar
pressure distribution and lower extremity muscle activity pat-
terns of backward compared to forward running on a treadmill.
Gait Posture 46:135–141
Sussman DH, Alrowayeh H, Walker ML (2000) Patellofemoral joint
compressive forces during backward and forward running at the
same speed. J Musculoskelet Res 4(2):107–118
Threlkeld AJ, Horn TS, Wojtowicz G, Rooney JG, Shapiro R
(1989) Kinematics, ground reaction force, and muscle balance
produced by backward running. J Orthop Sports Phys Ther
11(2):56–63
Uthoff A, Oliver J, Cronin J, Harrison C, Winwood P (2018) A new
direction to athletic performance: understanding the acute
and longitudinal responses to backward running. Sports Med
48:1083–1096
Willems PA, Cavagna GA, Heglund NC (1995) External, internal
and total work in human locomotion. J Exp Biol 198:379–393
Winter DA (1979) Biomechanics of human movement. Wiley, New
York
Woledge RC, Curtin NA, Homsher E (1985) Energetic aspects of
muscle contraction. Monogr Physiol Soc 41:1–357
Wright S, Weyand PG (2001) The application of ground force explains
the energetic cost of running backward and forward. J Exp Biol
204:1805–1815
Publisher’s Note Springer Nature remains neutral with regard to
jurisdictional claims in published maps and institutional affiliations.
| Biomechanical and metabolic aspects of backward (and forward) running on uphill gradients: another clue towards an almost inelastic rebound. | 08-25-2020 | Rasica, L,Porcelli, S,Minetti, A E,Pavei, G | eng |
PMC6466240 | International Journal of
Environmental Research
and Public Health
Article
Celebrating 40 Years of Ironman: How the
Champions Perform
Lucas Pinheiro Barbosa 1,†, Caio Victor Sousa 1,2,†
, Marcelo Magalhães Sales 3
,
Rafael dos Reis Olher 1, Samuel Silva Aguiar 1
, Patrick Anderson Santos 1, Eduard Tiozzo 2,
Herbert Gustavo Simões 1, Pantelis Theodoros Nikolaidis 4
and Beat Knechtle 5,6,*
1
Graduate Program in Physical Education, Catholic University of Brasília, 71966-700 Brasília, Brazil;
[email protected] (L.P.B.); [email protected] (C.V.S.); rfl[email protected] (R.d.R.O.);
[email protected] (S.S.A.); [email protected] (P.A.S.); [email protected] (H.G.S.)
2
Miller School of Medicine, University of Miami, Coral Gables, FL 33124, USA; [email protected]
3
Physical Education Department, Goias State University, Quirinópolis, 75860-000 GO, Brazil;
[email protected]
4
Exercise Physiology Laboratory, 18450 Nikaia, Greece; [email protected]
5
Medbase St. Gallen Am Vadianplatz, 9001 St. Gallen, Switzerland
6
Institute of Primary Care, University of Zurich, 8006 Zurich, Switzerland
*
Correspondence: [email protected]; Tel.: +41-(0)71-226-93-00
†
These authors contributed equally to this work.
Received: 7 February 2019; Accepted: 16 March 2019; Published: 20 March 2019
Abstract: We aimed to determine which discipline had the greater performance improvements in the
history of Ironman triathlon in Hawaii and also which discipline had the greater influence in overall
race time. Data from 1983 to 2018 of the top three women and men of each year who competed in
the Ironman World Championship were included. In addition to exploratory data analyses, linear
regressions between split times and years of achievement were performed. Further, a stepwise
multiple linear regression was applied using total race time as the dependent variable and split
times as the independent variables. Both women and men significantly improved their performances
from 1983 to 2018 in the Ironman World Championship. Swimming had the largest difference in
improvements between men and women (3.0% versus 12.1%, respectively). A negative and significant
decrease in each discipline was identified for both women and men, with cycling being the discipline
with the greatest reduction. The results from the stepwise multiple regression indicated that cycling
was the discipline with the highest influence on overall race time for both sexes. Based on the findings
of this study, cycling seems to be the Ironman triathlon discipline that most improved overall race
times and is also the discipline with the greatest influence on the overall race time of elite men and
women in the Ironman World Championship.
Keywords: triathlon; cycling; running; swimming; endurance
1. Introduction
The Ironman triathlon consists of swimming 2.4 miles (3.8 km), cycling 112 miles (180 km),
and running 26.2 miles (42.2 km) and is considered as one of the most challenging ultra-endurance
events worldwide [1,2]. Although triathlon started in San Diego, California, the history of Ironman
triathlon started in 1978 in Hawaii, with the first Ironman World Championship being held in
Kailua-Kona (Big Island) three years later, also in Hawaii [1–3].
Int. J. Environ. Res. Public Health 2019, 16, 1019; doi:10.3390/ijerph16061019
www.mdpi.com/journal/ijerph
Int. J. Environ. Res. Public Health 2019, 16, 1019
2 of 9
Nowadays, the Ironman events take place all over the world, with amateur and professional athletes
competing in these events to qualify for the World Championship in Kailua-Kona. Ironman Hawaii in
considered as the toughest Ironman race in the world due to the course, the environmental conditions,
and the competitiveness of the event [2,4]. The race itself is one of the most popular triathlon events in
the world, with a growing competitiveness and performance improvement in non-elite triathletes [1,5,6].
In addition, it should be highlighted that the best professional triathletes in the world often achieve new
records in Kailua-Kona [7].
In order to help coaches and athletes with both training plans and race strategy, performance
trends have been analyzed in the past few years in many endurance sports [8–11]. Specifically in
triathlon, relevant studies have been conducted for Olympic distance (1.5 km swim/40 km cycle/10 km
run) [12,13], half-distance (half-Ironman: 1.9 km swim/90 km cycle/21 km run) [13,14], full-distance
(3.8 km swim/180 km cycle/42.195 km run) [14,15], and ultra-triathlons (distance > Ironman) [16,17].
To date, two studies investigated the performance of amateur triathletes [2,5], but none of them
included only elite women and men.
Ofoghi et al. [18] investigated which discipline would have the greater influence on overall
performance in an Olympic triathlon and concluded that running was the most decisive, followed
by swimming and cycling. On the other hand, Sousa et al. [19] analyzed all sub-8-h performances in
full-distance triathlon (i.e., Ironman) and reported that cycling was the discipline with the greatest
influence on the overall result, followed by running and swimming. Additionally, it is noteworthy that
in 2018 the female and male winners of the 2018 World Championship improved the course records,
showing that the fastest Ironman triathletes worldwide can further improve their performances.
However, to the best of our knowledge, the only two studies analyzing the Ironman World
Championship results concerned amateur athletes in the analysis, with one of the studies analyzing
races up to 2007 [2] and the other analyzing races from 2002 to 2015 [5]. Therefore, we aimed to analyze
only elite men and women competing in the Ironman World Championship from 1983 to 2018 in
order to determine (i) which discipline had the greatest performance improvement in the last 35 years;
(ii) which discipline had the greatest influence on overall result; and (iii) whether women were really
closing the gap to men.
2. Methods
2.1. Ethical Approval
All procedures used in the study were approved by the Institutional Review Board of Kanton
St. Gallen, Switzerland, with a waiver of the requirement for informed consent of the participants
given the fact that the study involved the analysis of publicly available data (1 June 2010).
2.2. Data
All data were obtained from a publicly available database (www.ironman.com). All official
overall race and split times from the top three women’s and men’s finishers of the Ironman World
Championship from 1983 to 2018 were included in the analysis. Table 1 presents the descriptive
distribution of women’s races including the Ironman World Championship Race/Split Record (among
top three finishers), whereas Table 2 presents men’s data.
Table 1. Women’s total and split race times in the Ironman World Championship from 1983 to 2018.
Race Time
Median
(25–75 Percentile)
Mean
(±SD)
Ironman World
Championship Race/Split Record *
Overall
09:16:48
(09:03:51–09:26:18)
09:19:06
(00:26:14)
08:26:18
Swimming
00:57:00
(00:55:26–01:00:09)
00:57:34
(00:03:24)
00:48:14
Int. J. Environ. Res. Public Health 2019, 16, 1019
3 of 9
Table 1. Cont.
Race Time
Median
(25–75 Percentile)
Mean
(±SD)
Ironman World
Championship Race/Split Record *
Cycling
05:08:39
(05:00:14–05:17:50)
05:11:22
(00:18:06)
04:26:07
Running
03:08:10
(03:04:09–03:16:31)
03:10:10
(00:10:37)
02:50:26
* Within top three finishers from 1983 to 2018.
Table 2. Men’s total and split race times in the Ironman World Championship from 1983 to 2018.
Race Time
Median
(25–75 Percentile)
Mean
(±SD)
Ironman World
Championship Race/Split Record *
Overall
08:22:02
(08:14:37–08:33:02)
08:26:28
(00:18:26)
07:52:39
Swimming
00:51:43
(00:51:00–00:53:02)
00:53:18
(00:06:07)
00:48:02
Cycling
04:37:47
(04:30:16–04:46:15)
04:42:15
(00:21:01)
04:12:25
Running
03:08:10
(02:46:42–02:57:00)
02:55:21
(± 00:18:57)
02:39:59
* Within top three finishers from 1983 to 2018.
2.3. Statistical Analysis
Initially, an exploratory analysis of the data was carried out, in which central tendency (median
and mean), dispersion (interquartile ranges (25 and 75 percentiles and standard deviation), and extreme
(lowest value) measures were calculated (Tables 1 and 2). Furthermore, all data were transformed in
seconds and non-linear regressions (second order) were performed between each split time and year
of achievement. Linear regressions were used for splits because the non-linear fitting line was the
same as the linear. The relative difference (percentage) between the first (1983) and last (2018) World
Championship’s top three performances was calculated for both women and men. Regarding overall
race time, non-linear regression analyses were performed since the trend line had a better fit than
linear regression. A comparison of average race times between the top three athletes and the chasing
group (4th to 10th place finishers) was performed. Finally, a stepwise multiple linear regression was
performed using overall race time as the dependent variable and split times as independent variables.
The significance level was set as 5% (p < 0.05), and all procedures were performed using SPSS v21.0
(IBM SPSS Statistics for Windows. Armonk, NY: IBM Corp).
3. Results
Men improved in overall race time by 13.3% from 1983 to 2018, whereas women improved by
20.8% (Table 3). Swimming showed the largest difference in improvements between men and women
(3.0% versus 12.1%, respectively), and running showed the smallest difference (12.5% versus 15.5%,
respectively) for the three split disciplines.
Table 3.
Women’s and men’s percentage performance improvements in the Ironman World
Championship from 1983 to 2018.
Total
Total Difference
Decade Average
Decade Average Difference
Overall
Women
20.8%
7.5%
5.20%
1.87%
Men
13.3%
3.33%
Swimming
Women
12.1%
9.1%
3.25%
2.50%
Men
3%
0.75%
Int. J. Environ. Res. Public Health 2019, 16, 1019
4 of 9
Table 3. Cont.
Total
Total Difference
Decade Average
Decade Average Difference
Cycling
Women
26.4%
9.5%
6.60%
2.37%
Men
16.9%
4.23%
Running
Women
15.5%
3%
3.88%
0.75%
Men
12.5%
3.13%
Both women and men significantly improved their performances from 1983 to 2018 in the Ironman
World Championship in Kona, Hawaii (Figures 1 and 2). The world record was improved almost every
three years (see Supplementary Table S1 for accurate race time values from each year’s champions).
Int. J. Environ. Res. Public Health 2018, 15, x FOR PEER REVIEW
4 of 9
almost every three years (see Supplementary Table S1 for accurate race time values from each
year’s champions).
Figure 1. Dispersion and non-linear regression of overall race time performances in the Ironman
World Championship from 1983 to 2018 of women and men. Gold trophies represent the champion
in each year.
Figure 1. Dispersion and non-linear regression of overall race time performances in the Ironman
World Championship from 1983 to 2018 of women and men. Gold trophies represent the champion in
each year.
Int. J. Environ. Res. Public Health 2019, 16, 1019
5 of 9
Figure 1. Dispersion and non-linear regression of overall race time performances in the Ironman
World Championship from 1983 to 2018 of women and men. Gold trophies represent the champion
in each year.
Figure 2. Dispersion and non-linear regression overall race time performances between the top three
finishers and the chasing group (4th to 10th place finishers) from women and men in the Ironman
World Championship from 1983 to 2018.
Figure 2. Dispersion and non-linear regression overall race time performances between the top three
finishers and the chasing group (4th to 10th place finishers) from women and men in the Ironman
World Championship from 1983 to 2018.
The linear regression of split disciplines shows a negative and significant slope for all disciplines
for both women (swimming: −6.94 to 0.47; cycling: −71.06 to −36.98 *; running: −45.79 to −26.86 *;
Figure 3) and men (swimming: −19.37 to −6.77 *; cycling: −96.01 to −60.52 *; running: −65.06 to
−30.58 *; Figure 4) (* indicates p < 0.001). The greatest slope in both sexes was for cycling.
Int. J. Environ. Res. Public Health 2018, 15, x FOR PEER REVIEW
5 of 9
The linear regression of split disciplines shows a negative and significant slope for all
disciplines for both women (swimming: −6.94 to 0.47; cycling: −71.06 to −36.98*; running: −45.79 to
−26.86*; Figure 3) and men (swimming: −19.37 to −6.77*; cycling: −96.01 to −60.52*; running: −65.06 to
−30.58*; Figure 4) (* indicates p < 0.001). The greatest slope in both sexes was for cycling.
Figure 3. Dispersion and linear regression of split-times performances in the Ironman World
Championship from 1983 to 2018 of women.
Figure 3.
Dispersion and linear regression of split-times performances in the Ironman World
Championship from 1983 to 2018 of women.
Int. J. Environ. Res. Public Health 2019, 16, 1019
6 of 9
Figure 3. Dispersion and linear regression of split-times performances in the Ironman World
Championship from 1983 to 2018 of women.
Figure 4. Dispersion and linear regression of split-times performances in the Ironman World
Championship from 1983 to 2018 of men.
Figure 4.
Dispersion and linear regression of split-times performances in the Ironman World
Championship from 1983 to 2018 of men.
The best-fitting model from the stepwise multiple regression included swimming, cycling,
and running split times for both women and men (Table 4). Cycling was the discipline with the
greatest standardized beta for both sexes. The swimming discipline resulted in a negative standardized
coefficient for the men.
Table 4. Standardized coefficient from stepwise multiple regression using total race time as the
dependent variable of Ironman World Championship from 1983 to 2018.
Standardized β Coefficient
R2
R2aj
Swimming
Cycling
Running
Women
0.129
0.690
0.405
0.857
0.856
Men
−0.290
0.895
0.250
0.781
0.775
4. Discussion
The main finding of this manuscript was that cycling has been the Ironman triathlon discipline
with the greatest improvement rate throughout the years and also has had the greatest influence on
overall race time for both women and men. However, apparently both women and men have improved
their performances over the years in all triathlon disciplines. It is worth mentioning that women had a
greater improvement than men in all triathlon disciplines and consequently in total race times.
Jeukendrup and Martin [20] had previously reported that cycling in aero position and the use of
lighter wheels (i.e., elbows on handlebars and carbon wheels, respectively, which had developed for use
in time-trial and triathlon bicycles) makes an athlete significantly faster. Thus, cycling performance also
had new technologies that could influence the performance increase, from the outfit to the bicycle itself,
all of which contributed to make the athlete more comfortable, aerodynamic, and consequently faster.
Although cycling is the discipline that encompasses more time in comparison to swimming
and running in Olympic distance and short distances, it does not have an influence in overall race
time, being the least important of the three disciplines [21]. In Olympic distance and short distances,
athletes normally swim really fast to be able to leave transition one with the first pack of cyclists and
stay within the leading and chasing peloton, thus saving the energy for the running [18]. However,
Int. J. Environ. Res. Public Health 2019, 16, 1019
7 of 9
in Ironman races drafting during cycling is not allowed, making cycling a more competitive discipline,
which means that athletes have to apply some strategy in order to cycle fast enough to remain in a
competitive position but still save energy for the running leg. Similarly, in an analysis only including
top full-distance triathlon performances, the authors reported that cycling was the discipline that most
influenced overall performance in elite men racing below 8 h of overall race time, followed by running
and swimming [19].
A performance analysis on Ironman races investigated more than 340,000 triathletes racing in
253 different race locations and concluded that the race tactics in an Ironman triathlon should focus on
saving energy during the first two disciplines for the running split [22]. This conclusion is different
from the findings of the present study, which suggest that athletes seem to apply greater effort in cycling
than during running. It is worth mentioning that this analysis was carried out with a majority of age
groupers (non-elite), whereas the present study only considered the top three elite professionals from
each year. It is noteworthy that 4th to 10th place finishers in the Ironman World Championship seemed
to have a substantial performance improvement in the last decade of the event, with consistently much
closer groupings in the top ten athletes for both men and women.
With regard to performance throughout the race, the performance analysis of the Ironman
World Championship with amateurs reported a performance increase in all disciplines for men and
women [23]. However, the authors suggest that this improvement in performance may be due to
an increased number of athletes and morphological changes [23]. The overall performance increase
throughout the years can be mostly attributed to the development of new nutrition and training
strategies [24–27]. A controversial result was the negative coefficient for the swim split in men, which
would mean that a slower swim could lead to a better overall race time. We believe that this statistical
outcome is due to the specificity of the sample, as only the top three athletes in the overall race were
considered, and these athletes are not always the best swimmers. For example, in the 2018 World
Championship, none of the top three overall athletes were among the top 10 swimmers.
Concerning the performance gap between men and women, it has markedly reduced in the last
decades. At the 2018 Ironman World Championship, women improved by 21% while men improved
by 13%; the absolute gap between them reduced from 1 h and 38 min to 33 min. Indeed, in the 2018
Championship, the female champion Daniela Ryf crossed the finish line ahead of 20 elite professional
men who finished the race. Some previous studies have concluded that women have been closing the
gap in swimming [28,29], in running [30,31], and even in triathlon [2]. We believe that women may
still close the gap in an Ironman someday despite the body composition and physiological differences
that exist between men and women. One of the possible explanations for this can be attributed to
cultural changes that have favored a greater participation of women in all sports, including triathlon,
thus increasing competitiveness and therefore performance [7,9,10,32].
When comparing performance with other endurance modalities such as ultra-triathlon and
marathon running, the performance gaps between women and men are getting smaller each year.
Knechtle et al. [33] investigated the performance trends of Double Iron ultra-triathlon (2I; 2x Ironman
distance), Triple Iron ultra-triathlon (3I; 3x Ironman distance), and Deca Iron ultra-triathlon (10I;
10x Ironman distance) from 1985 to 2009 and reported a smaller sex difference in 2I and 3I. Conversely,
Nikolaidis et al. [31] investigated the performance of male and female athletes running the marathon
and concluded that men are still faster than women, but the performance gap remained unchanged for
the past few years.
Regarding the specific Ironman World Championship, Kailua-Kona is one of the toughest races
within the entire Ironman circuit, which typically requires athletes to swim in choppy waters, cycle with
a lot of wind, and run in hot and sunny weather [34,35]. The race course has not always been the
same in Ironman Hawaii, with small changes every two or three years in order to accommodate safety
precautions and/or local transit logistics. Although this may affect the overall race time, the distances
remained standard and we believe that any small course changes affecting a specific split race time are
diluted within the sample and do not represent a great confounder to the general results of this study.
Int. J. Environ. Res. Public Health 2019, 16, 1019
8 of 9
5. Conclusions
In conclusion, cycling seems to be the triathlon discipline that most improved overall race times
and is also the discipline that had the greatest influence on the overall race time in elite men and
women in the Ironman World Championship. Furthermore, within the last 40 years of Ironman Hawaii,
both men and women improved overall time performance, but women improved more, thereby closing
the gap to men.
Supplementary Materials: The following are available online at http://www.mdpi.com/1660-4601/16/6/1019/
s1, Table S1: Overall times in the Ironman World Championship from 1983 to 2018 for women and men.
Author Contributions: Conceptualization: C.V.S., L.P.B., P.T.N., and B.K.; methodology: C.V.S., L.P.B., M.M.S.,
R.d.R.O., S.S.A., P.A.S., P.T.N., and B.K.; formal analysis: C.V.S., L.P.B., and M.M.S.; writing—original draft
preparation: C.V.S. and L.P.B.; writing—review and editing: C.V.S., L.P.B., M.M.S., P.A.S., E.T., H.G.S., P.T.N.,
and B.K.; visualization: C.V.S., L.P.B., M.M.S., P.A.S., E.T., H.G.S., P.T.N., and B.K.; supervision: P.T.N. and B.K.;
project administration: P.T.N., and B.K.
Funding: This research received no external funding.
Conflicts of Interest: The authors report no conflicts of interest in this work.
References
1.
Stiefel, M.; Rüst, C.A.; Rosemann, T.; Knechtle, B. A comparison of participation and performance in
age-group finishers competing in and qualifying for Ironman Hawaii. Int. J. Gen. Med. 2013, 6, 67.
2.
Lepers, R. Analysis of Hawaii ironman performances in elite triathletes from 1981 to 2007. Med. Sci.
Sports Exerc. 2008, 40, 1828–1834. [CrossRef] [PubMed]
3.
Corporation, W.T. The Ironman Story. Available online: http://www.ironman.com/triathlon/history.aspx#
axzz2BrLoGbOE (accessed on 22 October 2018).
4.
Sparks, S.; Cable, N.; Doran, D.; Maclaren, D. The influence of environmental temperature on duathlon
performance. Ergonomics 2005, 48, 1558–1567. [CrossRef] [PubMed]
5.
Käch, I.W.; Rüst, C.A.; Nikolaidis, P.T.; Rosemann, T.; Knechtle, B. The age-related performance decline
in Ironman triathlon starts earlier in swimming than in cycling and running. J. Strength Cond. Res. 2018,
32, 379–395. [CrossRef] [PubMed]
6.
Stiefel, M.; Knechtle, B.; Lepers, R. Master triathletes have not reached limits in their Ironman triathlon
performance. Scand. J. Med. Sci. Sports 2014, 24, 89–97. [CrossRef]
7.
Knechtle, B.; Zingg, M.A.; Rosemann, T.; Rüst, C.A. Sex difference in top performers from Ironman to Double
Deca Iron ultra-triathlon. Open Access J. Sports Med. 2014, 5, 159. [CrossRef]
8.
Knechtle, B.; Knechtle, P.; Rüst, C.A.; Rosemann, T. A comparison of anthropometric and training
characteristics of Ironman triathletes and Triple Iron ultra-triathletes. J. Sports Sci. 2011, 29, 1373–1380.
[CrossRef]
9.
Etter, F.; Knechtle, B.; Bukowski, A.; Rüst, C.A.; Rosemann, T.; Lepers, R. Age and gender interactions in
short distance triathlon performance. J. Sports Sci. 2013, 31, 996–1006. [CrossRef]
10.
Le Meur, Y.; Hausswirth, C.; Dorel, S.; Bignet, F.; Brisswalter, J.; Bernard, T. Influence of gender on pacing
adopted by elite triathletes during a competition. Eur. J. Appl. Physiol. 2009, 106, 535–545. [CrossRef]
11.
Esteve-lanao, J.; San Juan, A.F.; Earnest, C.P.; Foster, C.; Lucia, A. How do endurance runners actually train?
Relationship with competition performance. Med. Sci. Sports Exerc. 2005, 37, 496–504. [CrossRef]
12.
Wonerow, M.; Rüst, C.A.; Nikolaidis, P.T.; Rosemann, T.; Knechtle, B. Performance Trends in Age Group
Triathletes in the Olympic Distance Triathlon at the World Championships 2009-2014. Chin. J. Physiol. 2017,
60, 137–150. [CrossRef]
13.
Wu, S.S.X.; Peiffer, J.J.; Brisswalter, J.; Nosaka, K.; Lau, W.Y.; Abbiss, C.R. Pacing strategies during the
swim, cycle and run disciplines of sprint, Olympic and half-Ironman triathlons. Eur. J. Appl. Physiol. 2015,
115, 1147–1154. [CrossRef]
14.
Knechtle, R.; Rüst, C.A.; Rosemann, T.; Knechtle, B. The best triathletes are older in longer race distances–a
comparison between Olympic, Half-Ironman and Ironman distance triathlon. Springerplus 2014, 3, 538.
[CrossRef] [PubMed]
Int. J. Environ. Res. Public Health 2019, 16, 1019
9 of 9
15.
Dallam, G.M.; Jonas, S.; Miller, T.K. Medical considerations in triathlon competition. Sports Med. 2005,
35, 143–161. [CrossRef]
16.
Zaryski, C.; Smith, D.J. Training principles and issues for ultra-endurance athletes. Curr. Sports Med. Rep.
2005, 4, 165–170. [CrossRef] [PubMed]
17.
Abbiss, C.R.; Quod, M.J.; Martin, D.T.; Netto, K.J.; Nosaka, K.; Lee, H.; Surriano, R.; Bishop, D.; Laursen, P.B.
Dynamic pacing strategies during the cycle phase of an Ironman triathlon. Med. Sci. Sports Exerc. 2006,
38, 726–734. [CrossRef]
18.
Ofoghi, B.; Zeleznikow, J.; Macmahon, C.; Rehula, J.; Dwyer, D.B. Performance analysis and prediction in
triathlon. J. Sports Sci. 2016, 34, 607–612. [CrossRef] [PubMed]
19.
Sousa, C.V.; Barbosa, L.P.; Sales, M.M.; Santos, P.A.; Tiozzo, E.; Simões, H.G.; Nikolaidis, P.T.; Knechtle, B.
Cycling as the Best Sub-8-Hour Performance Predictor in Full Distance Triathlon. Sports 2019, 7, 24.
[CrossRef]
20.
Jeukendrup, A.E.; Martin, J. Improving cycling performance. Sports Med. 2001, 31, 559–569. [CrossRef]
[PubMed]
21.
Figueiredo, P.; Marques, E.A.; Lepers, R. Changes in contributions of swimming, cycling, and running
performances on overall triathlon performance over a 26-year period.
J. Strength Cond.
Res.
2016,
30, 2406–2415. [CrossRef] [PubMed]
22.
Knechtle, B.; Käch, I.; Rosemann, T.; Nikolaidis, P.T. The effect of sex, age and performance level on pacing
of Ironman triathletes. Res. Sports Med. 2019, 27, 99–111. [CrossRef] [PubMed]
23.
Lepers, R.; Rust, C.A.; Stapley, P.J.; Knechtle, B. Relative improvements in endurance performance with age:
Evidence from 25 years of Hawaii Ironman racing. Age 2013, 35, 953–962. [CrossRef] [PubMed]
24.
Issurin, V.B. Biological Background of Block Periodized Endurance Training: A Review. Sports Med. 2018.
[CrossRef]
25.
McCormick, A.; Meijen, C.; Marcora, S. Psychological Determinants of Whole-Body Endurance Performance.
Sports Med. 2015, 45, 997–1015. [CrossRef] [PubMed]
26.
Doma, K.; Deakin, G.B.; Bentley, D.J. Implications of Impaired Endurance Performance following Single
Bouts of Resistance Training: An Alternate Concurrent Training Perspective. Sports Med. 2017, 47, 2187–2200.
[CrossRef]
27.
Burke, L.M.; Hawley, J.A.; Jeukendrup, A.; Morton, J.P.; Stellingwerff, T.; Maughan, R.J. Toward a Common
Understanding of Diet-Exercise Strategies to Manipulate Fuel Availability for Training and Competition
Preparation in Endurance Sport. Int. J. Sport Nutr. Exerc. Metab. 2018, 28, 451–463. [CrossRef] [PubMed]
28.
Nikolaidis, P.T.; Di Gangi, S.; de Sousa, C.V.; Valeri, F.; Rosemann, T.; Knechtle, B. Sex difference in open-water
swimming-The Triple Crown of Open Water Swimming 1875–2017. PLoS ONE 2018, 13, e0202003. [CrossRef]
[PubMed]
29.
Nikolaidis, P.T.; de Sousa, C.V.; Knechtle, B. Sex difference in long-distance open-water swimming races -
does nationality play a role? Res. Sports Med. 2018, 26, 332–344. [CrossRef] [PubMed]
30.
Hoffman, M.D.; Wegelin, J.A. The Western States 100-mile endurance run: Participation and performance
trends. Med. Sci. Sports Exerc. 2009, 41, 2191. [CrossRef]
31.
Nikolaidis, P.T.; Rosemann, T.; Knechtle, B. Sex Differences in the Age of Peak Marathon Race Time.
Chin. J. Physiol. 2018, 61, 85–91. [CrossRef]
32.
Tarnopolsky, M.A. Gender differences in substrate metabolism during endurance exercise.
Can.
J.
Appl. Physiol. 2000, 25, 312–327. [CrossRef]
33.
Knechtle, B.; Knechtle, P.; Lepers, R. Participation and performance trends in ultra-triathlons from 1985 to
2009. Scand. J. Med. Sci. Sports 2011, 21, e82–e90. [CrossRef]
34.
Sousa, C.; Aguiar, S.; Olher, R.; Sales, M.; Moraes, M.; Nikolaidis, P.; Rosemann, T.; Knechtle, B.; Simões, H.
Hydration status after an Ironman triathlon: A meta-analysis. J. Hum. Kinet. 2018. [CrossRef]
35.
Laird, R.H.; Johnson, D. The medical perspective of the Kona Ironman Triathlon. Sports Med. Arthrosc. Rev.
2012, 20, 239. [CrossRef]
© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access
article distributed under the terms and conditions of the Creative Commons Attribution
(CC BY) license (http://creativecommons.org/licenses/by/4.0/).
| Celebrating 40 Years of Ironman: How the Champions Perform. | 03-20-2019 | Barbosa, Lucas Pinheiro,Sousa, Caio Victor,Sales, Marcelo Magalhães,Olher, Rafael Dos Reis,Aguiar, Samuel Silva,Santos, Patrick Anderson,Tiozzo, Eduard,Simões, Herbert Gustavo,Nikolaidis, Pantelis Theodoros,Knechtle, Beat | eng |
PMC6239296 | S1 Appendix. Solution of the integral equation for
Pmax(T).
The maximal power Pmax(T) is determined by the integral equation
Pmax(T) + Psup(T) = 1
T
Z T
0
Pmax(T − t)dt = 1
T
Z T
0
Pmax(t)dt
(1)
with Psup(T) given by Eq. (3). This equation can be easily transformed into a
differential equation by defining the indefinite integral E(T) of Pmax(T) so that the
derivative E′(T) = Pmax(T). Without loss of generality, we can chose the initial
condition E(0) = 0. The differential equation for E(T) is then
E′(T) + Psup(T) = E(T)
T
(2)
which has the general solution
E(T) = TPm + TPsup(tc) − T
Z T
tc
Psup(t)
t
dt
(3)
where we imposed the initial condition E′(T = tc) = Pm so that Pmax(T = tc) = Pm as
required by definition of Pm. Performing the integral with the constant function
Psup(t) = Ps for T ≤ tc yields
E(T) = T [Pm + Ps − Ps log(T/tc)]
(4)
and using Psup(t) = Pl(t − tc)/t + Pstc/t for T ≥ tc yields
E(T) = T
Pm + Ps + (Pl − Ps)T − tc
T
− Pl log(T/tc)
.
(5)
Taking the derivative of this solution, we finally obtain the solution
Pmax(T) = Pm − Ps log(T/tc)
(6)
for T ≤ tc and
Pmax(T) = Pm − Pl log(T/tc)
(7)
for T ≥ tc. This is the result given in Eq. (6).
PLOS
1/??
| A minimal power model for human running performance. | 11-16-2018 | Mulligan, Matthew,Adam, Guillaume,Emig, Thorsten | eng |
PMC4919094 | RESEARCH ARTICLE
Prediction and Quantification of Individual
Athletic Performance of Runners
Duncan A. J. Blythe1,2☯¤a*, Franz J. Király3☯¤b*
1 African Institute for Mathematical Sciences, Bagamoyo, Tanzania, 2 Bernstein Centre for Computational
Neuroscience, Berlin, Germany, 3 Department of Statistical Science, University College London, London,
United Kingdom
☯ These authors contributed equally to this work.
¤a Current address: African Institute of Mathematical Sciences, Tanzania, P.O. Box 176, Alpha Zulu,
Chunguuni Street (off Indian Street), Mwambao—Bagamoyo, Pwani—Tanzania
¤b Current address: Department of Statistical Science, University College London, Gower Street, London
WC1E 6BT, United Kingdom
* [email protected] (DAJB); [email protected] (FJK)
Abstract
We present a novel, quantitative view on the human athletic performance of individual run-
ners. We obtain a predictor for running performance, a parsimonious model and a training
state summary consisting of three numbers by application of modern validation techniques
and recent advances in machine learning to the thepowerof10 database of British runners’
performances (164,746 individuals, 1,417,432 performances). Our predictor achieves an
average prediction error (out-of-sample) of e.g. 3.6 min on elite Marathon performances and
0.3 seconds on 100 metres performances, and a lower error than the state-of-the-art in per-
formance prediction (30% improvement, RMSE) over a range of distances. We are also the
first to report on a systematic comparison of predictors for running performance. Our model
has three parameters per runner, and three components which are the same for all runners.
The first component of the model corresponds to a power law with exponent dependent on
the runner which achieves a better goodness-of-fit than known power laws in the study of
running. Many documented phenomena in quantitative sports science, such as the form of
scoring tables, the success of existing prediction methods including Riegel’s formula, the
Purdy points scheme, the power law for world records performances and the broken power
law for world record speeds may be explained on the basis of our findings in a unified way.
We provide strong evidence that the three parameters per runner are related to physiologi-
cal and behavioural parameters, such as training state, event specialization and age, which
allows us to derive novel physiological hypotheses relating to athletic performance. We con-
jecture on this basis that our findings will be vital in exercise physiology, race planning, the
study of aging and training regime design.
PLOS ONE | DOI:10.1371/journal.pone.0157257
June 23, 2016
1 / 16
a11111
OPEN ACCESS
Citation: Blythe DAJ, Király FJ (2016) Prediction and
Quantification of Individual Athletic Performance of
Runners. PLoS ONE 11(6): e0157257. doi:10.1371/
journal.pone.0157257
Editor: Nir Eynon, Victoria University, AUSTRALIA
Received: February 26, 2016
Accepted: May 26, 2016
Published: June 23, 2016
Copyright: © 2016 Blythe, Király. This is an open
access article distributed under the terms of the
Creative Commons Attribution License, which permits
unrestricted use, distribution, and reproduction in any
medium, provided the original author and source are
credited.
Data Availability Statement: Data are available from
https://figshare.com/articles/thepowerof10/3408202
and https://figshare.com/articles/Ful_code_to_
Prediction_and_Quantification_of_Individual_
Athletic_Performance_of_Runners_/3408250.
Funding: DAJB was supported by a grant from the
German Research Foundation, research training
group GRK 1589/1 “Sensory Computation in Neural
Systems.” FJK was partially supported by
Mathematisches Forschungsinstitut Oberwolfach
(MFO). This research was partially carried out at
MFO with the support of FJK’s Oberwolfach Leibniz
Fellowship.
Introduction
Performance prediction and modeling are cornerstones of sports medicine, essential in training
and assessment of athletes with implications beyond sport, for example in the understanding
of aging, muscle physiology, and the study of the cardiovascular system. Existing research on
running performance focuses either on (A) explaining world records [1–6], (B) equivalent
scoring [7, 8], or (C) modelling of individual physiology [9–16]. Currently, however, there is
no parsimonious model which simultaneously explains individual physiology (C) and collec-
tive performance (A,B) of runners.
We present such a model, a non-linear low-rank model derived from a database of UK run-
ners. It levers an individual power law which explains the power laws known to apply to world
records, and which allows us to derive runner-individual training parameters from prior per-
formance data. Performance predictions obtained using our approach are the most accurate to
date, with an average prediction error of under 4 minutes (2% rel.MAE and 3% rel.RMSE out-
of-sample) for elite performances. We anticipate that our framework will allow researchers to
leverage existing insights in the study of world record performances and sports medicine for an
improved understanding of human physiology.
Our work builds on the three major research strands in prediction and modeling of running
performance, which we briefly summarize:
(A) Power law models of performance posit a power law dependence t = c sα between the
time elapsed running t and the distance s, for constants c and α. This is equivalent to assuming
a linear dependence log t = α log s + log c of log-time on log-distance. Power law models have
been known to describe world record performances across sports for over a century [17], and
have been applied extensively to running performance [1–6]. These power laws have been
applied to prediction by practitioners: the Riegel formula [18] predicts performance by fitting c
to each runner and fixing α = 1.06 (derived from world-record performances). The power law
approach has the benefit of modelling performances in a scientifically parsimonious way.
(B) Scoring tables, such as those of the international association of athletics federations
(IAAF), render performances over disparate distances comparable by presenting them on a
single scale. These tables have been published by sports associations for almost a century [19]
and catalogue, rather than model, performances of equivalent standard. Performance predic-
tions may be obtained from scoring tables by forecasting a time with the same score as an exist-
ing attempt, as implemented in the Purdy Points scheme [7, 8]. The scoring table approach has
the benefit of describing performances in an empirically accurate way.
(C) Explicit modeling of performance related physiology is an active subfield of sports sci-
ence. Several physiological parameters are known to be related to athletic performance; these
include maximal oxygen uptake ( _VO2-max) and critical speed (speed at _VO2-max) [9, 10],
blood lactate concentration, and the anaerobic threshold [11, 20]. Physiological parameters
may be used (C.i) to make direct predictions when clinical measurements are available [12, 13,
21], or (C.ii) to obtain theoretical models describing physiological processes [14–16, 22]. These
approaches have the benefit of explaining performances physiologically.
All three approaches (A), (B), (C) have appealing properties, as explained above, but none
provides a complete treatment of running performance prediction: (A) individual perfor-
mances do not follow the parsimonious power law perfectly; (B) the empirically accurate scor-
ing tables do not provide a simple interpretable relationship. Neither (A) nor (B) can deal with
the fact that runners may differ from one another in multiple ways. The clinical measurements
in (C.i) are informative but usually available only for a few select runners, typically at most a
few dozen (as opposed to the 164,746 considered in our study). The interpretable models in (C.
ii) are usually designed not with the aim of predicting performance but to explain physiology
Prediction and Quantification of Individual Athletic Performance of Runners
PLOS ONE | DOI:10.1371/journal.pone.0157257
June 23, 2016
2 / 16
Competing Interests: The authors have declared
that no competing interests exist.
or to estimate physiological parameters from performances; thus these methods are not directly
applicable to running performance prediction without additional work.
The approach we present unifies the desirable properties of (A), (B) and (C), while avoiding
the aforementioned shortcomings. We obtain (A) a parsimonious model for individual athletic
performance that is (B) empirically derived from a large database of UK runners. It yields the
best performance predictions to date (2% average error for elite runners on all events, average
error 3.6 min for Marathon 0.3 seconds for 100 metres) and (C) unveils hidden descriptors for
individuals which we find to be related to physiological and behavioural characteristics.
Our approach bases predictions on Local Matrix Completion (LMC), a machine learning
technique which posits the existence of a small number of explanatory variables which describe
the performance of individual runners. Application of LMC to a database of runners allows us,
in a second step, to derive a parsimonious physiological model describing the running perfor-
mance of individual runners. We discover that a three number-summary for each individual
explains performance over the full range of distances from 100m to the Marathon. The three-
number-summary relates to: (1) the endurance of a runner, (2) the relative balance between
speed and endurance, and (3) specialization over middle distances. The first number explains
most of the individual differences over distances greater than 800m, and may be interpreted as
the exponent of an individual power law for each runner, which holds with remarkably high
precision, on average. The other two numbers describe individual, non-linear corrections to
this individual power law. Vitally, we show that the individual power law with its non-linear
corrections reflects the data more accurately than the power law for world records. We antici-
pate that the individual power law and three-number summary will allow for exact quantitative
assessment in the science of running and related sports.
Materials and Methods
Local Matrix Completion and the Low-Rank Model
It is well known that world records over distinct distances are held by distinct runners—no one
single runner holds all running world records. Since world record data obey an approximate
power law (see above), this implies that the individual performance of each runner deviates
from this power law. The left top panel of Fig 1 displays world records and the corresponding
individual performances of world record holders in logarithmic coordinates—an exact power
law would follow a straight line. The world records align closely to a straight line, while individ-
uals deviate non-linearly. Notable is also the kink in the world records which causes them to
deviate from an exact straight line, yielding a “broken power law” for world records [5].
Any model for individual performances must model this individual, non-linear variation,
and will, optimally, explain the broken power law observed for world records as an epiphenom-
enon of such variation over individuals. In the following paragraphs we explain how the LMC
scheme captures individual variation in a typical scenario.
Consider three runners (taken from the database) as shown in the top-right panel of Fig 1.
The 1500m performance of the green runner is not known and is to be predicted. All three run-
ners, green, blue and red, achieve similar performance over 800m. Any classical method for
performance prediction which only takes this information into account will predict that green
performs similarly over 1500m to blue and red. However, this is unrealistic, since it does not
take into account event specialization: looking at the 400m performance, we see that red is
slowest over short distances, followed by blue and then by green. Thus red is more of an endur-
ance type runner than blue, and blue is more of a speed type runner (sprinter) than red; green
specializes to a greater extent in speed than both red and blue. Using this additional informa-
tion leads to the more realistic prediction that green will be out-performed by red and blue
Prediction and Quantification of Individual Athletic Performance of Runners
PLOS ONE | DOI:10.1371/journal.pone.0157257
June 23, 2016
3 / 16
over 1500m. Supplementary analysis (IV) in S1 Supplement validates that this phenomenon
illustrated in the example is prevalent throughout the data set.
LMC is a quantitative method for taking into account this event specialization. A schematic
overview of the simplest variant is displayed in the bottom panel of Fig 1: to predict an event
for a runner (figure: 1500m for green) we find a 3-by-3-pattern of performances, denoted by A,
with exactly one missing entry—this means the two other runners (figure: red and blue) have
attempted similar events and their data are available. Explanation of the green runner’s curve
by the red and the blue is mathematically modelled by demanding that the data of the green
runner is given as a weighted sum of the data of the red and the blue; i.e., more mathematically,
the green row is a linear combination of the blue and the red row. A classical result in matrix
algebra implies that the green row is a linear combination of red and blue whenever the deter-
minant of A, a polynomial function in the entries of A, vanishes; i.e., det(A) = 0.
A prediction is made by solving the Eq det(A) = 0 for “?”. To increase accuracy, candidate
solutions from multiple 3-by-3-patterns (obtained from many triples of runners) are averaged
in a way that minimizes the expected error in approximation. We shall consider variants of the
algorithm which use n-by-n-patterns, n corresponding to the complexity of the model (we later
show n = 4 to be optimal). See the methods appendix for an exact description of the algorithm
used.
Fig 1. Central phenomenon: non-linear deviation from the power law in individuals. Top left:
performances of world record holders and a selection of random runners. Curves labelled by runners are their
known best performances (y-axis) at that event (x-axis). Black crosses are world record performances.
Individual performances deviate non-linearly from the world record power law. Top right: a good model should
take into account specialization, illustration by example. Hypothetical performance curves of three runners,
green, red and blue are shown, the task is to predict green on 1500m from all other performances. Dotted
green lines are predictions. State-of-art methods such as Riegel or Purdy predict green performance on
1500m close to blue and red; a realistic predictor for 1500m performance of green—such as LMC—will
predict that green is outperformed by red and blue on 1500m; since blue and red being worse on 400m
indicates that out of the three runners, green specializes most on shorter distances. Bottom: using local
matrix completion as a mathematical prediction principle by filling in an entry in a (3 × 3) sub-pattern.
Schematic illustration of the algorithm.
doi:10.1371/journal.pone.0157257.g001
Prediction and Quantification of Individual Athletic Performance of Runners
PLOS ONE | DOI:10.1371/journal.pone.0157257
June 23, 2016
4 / 16
The LMC prediction scheme is an instance of the more general local low-rank matrix com-
pletion framework introduced in [23], here applied to performances in the form of a numerical
table (or matrix) with columns corresponding to events and rows to runners. The cited frame-
work is the first matrix completion algorithm which allows prediction of single missing entries
as opposed to all entries. While matrix completion has proved vital in predicting consumer
behaviour and recommender systems, the results in the present study show that existing
approaches which predict all entries at once cannot cope with the non-uniform missingness
and the noise associated with performance prediction in the same way as LMC can (see find-
ings and supplemental analysis II.a in S1 Supplement). See the methods appendix for more
details of the method and an exact description.
In a second step, we use the LMC scheme to fill in all missing performances (over all events
considered—100m, 200m etc.) and obtain a parsimonious low-rank model—we remark that
first filling in the entries with LMC and only then fitting the model is crucial due to the fact
that data are non-uniformly missing (see supplemental analysis II.a in S1 Supplement). The
low-rank model explains individual running times t in terms of distance s and has the form:
log t ¼ l1f1ðsÞ þ l2f2ðsÞ þ þ lrfrðsÞ;
ð1Þ
with components f1, f2, . . ., fr that are the same for every runner, and coefficients λ1, λ2, . . ., λr
which summarize the runner under consideration. The number of components and coefficients
r is known as the rank of the model and measures its complexity. The Riegel power law is a
very special case, demanding that log t = 1.06log s + c; that is, a rank 2 model with λ1 = 1.06 for
every runner, f1(s) = log s, and a runner-specific constant λ2 f2(s) = c. Our analyses will show
that the best model has rank r = 3 (meaning above we consider patterns or matrices of size n ×
n = 4 since above n = r + 1). This means that the model has r = three universal components
f1(s), f2(s), f3(s), and every runner is described by their individual three-coefficient-summary
λ1, λ2, λ3. Remarkably, we find that f1(s) = log s (for a suitable unit of distance/time, see supple-
mental analysis II.b in S1 Supplement), yielding an individual power law; the corresponding
coefficient λ1 thus has the natural interpretation as an individual power law exponent.
Table 1 contains the exact form for the components f1, f2, f3 in our model; they are also dis-
played in Fig 2 top left. More details on how to obtain components and coefficients can be
found in the methods section, “obtaining the low-rank components and coefficients”, and in
supplementary analysis II.b in S1 Supplement.
Data Set, Analyses and Model Validation
The basis for our analyses is the online database www.thepowerof10.info, which catalogues
British individuals’ performances achieved in officially ratified athletics competitions since
1954. The excerpt we consider contains performances between 1954 and August 3, 2013. Our
study does not use performances prior to 1954 since the database does not contain perfor-
mances dating prior to 1954. It contains (after error removal) records of 164,746 individuals of
both genders, ranging from the amateur to the elite, young to old, comprising a total of
1,417,432 individual performances over 10 different distances: 100m, 200m, 400m, 800m,
1500m, the Mile, 5km, 10km, Half-Marathon, Marathon. All British records over the distances
considered are contained in the dataset; the 95th percentile for the 100m, 1500m and Marathon
are 15.9, 6:06.5 and 6:15:34, respectively. As performances for the two genders distribute differ-
ently, we present only results on the subset of 101,775 male runners in the main corpus of the
manuscript; female runners and further subgroup analyses are considered in the supplemen-
tary results. The data set is available upon request, subject to approval by British Athletics. Full
code and data for our analyses can be obtained from [24, 25].
Prediction and Quantification of Individual Athletic Performance of Runners
PLOS ONE | DOI:10.1371/journal.pone.0157257
June 23, 2016
5 / 16
Adhering to state-of-the-art statistical practice (see [26–29]), all prediction methods are val-
idated out-of-sample, i.e., by using only a subset of the data for estimation of parameters (train-
ing set) and computing the error on predictions made for a distinct subset (validation or test
set). As error measures, we use the root mean squared error (RMSE) and the mean absolute
error (MAE), estimated by leave-one-out validation for 1000 single performances omitted at
random.
We would like to stress that out-of-sample prediction error is the correct way to evaluate
the quality of prediction, as opposed to merely reporting goodness-of-fit in-sample, since out-
putting an estimate for an instance that the method has already seen does not qualify as
prediction.
More details on the data set and our validation setup can be found in the supplementary
material.
Table 1. The three components of the low-rank model of Eq (1).
s
100m
200m
400m
800m
1500m
Mile
5km
10km
HM
Mar
f1
2.254
2.875
3.574
4.305
4.964
5.049
6.179
6.844
7.555
8.243
f2
0.4473
0.4721
0.5265
0.3045
0.0798
0.0806
-0.1597
-0.1983
-0.2279
-0.2785
f3
-0.1750
-0.2004
-0.1145
0.2224
0.3263
0.3092
0.3157
0.2717
-0.1153
-0.6912
v
0.1291
0.1647
0.2047
0.2466
0.2843
0.2892
0.3539
0.3920
0.4327
0.4721
An entry in the rows i = 1,2,3 is fi(s), where s is the column header; HM is the half-Marathon, Mar is the Marathon. The components are obtained as
described in methods, “obtaining the low-rank components and coefficients”. v is the raw singular vector described there from which f1 is obtained by
rescaling. v, f2, f3 are displayed in Fig 2 top left with standard error tubes per entry. The entries for v have, on average, an estimated standard error of
0.005, the entries for f2 have, on average, an estimated standard error of 0.02, and the entries for f3 have, on average, an estimated standard error of
0.04.
doi:10.1371/journal.pone.0157257.t001
Fig 2. The three components of the low-rank model, and explanation of the world record data. Left: the components displayed (unit norm, log-time vs
log-distance). Tubes around the components are one standard deviation, estimated by the bootstrap. The first component is an exact power law (straight line
in log-log coordinates); the last two components are non-linear, describing transitions at around 800m and 10km. Middle: Comparison of first component and
world record to the exact power law (log-speed vs log-distance). Right: Least-squares fit of rank 1-3 models to the world record data (log-speed vs log-
distance).
doi:10.1371/journal.pone.0157257.g002
Prediction and Quantification of Individual Athletic Performance of Runners
PLOS ONE | DOI:10.1371/journal.pone.0157257
June 23, 2016
6 / 16
Results
(I) Prediction accuracy. We evaluate prediction accuracy of ten methods, including our pro-
posed method, LMC. We include, as naive baselines: (1.a) imputing the event mean, (1.b)
imputing the average of the k-nearest neighbours; as representative of the state-of-the-art in
quantitative sports science: (2.a) the Riegel formula, (2.b) a power law predictor with exponent
estimated from the data, which is the same for all runners, (2.c) a power law predictor with
exponent estimated from the data, with one exponent per runner, (2.d) the Purdy points
scheme [7]; as representatives for the state-of-the-art in matrix completion: (3.a) imputation by
expectation maximization on a multivariate Gaussian [30] (3.b) nuclear norm minimization
[31, 32].
We instantiate our low-rank local matrix completion (LMC) in two variants: (4.a) rank 1,
and (4.b) rank 2.
Methods (1.a), (1.b), (2.a), (2.b), (2.d), (4.a) require at least one observed performance per
runner, methods (2.c), (4.b) require at least two observed performances in distinct events. Meth-
ods (3.a), (3.b) will return a result for any number of observed performances (including zero).
Prediction accuracy is therefore measured by evaluating the RMSE and MAE out-of-sample on
the runners who have attempted at least three distances, so that the two necessary performances
remain to calculate the prediction when one is removed for leave-one-out validation. Prediction
is further restricted to the best 95-percentile of runners (measured by performance in the best
event) to reduce the effect of outliers. Whenever the method demands that the predicting events
need to be specified, the events which are closest in log-distance to the event to be predicted are
taken. The accuracy of predicting time (normalized w.r.t. the event mean), log-time, and speed
are measured. We repeat this validation setup for the year of best performance and a random cal-
endar year. Moreover, for completeness and comparison we treat 2 additional cases: the top 25%
of runners and runners who have attempted at least 4 events, each in log time. More details on
methods and validation are presented in the methods appendix.
The results are displayed in Table 2 (RMSE of log-time prediction) and supplementary
Table B in S1 Supplement (MAE of log-time prediction), S4 (rel.RMSE of time prediction) and
S5 (rel. MAE of time prediction). Of all benchmarks, k-nearest neighbours (1.b), Purdy points
(2.d) and Expectation Maximization (3.a) perform best. LMC rank 2 substantially outperforms
k-nearest neighbours, Purdy points and Expectation Maximization (two-sided Wilcoxon
signed-rank test significant on the validation samples of absolute prediction errors; p2.0e-8
on top 95% in log-time and p1.4e-11 for top 25% in log-time); rank 1 outperforms Purdy
points on the year of best performance data (p3.0e-3) for the best runners, and is on a par on
runners up to the 95th percentile. Both rank 1 and 2 outperform the power law models
(p1.1e-42), the improvement in RMSE of LMC rank 2 over the power law models reaches
over 50% for data from the fastest 25% of runners.
(II) The rank (number of components) of the model. Paragraph (I) establishes that LMC
is the best method for prediction. LMC assumes a fixed number of prototypical runners, viz.
the rank r, which is the complexity parameter of the model. We establish the optimal rank by
comparing prediction accuracy of LMC with various ranks. The rank r algorithm requires r
attempted events for prediction, thus r + 1 observed events are needed for validation. Table F
in S1 Supplement displays prediction accuracies for LMC ranks r = 1 to r = 4, on the runners
who have attempted k > r events, for all k 5. The data is restricted to the top 25% in the year
of best performance in order to obtain a high signal to noise ratio. We observe that rank 3 out-
performs all other ranks, when applicable; rank 2 aways outperforms rank 1 (both p1e-4).
We also find that the improvement of rank 2 over rank 1 depends on the event predicted:
improvement is 26.3% for short distances (100m, 200m), 29.3% for middle distances (400m,
Prediction and Quantification of Individual Athletic Performance of Runners
PLOS ONE | DOI:10.1371/journal.pone.0157257
June 23, 2016
7 / 16
Table 2. Out-of-sample RMSE for prediction methods on different data setups.
Generic Baselines
State of art Performance Predictors
State of art Matrix
Completion
Proposed Method:
LMC
evaluation percentiles no.events
data type
r.mean
k-NN
individual
power law
riegel
power law
purdy
nuclear
norm
EM
LMC rank
1
LMC rank
2
log time
0-95
3
best
0.1308
± 0.0032
0.0618
± 0.0027
0.1033
± 0.0042
0.0982
± 0.0046
0.0973
± 0.0046
0.0610
± 0.0031
0.3909
± 0.0457
0.0566
± 0.0028
0.0586
± 0.0026
0.0515
± 0.0027
normalized
0-95
3
best
0.1364
± 0.0044
0.0716
± 0.0046
0.1067
± 0.0048
0.1059
± 0.0066
0.1050
± 0.0065
0.0684
± 0.0043
0.1900
± 0.0120
0.0634
± 0.0045
0.0643
± 0.0038
0.0576
± 0.0039
speed
0-95
3
best
0.6655
± 0.0147
0.3057
± 0.0146
0.6096
± 0.0245
0.5467
± 0.0251
0.5415
± 0.0243
0.3077
± 0.0176
26.6210
± 11.4828
0.2922
± 0.0165
0.3123
± 0.0149
0.2530
± 0.0129
log time
0-95
3
random
0.1380
± 0.0032
0.0544
± 0.0025
0.0931
± 0.0035
0.0931
± 0.0038
0.0919
± 0.0038
0.0591
± 0.0028
0.4416
± 0.0428
0.0561
± 0.0031
0.0567
± 0.0027
0.0471
± 0.0024
normalized
0-95
3
random
0.1450
± 0.0043
0.0623
± 0.0037
0.0951
± 0.0039
0.1011
± 0.0048
0.0998
± 0.0046
0.0682
± 0.0038
0.2046
± 0.0117
0.0634
± 0.0039
0.0640
± 0.0037
0.0538
± 0.0035
speed
0-95
3
random
0.6935
± 0.0143
0.2585
± 0.0117
0.5917
± 0.0312
0.5052
± 0.0176
0.4979
± 0.0167
0.2835
± 0.0137
24.7206
± 10.9157
0.2801
± 0.0196
0.2863
± 0.0120
0.2261
± 0.0105
log time
0-95
4
best
0.1268
± 0.0032
0.0735
± 0.0030
0.0777
± 0.0024
0.0819
± 0.0032
0.0822
± 0.0032
0.0581
± 0.0023
0.1779
± 0.0199
0.0529
± 0.0024
0.0536
± 0.0021
0.0467
± 0.0022
log time
0-25
3
best
0.0557
± 0.0015
0.0416
± 0.0014
0.0806
± 0.0031
0.0683
± 0.0026
0.0720
± 0.0026
0.0411
± 0.0012
0.3008
± 0.0275
0.0383
± 0.0013
0.0411
± 0.0014
0.0306
± 0.0011
Predicted performance is of the 95 and 25 top percentiles of male runners who attempted 3 or 4 events resp., in their best year and a random calendar year. Standard errors are
bootstrap estimates over 1000 repetitions. Compared method classes are (1) generic baselines, (2) state-of-the-art in performance prediction, (3) state-of-the-art in matrix
completion, (4) local matrix completion (columns). Methods are (1.a) r.mean: predicting the mean of all runners (1.b) k-NN: predicting the nearest neighbor. (2.a) riegel: Riegel’s
formula (2.b) power law: power law with free exponent and coefficient. Exponent is the same for all runners. (2.c) ind.power law: power law with free exponent and coefficient. (2.d)
purdy: Purdy points scheme (3.a) EM: expectation maximization (3.b) nuclear norm: nuclear norm minimization (4.a) LMC with rank 1 (4.b) LMC with rank 2. Data setup is specified
by (i) evaluation: what is predicted. log-time = natural logarithm of time in seconds, normalized = time relative to mean performance, speed = average speed in meters per seconds,
(ii) percentiles: selected percentile range of runners, (iii) no.events tried = sub-set of runners who have attempted at least that number of different events, (iv) data type: collation
mode of performance matrix; best = 1 year around best performance, random = random 2 year period. LMC rank 2 significantly outperforms all competitors in either setting.
doi:10.1371/journal.pone.0157257.t002
Prediction and Quantification of Individual Athletic Performance of Runners
PLOS ONE | DOI:10.1371/journal.pone.0157257
June 23, 2016
8 / 16
800m, 1500m), 12.8% for the mile to half-marathon, and 3.1% for the Marathon (all significant
at p = 1e-3 level) (see Fig A in S1 Supplement). These results indicate that inter-runner variabil-
ity is greater for short and middle distances than for Marathon.
(III) The three components of the model. The findings in (II) imply that the best low-rank
model assumes 3 components. To estimate the components (fi in Eq (1)) we impute all missing
entries in the data matrix of the top 25% runners who have attempted 4 events and compute its
singular value decomposition (SVD) [33]. From the SVD, the exact form of components may
be directly obtained as the right singular vectors (in a least-squares sense, and up to scaling, see
supplemental analysis II.b in S1 Supplement). We obtain three components in log-time coordi-
nates, which are displayed in the left hand panel of Fig 2. The first component for log-time pre-
diction is linear (i.e., f1(s) / log s in Eq (1)) to a high degree of precision (R2 = 0.9997) and
corresponds to an individual power law, applying distinctly to each runner. The second and
third components are non-linear; the second component decreases over short sprints and
increases over the remainder, and the third component resembles a parabola with extremum
positioned around the middle distances.
In speed coordinates, the first, individual power law component does not display the “bro-
ken power law” behaviour of the world records (rank 1 component: goodness-of-fit for linear
model R2 = 0.99; world-record data: R2 = 0.93). Deviations from an exact line can be explained
by the second and third component (Fig 2 middle).
The three components together explain the world record data and its “broken power law”
far more accurately than a simple linear power law trend—with the rank 3 model fitting the
world records almost exactly (Fig 2 right).
(IV) The three runner-specific coefficients. The three summary coefficients for each run-
ner (λ1, λ2, λ3 in Eq (1)) are obtained from the entries of the left singular vectors (see methods
appendix). Since all three coefficients summarize the runner, we refer to them collectively as
the three-number-summary.
(IV.i) Fig 3 displays scatter plots and Spearman correlations between the coefficients and
performance over the full range of distances. The individual exponent correlates with perfor-
mance over distances greater than 800m. The second coefficient correlates positively with per-
formance over short distances and displays a non-linear association with performance over
Fig 3. Matrix scatter plot of the three-number-summary vs performance. For each of the scores in the
three-number-summary (rows) and each event distance (columns), the plot matrix shows: a scatter plot of
performances (time) vs the coefficient score of the top 25% (on the best event) runners who have attempted
at least 4 events. Each scatter plot in the matrix is colored on a continuous color scale according to the
absolute value of the scatter sample’s Spearman rank correlation (red = 0, green = 1).
doi:10.1371/journal.pone.0157257.g003
Prediction and Quantification of Individual Athletic Performance of Runners
PLOS ONE | DOI:10.1371/journal.pone.0157257
June 23, 2016
9 / 16
middle distances. The third coefficient correlates with performance over middle distances. (All
correlations are significant at p1.0e-4; t-distribution approximation to the distribution of
Spearman’s correlation coefficient.) The associations for all three coefficients are non-linear,
with the notable exception of the individual exponent on distances exceeding 800m.
(IV.ii) Fig 4 top displays the three-number-summary for the top 95% runners in the data-
base. The runners separate into (at least) four classes, which are associated with the runner’s
preferred distance. A qualitative transition can be observed over middle distances. Three-num-
ber-summaries of world class runners (not all in the UK runners database), computed from
their personal bests, are listed in Table 3; they and also shown as highlighted points in Fig 4 top
right. The elite runners trace a frontier around the population: all elite runners are subject to a
low individual exponent. A hypothetical runner holding all the world records is also shown in
Fig 4 top right, obtaining an individual exponent which comes close to the world record expo-
nent estimated by Riegel [3] (1.08 for elite runners, 1.06 for senior runners).
(IV.iii) Fig 4 bottom left shows that a low individual exponent correlates positively with per-
formance in a runner’s preferred event. The individual exponents are higher on average
(median = 1.12; 5th, 95th percentiles = 1.10, 1.15) than the world record exponents estimated
by Riegel.
Fig 4. Scatter plots exploring the three number summary. Top left and right: 3D scatter plot of three-
number-summaries of runners in the data set, colored by preferred distance and shown from two angles. A
negative value for the second score is a indicates that the runner is a sprinter, a positive value an endurance
runner. In the top right panel, the summaries of the elite runners Usain Bolt (world record holder, 100m,
200m), Mo Farah (world beater over distances between 1500m and 10km), Haile Gabrselassie (former world
record holder from 5km to Marathon) and Takahiro Sunada (100km world record holder) are shown;
summaries are estimated from their personal bests. For comparison we also display the hypothetical data of
a runner who holds all world records. Bottom left: preferred distance vs individual exponents, color is
percentile on preferred distance. Bottom right: age vs. exponent, colored by preferred distance.
doi:10.1371/journal.pone.0157257.g004
Prediction and Quantification of Individual Athletic Performance of Runners
PLOS ONE | DOI:10.1371/journal.pone.0157257
June 23, 2016
10 / 16
(IV.iv) Fig 4 bottom right shows that in cross-section, the individual exponent decreases
with age until 20 years, and subsequently increases. (All correlations significant at p1.0e-4; t-
distribution approximation to the distribution of Spearman’s correlation coefficient.)
(V) Phase transitions. We observe two transitions in behaviour between short and long dis-
tances. The data exhibit a phase transition around 800m: the second component exhibits a
kink and the third component makes a zero transition (Fig 2); the association of the first two
scores with performance shifts from the second to the first score (Fig 3). The data also exhibits
a transition around 5000m. We find that for distances shorter than 5000m, holding the event
performance constant and increasing the standard of shorter events leads to a decrease in the
predicted standard of longer events and vice versa. On the other hand for distances greater
than 5000m this behaviour reverses; holding the event performance constant, and increasing
the standard of shorter events leads to an increase in the predicted standard of longer events.
See supplementary analysis IV in S1 Supplement for details.
(VI) Universality over subgroups. Qualitatively and quantitatively similar results to the
above can be deduced for female runners, and subgroups stratified by age or training standard;
LMC remains an accurate predictor, and the low-rank model has similar form. See supplemen-
tal analysis II.c in S1 Supplement.
Discussion
We have presented the most accurate existing predictor for running performance to date—
local low-rank matrix completion (finding I); its predictive power confirms the validity of a
three-component model (finding II) that offers a parsimonious explanation for many known
phenomena in the quantitative science of running, including answers to some of the major
open questions of the field. More precisely, we establish:
The individual power law. In log-time coordinates, the first component of our physiologi-
cal model is linear with high accuracy, yielding an individual power law (finding III). This is a
novel and rather surprising finding, since, although world-record performances are known to
obey a power law [1–6], there is no reason to suppose a-priori that the performance of individ-
uals is governed by a power law. Striking is that the power-law derived is considerably more
accurate when considered in log-distance—log-speed coordinates than the power-law which
applies to world-record data. This parsimony a-posteriori unifies (A) the parsimony of the
Table 3. Estimated three-number-summary (λi) for a selection of elite runners.
runner
Specialization
Individual
Exponent (λ1)
Score 2 (λ2)
Score 3 (λ3)
Usain Bolt
Sprints
1.11
-0.367
0.081
Mo Farah
Middle-Long
1.08
0.033
-0.076
Haile Gabrselassie
Long
1.08
0.114
-0.056
Galen Rupp
Long
1.08
0.104
-0.040
Seb Coe
Middle
1.09
-0.085
-0.036
Takahiro Sunada
Ultra-Long
1.09
0.138
-0.010
Paula Radcliffe
Long (Female)
1.10
0.189
0.025
The scores λ1, λ2, λ3 are as in Eq (1). Since component 1 is a power law (see the top-left of Fig 2), λ1 may
be interpreted as an individual exponent. See the bottom right panel of Fig 4 for a scatter plot of the
runners in the database.
doi:10.1371/journal.pone.0157257.t003
Prediction and Quantification of Individual Athletic Performance of Runners
PLOS ONE | DOI:10.1371/journal.pone.0157257
June 23, 2016
11 / 16
power law with the (B) empirical correctness of scoring tables. To what extent this individual
power law is exact is to be determined in future studies.
An explanation of the world record data. The broken power law [5] of world record data
can be seen as a consequence of the individual power law and the non-linearity in the second
and third component (finding III) of our low-rank model. The breakage point in the world rec-
ords can be explained by the differing contributions in the non-linear components of the dis-
tinct individuals holding the world records. Savaglio and Carbone [5] hypothesize that the
breakpoint in the log-speed—log-distance slope of world-record data, which occurs between
short and long distance events, is due to a transition in the physiology required between short
and long-distance events. Our analyses indeed show that their exist breakpoints, manifested in
the second and third components of the low-rank model. However our findings show that the
claim that there is a universal physiological transition manifesting itself in the differing slopes
of short and long-distance world-record data is unwarranted. Runners who exhibit small values
for the 2nd and 3rd numbers in their three number summaries will exhibit performances close
to log-log with little or no transition; this is because the first component of the model is much
closer to scale-free (log-log straight line) than world-record data. Some runners will indeed dis-
play an upward kink in their average speed as is the case with world-record data. Other runners
will exhibit transitions corresponding to a quicker fall off in average speed rather than faster,
i.e. a downwards kink. Thus the validity of the three component model points to a far more
complex description and diversity of average speed than world record data suggest.
Crucially both the power law and the broken power law on world record data can be under-
stood as epiphenomena of the individual power law and its non-linear corrections.
Universality of our model. The low-rank model remains unchanged when considering dif-
ferent subgroups of runners, stratified by gender, age, or calendar year; only the individual
three-number-summaries change (finding VI). This shows the low-rank model to be universal
for running.
The three-number-summary reflects a runner’s training state. Our predictive validation
implies that the number of components of our model is three (finding II), which yields three
numbers describing the training state of a given runner (finding IV). The most important sum-
mary is the individual exponent for the individual power law which describes overall perfor-
mances for distances longer than 400m (IV.iii). The second coefficient describes whether the
runner has greater endurance (positive) or speed (negative) and predicts performances over
the sprint distances, the third describes specialization over middle distances (negative) vs. short
and long distances (positive). All three numbers together clearly separate the runners into four
clusters, which fall into two clusters of short-distance runners and one cluster of middle-and
long-distance runners respectively (IV.i). Our analysis provides strong evidence that the three-
number-summary captures physiological and/or social/behavioural characteristics of the run-
ners, e.g., training state, specialization, and which distance a runner chooses to attempt. While
the data set does not allow us to separate these potential influences or to make statements
about cause and effect, we conjecture that combining the three-number-summary with specific
experimental paradigms will lead to a clarification; further, we conjecture that a combination
of the three-number-summary with additional data, e.g. training logs, high-frequency training
measurements or clinical parameters, will lead to a better understanding of (C) existing physio-
logical models.
Some novel physiological insights can be deduced from leveraging our model on the UK
runners database:
• We find that the individual exponent correlates with performances over distances greater
than 400m and especially long distances above 5km (finding III). We also find that LMC is
Prediction and Quantification of Individual Athletic Performance of Runners
PLOS ONE | DOI:10.1371/journal.pone.0157257
June 23, 2016
12 / 16
most effective for the longer-sprints and middle distances; the improvement of the higher
rank over the rank 1 version is lowest over the marathon distance (supplemental analysis I.c
in S1 Supplement). This indicates that the variability in performances on long distances may
to a large extent be explained by a single factor, which may imply that there is only one way
to be a fast marathoner. On the other hand since we find that the rank-2 and 3 versions far
outperform the rank-1 version over middle distances, this may be interpreted in terms of
some runners using a high maximum velocity to coast whereas other runners using greater
endurance to run closer to their maximum speed for the duration of the race; if the type of
running (coasting vs. endurance) is a physiological correlate to the specialization summary
(as hypothesized above), it could imply that the “one way” corresponds to possessing a high
level of endurance—as opposed to being able to coast relative to a high maximum speed. In
any case, the low-rank model predicts that a marathoner who is not close to world class over
10km is unlikely to be a world class marathoner.
• The phase transitions which we observe (finding V) provide additional observational evi-
dence for a transition in the complexity of the physiology underlying performance between
long and short distances. This finding is bolstered by the difference we observe between the
increase in performance of the rank 2 predictor over the rank 1 predictor for short/middle
distances over long distances. Notice, however, that this is quite different evidence than the
kink in the power-law of world-record speeds [5], which we argued above does not necessar-
ily imply the presence of transitions at the level of the individual runner. Our results may
have implications for existing hypotheses and findings in sports science on the differences in
physiological determinants of long and short distance running respectively. These include
differences in the muscle fibre types contributing to performance (type I vs. type II) [34, 35],
whether the race length demands energy primarily from aerobic or anaerobic metabolism
[20, 36], which energy systems are mobilized (glycolysis vs. lipolysis) [37, 38] and whether
the race terminates before the onset of a _VO2 slow component [39, 40]. We conjecture that
the combination of our methodology with experiments will shed further light on these
differences.
• An open question in the physiology of aging is whether sprinting power or endurance capa-
bilities diminish faster with age. Our analysis provides cross-sectional evidence that training
standard decreases with age, and specialization shifts away from endurance: a larger expo-
nent is correlated with worse performance on endurance type events (finding IV.i), and
exponents increase, in cross-section, with age (finding IV.iv). This confirms observations of
Rittweger et al. [41] on masters world-record data. There are multiple possible explanations
for this, for example longitudinal changes in specialization, or selection bias due to the dis-
tances older runners prefer; our model renders these hypotheses amenable to quantitative
validation.
• We find that there are a number of high-standard runners who attempt distances different
from their inferred best distance; most notably a cluster of young runners (<25 yrs.) who
run short distances (mostly in accordance with legal limitations of participation), and a clus-
ter of older runners (>40 yrs.) who run long distances, but who we predict would perform
better on longer resp. shorter distances. Moreover, the third component of our model implies
the existence of runners with very strong specialization in their best event; there are indeed
high profile examples of such runners, such as Zersenay Tadese, who holds the half-mara-
thon world best performance (58:23) but has as yet to produce a marathon performance even
close to this in quality (best performance, 2:10:41).
Prediction and Quantification of Individual Athletic Performance of Runners
PLOS ONE | DOI:10.1371/journal.pone.0157257
June 23, 2016
13 / 16
We also anticipate that our framework will prove fruitful in equipping the practioner with
new methods for prediction and quantification:
• Individual predictions are crucial in race planning, especially for predicting a target perfor-
mance for events such as the Marathon for which months of preparation are needed; the abil-
ity to accurately select a realistic target speed could potentially make the difference between a
runner achieving a personal best performance and “hitting the wall” or at worst dropping out
of the race from exhaustion.
N.B.: We would like to stress that using a prediction as part of marathon preparation without
professional support may lead to injury and is entirely at the risk of the user.
• Predictions and the three-number-summary yield a concise description of the runner’s spe-
cialization and training state and are thus of immediate use in training assessment and plan-
ning, for example in determining the potential effect of a training scheme or finding the
optimal event(s) for which to train.
N.B.: We would like to stress that our study is not able to assign a conclusive meaning to the
three-number summary, due to the limitations of the data set; therefore decisions should not be
based on a hypothesized interpretation without consideration.
• The presented framework allows, in principle, for the derivation of novel and more accurate
scoring schemes, including scoring tables for any type of population.
N.B.: We would like to stress that the form of the derived scoring tables may depend on the
selection of the data from which they are derived.
• Predictions for elite runners allow for a more precise estimation of quotas and betting risk.
For example, we predict that a fair race between Mo Farah and Usain Bolt is over 492m (374-
594m with 95% confidence), Chris Lemaitre and Adam Gemili have the calibre to run 43.5
(±1.3) and 43.2 (±1.3) resp. seconds over 400m. Kenenisa Bekele is capable, in a training
state where he can achieve his personal bests over 5km, 10km and the half-marathon, of a
2:00:36 marathon (±3.6 mins).
N.B.: We would like to stress that such predictions need to be taken with much caution, as they
are only correct insofar as our model extends, from the top 25% of UK runners (who success-
fully participated in official events), to the very extremes of human performance.
We further conjecture that the physiological laws we have validated for running will be
immediately transferable to any sport where a power law has been observed on the collective
level, such as swimming, cycling, and horse racing.
Supporting Information
S1 Supplement. Additional analyses and method details with corresponding figures and
tables.
(PDF)
Acknowledgments
DAJB was supported by a grant from the German Research Foundation, research training
group GRK 1589/1 “Sensory Computation in Neural Systems”. FJK was partially supported by
Mathematisches Forschungsinstitut Oberwolfach (MFO). This research was partially carried
out at MFO with the support of FJK’s Oberwolfach Leibniz Fellowship. We thank thepo-
werof10.info for permission to use their database for this paper, Ryota Tomioka for providing
us with his code for matrix completion via nuclear norm minimization, and for advice on its
Prediction and Quantification of Individual Athletic Performance of Runners
PLOS ONE | DOI:10.1371/journal.pone.0157257
June 23, 2016
14 / 16
use, Louis Theran for advice regarding the implementation of local matrix completion in
higher ranks. We thank Denis Bafounta, Renato Canova, Tim Grose, Florian Lorenz, Klaus-
Robert Müller and Franz Wölfle for remarks, and discussion of the concepts and results pre-
sented in the manuscript.
Author Contributions
Conceived and designed the experiments: DAJB FJK. Performed the experiments: DAJB. Ana-
lyzed the data: DAJB FJK. Contributed reagents/materials/analysis tools: DAJB FJK. Wrote the
paper: DAJB FJK. Conceived the LMC algorithm in higher ranks: FJK. Adapted the LMC algo-
rithm and designed its concrete implementation for performance prediction: DAJB.
References
1.
Lietzke MH. An analytical study of world and olympic racing records. Science. 1954; 119(3089):333–
336. doi: 10.1126/science.119.3089.333 PMID: 17756808
2.
Henry FM. Prediction of world records in running sixty yards to twenty-six miles. Research Quarterly
American Association for Health, Physical Education and Recreation. 1955; 26(2):147–158.
3.
Riegel PS. Athletic records and human endurance. American Scientist. 1980; 69(3):285–290.
4.
Katz L, Katz JS. Fractal (power-law) analysis of athletic performance. Research in Sports Medicine: An
International Journal. 1994; 5(2):95–105.
5.
Savaglio S, Carbone V. Human performance: scaling in athletic world records. Nature. 2000; 404
(6775):244–244. doi: 10.1038/35005165 PMID: 10749198
6.
García-Manso JM, Martín-González JM, Dávila N, Arriaza E. Middle and long distance athletics races
viewed from the perspective of complexity. Journal of theoretical biology. 2005; 233(2):191–198. doi:
10.1016/j.jtbi.2004.10.014 PMID: 15619360
7.
Purdy JG. Computer generated track and field scoring tables: II. Theoretical foundation and develop-
ment of a model. Medicine and science in sports. 1974; 7(2):111–115.
8.
Purdy JG. Computer generated track and field scoring tables: III. Model evaluation and analysis. Medi-
cine and science in sports. 1976; 9(4):212–218.
9.
Hill AV, Long C, Lupton H. Muscular exercise, lactic acid, and the supply and utilisation of oxygen. Pro-
ceedings of the Royal Society of London Series B, Containing Papers of a Biological Character. 1924;
p. 84–138. doi: 10.1098/rspb.1924.0045
10.
Billat LV, Koralsztein JP. Significance of the velocity at VO2-max and time to exhaustion at this velocity.
Sports Medicine. 1996; 22(2):90–108. doi: 10.2165/00007256-199622020-00004 PMID: 8857705
11.
Wasserman K, Whipp BJ, Koyl S, Beaver W. Anaerobic threshold and respiratory gas exchange during
exercise. Journal of applied physiology. 1973; 35(2):236–243. PMID: 4723033
12.
Noakes TD, Myburgh KH, Schall R. Peak treadmill running velocity during the VO2max test predicts
running performance. Journal of sports sciences. 1990; 8(1):35–45. doi: 10.1080/02640419008732129
PMID: 2359150
13.
Billat LV. Use of blood lactate measurements for prediction of exercise performance and for control of
training. Sports medicine. 1996; 22(3):157–175. doi: 10.2165/00007256-199622030-00003 PMID:
8883213
14.
Keller JB. A theory of competitive running. Physics today. 1973; p. 43. doi: 10.1063/1.3128231
15.
Péronnet F, Thibault G. Mathematical analysis of running performance and world running records. Jour-
nal of Applied Physiology. 1989; 67(1):453–465. PMID: 2759974
16.
van Schenau GJI, de Koning JJ, de Groot G. Optimisation of sprinting performance in running, cycling
and speed skating. Sports Medicine. 1994; 17(4):259–275. doi: 10.2165/00007256-199417040-00006
17.
Kennelly AE. An approximate law of fatigue in the speeds of racing animals. In: Proceedings of the
American Academy of Arts and Sciences. JSTOR; 1906. p. 275–331.
18.
Riegel PS. Time predicting. Runner’s World Magazine. 1977;.
19.
Purdy JG. Computer generated track and field scoring tables: I. Historical development. Medicine and
science in sports. 1974; 6(4):287. PMID: 4618326
20.
Bosquet L, Léger L, Legros P. Methods to determine aerobic endurance. Sports Medicine. 2002; 32
(11):675–700. doi: 10.2165/00007256-200232110-00002 PMID: 12196030
Prediction and Quantification of Individual Athletic Performance of Runners
PLOS ONE | DOI:10.1371/journal.pone.0157257
June 23, 2016
15 / 16
21.
Bundle MW, Hoyt RW, Weyand PG. High-speed running performance: a new approach to assessment
and prediction. Journal of Applied Physiology. 2003; 95(5):1955–1962. doi: 10.1152/japplphysiol.
00921.2002 PMID: 14555668
22.
Di Prampero PE. Factors limiting maximal performance in humans. European journal of applied physi-
ology. 2003; 90(3-4):420–429. doi: 10.1007/s00421-003-0926-z PMID: 12910345
23.
Király FJ, Theran L, Tomioka R. The algebraic combinatorial approach for low-rank matrix completion.
Journal of Machine Learning Research. 2015;.
24.
Blythe DAJ, Király FJ. Full data to “Prediction and Quantification of Individual Athletic Performance of
Runners”; 2016. doi: 10.6084/m9.figshare.3408202.v1 Available from: https://figshare.com/articles/
thepowerof10/3408202
25.
Blythe DAJ, Király FJ. Full code to “Prediction and Quantification of Individual Athletic Performance of
Runners”; 2016. doi: 10.6084/m9.figshare.3408250.v1 Available from: https://figshare.com/articles/
Ful_code_to_Prediction_and_Quantification_of_Individual_Athletic_Performance_of_Runners_/
3408250
26.
Efron B. Estimating the error rate of a prediction rule: improvement on cross-validation. Journal of the
American Statistical Association. 1983; 78(382):316–331. doi: 10.1080/01621459.1983.10477973
27.
Kohavi R, et al. A study of cross-validation and bootstrap for accuracy estimation and model selection.
In: IJCAI. vol. 14; 1995. p. 1137–1145.
28.
Efron B, Tibshirani R. Improvements on cross-validation: the 632+ bootstrap method. Journal of the
American Statistical Association. 1997; 92(438):548–560. doi: 10.1080/01621459.1997.10474007
29.
Browne MW. Cross-validation methods. Journal of Mathematical Psychology. 2000; 44(1):108–132.
doi: 10.1006/jmps.1999.1279 PMID: 10733860
30.
Dempster AP, Laird NM, Rubin DB. Maximum likelihood from incomplete data via the EM algorithm.
Journal of the royal statistical society Series B (methodological). 1977; p. 1–38.
31.
Candès EJ, Recht B. Exact matrix completion via convex optimization. Foundations of Computational
mathematics. 2009; 9(6):717–772. doi: 10.1007/s10208-009-9045-5
32.
Candès EJ, Tao T. The power of convex relaxation: near-optimal matrix completion. Information The-
ory, IEEE Transactions on. 2010; 56(5):2053–2080. doi: 10.1109/TIT.2010.2044061
33.
Golub GH, Reinsch C. Singular value decomposition and least squares solutions. Numerische Mathe-
matik. 1970; 14(5):403–420. doi: 10.1007/BF02163027
34.
Saltin B, Henricksson J, Hygaard E, Andersen P. Fibre types and metabolic potentials of skeletal mus-
cles in sedentary man and endurance runners. Annals of the New York Academy of Sciences. 1977; p.
3–29. doi: 10.1111/j.1749-6632.1977.tb38182.x PMID: 73362
35.
Hoppeler H, Howald H, Conley K, Lindstedt SL, Claassen H, Vock P, et al. Endurance training in
humans: aerobic capacity and structure of skeletal muscle. Journal of Applied Physiology. 1985; 59
(2):320–327. PMID: 4030584
36.
Faude O, Kindermann W, Meyer T. Lactate threshold concepts. Sports Medicine. 2009; 39(6):469–
490. doi: 10.2165/00007256-200939060-00003 PMID: 19453206
37.
Brooks GA, Mercier J. Balance of carbohydrate and lipid utilization during exercise: the “crossover”
concept. Journal of Applied Physiology. 1994; 76(6):2253–2261. PMID: 7928844
38.
Venables MC, Achten J, Jeukendrup AE. Determinants of fat oxidation during exercise in healthy men
and women: a cross-sectional study. Journal of applied physiology. 2005; 98(1):160–167. doi: 10.1152/
japplphysiol.00662.2003 PMID: 15333616
39.
Borrani F, Candau R, Millet G, Perrey S, Fuchslocher J, Rouillon J. Is the VO2 slow component depen-
dent on progressive recruitment of fast-twitch fibers in trained runners? Journal of Applied Physiology.
2001; 90(6):2212–2220. PMID: 11356785
40.
Poole DC, Barstow TJ, Gaesser GA, Willis WT, Whipp BJ. VO2 slow component: physiological and
functional significance. Medicine and science in sports and exercise. 1994; 26(11):1354–1358. PMID:
7837956
41.
Rittweger J, di Prampero PE, Maffulli N, Narici MV. Sprint and endurance power and ageing: an analy-
sis of master athletic world records. Proceedings of the Royal Society B: Biological Sciences. 2009;
276(1657):683–689. doi: 10.1098/rspb.2008.1319 PMID: 18957366
Prediction and Quantification of Individual Athletic Performance of Runners
PLOS ONE | DOI:10.1371/journal.pone.0157257
June 23, 2016
16 / 16
| Prediction and Quantification of Individual Athletic Performance of Runners. | 06-23-2016 | Blythe, Duncan A J,Király, Franz J | eng |
PMC7573630 | Supplementary Information
Efficient trajectory optimization for curved running
using a 3D musculoskeletal model with implicit
dynamics
Marlies Nitschke1,*, Eva Dorschky1, Dieter Heinrich2, Heiko Schlarb3, Bjoern M.
Eskofier1, Anne D. Koelewijn1,4, and Antonie J. van den Bogert4
1Machine Learning and Data Analytics Lab, Department of Computer Science, Friedrich-Alexander-Universit¨at
Erlangen-N¨urnberg (FAU), Erlangen, Germany
2Department of Sport Science, University of Innsbruck, Innsbruck, Austria
3adidas AG, Herzogenaurach, Germany
4Department of Mechanical Engineering, Cleveland State University, Cleveland, USA
*[email protected]
S1 Model Adaptations
The proposed “running model for motions in all directions”, short “runMaD”, was adapted from a model proposed by Hamner
et al.1 in order to simulate running with directional changes. We changed the sequence of pelvis rotations from tilt (z), list (x),
rotation (y) to rotation (y), obliquity (x), tilt (z) (see Fig. 1 for the global coordinate system). The updated rotation sequence is
in agreement with clinical understanding2. Hence, results are interpretable in clinical analysis independently of the movement
direction. Furthermore, this sequence simplifies the specification of directional tasks in simulations, e.g. in the presented curved
running simulation.
Pronation and supination of the foot was enabled by unlocking the subtalar joint. The metatarsophalangeal (mtp) joint was
also unlocked for roll over of the foot. For both joints, a range of motion from −90◦ to 90◦ was allowed. In order to fit the
recorded data of fast running, the upper limit of knee flexion was increased from 120◦ to 160◦. Additionally, the range of motion
of the pronation/supination angle at the elbow was enlarged from [0◦,90◦] to [0◦,150◦] and the default pronation/supination
angle was set to 90◦ such that the palms were pointing towards the body for zero torque. The default elbow flexion was set to
5◦ to be within the range of motion.
S2 System Dynamics
The system dynamics were described implicitly as a function f() of the states x, the state derivatives ˙x, and the controls u
(Eq. 1). The function f() contained
• identities ˙q− dq
dt = 0 for each degree of freedom (DOF),
• multibody dynamics for each DOF (Eq. S1),
• activation dynamics for each muscle tendon unit (MTU) (Eq. S2), and
• contraction dynamics for each MTU (Eq. S3).
S2.1 Multibody Dynamics
The multibody dynamics were defined as follows:
M(q) ¨q + C(q, ˙q) ˙q + G(q) − JcT Fc − τ = 0,
(S1)
where q contained the global position, the global orientation, and the joint angles, ˙q contained the global velocities and joint
angular velocities, M(q) was the mass matrix, C(q, ˙q) contained the Coriolis forces, G(q) contained the gravity forces, and Jc
was the Jacobian of the contact forces Fc. τ was the sum of active joint torques generated by the muscles τmus (Eq. S12),
passive joint torques τpas (Eq. S14), and joint torques due to external actuation torques τext (Eq. S15). The talus was assumed
to be weightless to save computation within the multibody dynamics. This assumption can be made since the talus is a small
body which cannot move independently from the other segments since no MTU is connected to it.
S2.2 Muscle Dynamics
The model was operated using 92 MTUs (see Table S1). MTUs were modeled as three element Hill-type muscles with a
contractile element (CE), a parallel elastic element (PEE), and a series elastic element (SEE). The muscle dynamics were
described as a function of the activation state a and the CE length state s:
˙a − r(ne) (ne − a) = 0,
(S2)
FSEE(lMTU(q),s) − (FCE(a,s, ˙s) + FPEE(s)) cos(φ(s)) = 0,
(S3)
where the state variable s = lCE cos(φ) denoted the projection of CE length on the muscle line of action for a specific pennation
angle φ 3, ˙a denoted the time derivative of the activation a, ne denoted the neural excitation, and r(ne) denoted the activation
dynamics, which were determined as follows:
r(ne) =
ne
Tact
+ 1 − ne
Tdeact
.
(S4)
The activation time constant Tact = 10ms and the deactivation time constantand Tdeact = 40ms were identically to Hamner’s
model1. In Eq. S3, FSEE denoted the force in the SEE, FCE denoted the force in the CE, and FPEE was the force in the PEE.
The force in the CE was determined by
FCE = a fFL(lCE) fFV(vCE) FISO ,
(S5)
where FISO was the maximum isometric force, fFL(lCE) denoted the force-length relationship, and fFV(vCE) the force-velocity
relationship. The force-length relationship was described as
fFL(lCE) = exp
−
lCE − 1
w
2!
,
(S6)
where lCE was the length of the CE normalized to the optimal fibre length lCE,opt, w a width parameter of the muscle, equal to
the square root of the shape factor used by Thelen4. The force-velocity relationship was described as
fFV(vCE) =
(
λ vCE,max + vCE
λ vCE,max − vCE / A
if vCE < 0
gmax vCE + λ cFV
vCE + λ cFV
if vCE ≥ 0,
(S7)
where vCE was the normalized CE velocity, λ = 0.5025 + 0.5341a5 modeled the activation dependence of the normalized
maximum shortening velocity vCE,max = 10lCE,opt/s4. gmax = 1.8 was the maximum force amplification during lengthening4,
A = 0.25 was the Hill curve parameter6, and cFV was determined as follows7:
cFV = vCE,max A (gmax − 1)
A + 1
.
(S8)
The force in the SEE and PEE were modeled as non-linear springs7:
F(l) =
(
FISO k1 (l − lslack)
if l ≤ lslack
FISO k1 (l − lslack) + FISO k2 (l − lslack)2
if l > lslack ,
(S9)
where l denoted the length of the element normalized to optimal fiber length lCE,opt. It was equal to lMTU(q) − lCE for the
SEE, and equal to s for the PEE. lslack was the slack length normalized to optimal fiber length lCE,opt, k1 = 1 was a small linear
stiffness, which was added to aid the optimization such that the derivative with respect to the model states was never zero, and
k2 was the non-linear stiffness, which was equal to the following for the PEE and SEE, respectively:
k2,PEE =
1
w2 ,
(S10)
k2,SEE =
1
(h lslack,SEE)2 ,
(S11)
where h = 0.04 was the strain of the muscle at isometric force8,9.
2/8
Table S1. Abbreviations of muscles used in the model “runMaD”.
Abbreviation
Muscle Name
glut_min/med/max
gluteus minimus/medius/maximus
semimem
semimembranosus
semiten
semitendinosus
bifemsh/bifemlh
biceps femoris (short/long head)
sar
sartorius
add_long/brev/mag
adductor longus/brevis/magnus
tfl
tensor fascia latae
pect
pectineus
grac
gracilis
iliacus
iliacus
psoas
psoas
quad_fem
quadratus femoris
gem
gemellus
peri
piriformis
rect_fem
rectus femoris
vas_med/int/lat
vastus medialis/intermedius/lateralis
med/lat_gas
gastrocnemius (medial/lateral head)
soleus
soleus
tip_post/ant
tibialis posterior/anterior
flex_dig/hal
flexor digitorum/hallucis
per_long/brev/tert
peroneus longus/brevis/tertius
ext_dig/hal
extensor digitorum/hallucis
ercspn
erector spinae
intobl/extobl
internal/external abdominal oblique
S2.3 Muscle-Joint Coupling
The torque in DOF j generated by muscle i was determined using the following equation:
τmus,i,j = − ∂ lMTU,i(q)
∂ qj
FSEE,i (lMTU,i(q),lCE,i),
(S12)
where lMTU,i(q) denoted the normalized muscle-tendon length depending on the current pose defined by the generalized
coordinates q, and ∂ lMTU,i(q)
∂ qj
denoted the muscle moment arm. A constant muscle moment arm would not have been accurate
enough in the 3D model3. Hence, a polynomial function was fitted to describe the muscle-tendon length depending on the joint
angles to match the moment arms available in OpenSim, since polynomials have well defined derivatives. The order of the
polynomial was chosen such that the root mean square error in the moment arms was less than 5% of the maximum moment
arm for each muscle. The maximum possible order was set to four. Additionally, the range of motion used to determine the
polynomial was reduced if the muscles wrapped around the bones incorrectly for large joint angles. Linear interpolation was
used outside of this range of motion. The following ranges were used: hip flexion [−28◦,78◦], hip adduction [−13◦,13◦], hip
rotation [−3◦,18◦], knee angle [−118◦,8◦], ankle angle [−38◦,38◦], subtalar angle [−13◦,13◦], mtp angle [−8◦,48◦], lumbar
extension [−38◦,3◦], lumbar bending [−8◦,8◦], and lumbar rotation [−18◦,18◦].
S2.4 Passive Joint Torques
Passive torques were added to the joint torques when the joint angle was outside of the normal range of motion. These torques
were determined as follows:
τpas,out, j =
(
K2 (qj −qj,min)2
if q j < q j,min
−K2 (qj −qj,max)2
if q j > q j,max ,
(S13)
where K2 = 5000Nmrad−2. The range of motions, defined by qj,min and qj,max for the trunk and legs are the same as those
used for the muscle moment arms. For the arms, the following ranges were used: arm flexion [−88◦,88◦], arm adduction
[−118◦,88◦], arm rotation [−88◦,88◦], elbow flexion [2◦,148◦], and pronation/supination [2◦,148◦]. For numerical reasons, a
small stiffness was used for the full range of motion, such that the derivative of the joint moment with respect to the joint angle
was never zero:
τpas, j = τpas,out, j −K1 (qj − q j,neutral) − B ˙qj ,
(S14)
3/8
where the stiffness was K1 = 1Nmrad−1 and the damping was equal to B = 1Nmsrad−1. qj,neutral was the neutral position of
DOF j defined as default values in the OpenSim model file runMaD.osim (see supplementary material at www.simtk.org).
S2.5 Arm Torques
The DOF j of a arm was directly actuated by the torque
τext,j = m j 10Nm,
(S15)
with torque control m j. For numerical reasons, a scaling was performed to obtain states and controls of same magnitude.
S2.6 Penetration-Based Ground Contact Model
Eight contact points were used at each foot to describe the contact with the ground. Four contact points were located at the toe
segment and four at the calcaneus segment. In the OpenSim model, their location relative to the respective segment origin
was defined using marker objects (see runMaD.osim). The vertical ground reaction force (GRF) in each contact point c was
determined based on the ground penetration d:
Fc,y(d) = k d (1 − b ˙pc,y),
(S16)
where k = 100BWm−1 was the stiffness of the ground normalized to body weight (BW) per meter. The damping constant
b = 0.75sm−110 was multiplied with the vertical velocity of the contact point ˙pc,y. The ground penetration d was determined
from the vertical position of the contact point pc,y and the size of the transition region pc,y,0 = 10−3 m between contact and no
contact:
d = 1
2
q
p2c,y + p2
c,y,0 − pc,y
.
(S17)
The horizontal GRFs in x- and z-directions were determined using a continuous approximation of the Coulomb friction:
Fc,x(Fc,y, ˙pc,x) = − µk Fc,y
˙pc,x
q
˙p2c,x + ˙p2
c,x,0
,
(S18)
Fc,z(Fc,y, ˙pc,z) = − µk Fc,y
˙pc,z
q
˙p2c,z + ˙p2
c,z,0
,
(S19)
where µk = 1 was the kinetic friction coefficient, ˙pc,x and ˙pc,z were the sliding velocities of the contact point, and ˙pc,x,0 =
˙pc,z,0 = 10−2 ms−1 was a small velocity parameter that ensured that the force was differentiable around zero velocity.
S3 Simulations
The bounds of states x and controls u used in the three optimization examples are summarized in Table S2. For standing,
smaller ranges were chosen compared to running to avoid the simulation terminating in local optima.
S4 Results
Pelvis translation and joint moments of straight and curved running are shown in Figs. S1 and S2.
4/8
Table S2. Bounds used to simulate standing, straight running, and curved running. For straight running, the pelvis position at
the first node was fixed to qpel_tx[0] = 0 and qpel_tz[0] = 0. For curved running, the pelvis position at the first node was fixed to
qpel_tx[0] = −r and qpel_tz[0] = 0. ∆t denotes the duration between two collocation nodes.
Parameter
Unit
Standing
Running
Lower
Upper
Lower
Upper
qpelvis_rotation
degree
0
0
-90
90
qpelvis_obliquity
degree
-5
5
-90
90
qpelvis_tilt
degree
-30
30
-90
90
qpel_tx
m
0
0
-5
7
qpel_ty
m
0.5
1.5
-1
2
qpel_tz
m
0
0
-3
3
qhip_flexion
degree
-30
30
-120
120
qhip_adduction
degree
-10
10
-120
120
qhip_rotation
degree
-30
30
-120
120
qknee_angle
degree
-30
10
-160
10
qankle_angle
degree
-30
30
-90
90
qsubtalar_angle
degree
-30
30
-90
90
qmtp_angle
degree
-30
30
-90
90
qlumbar_extension
degree
-10
10
-90
90
qlumbar_bending
degree
-10
10
-90
90
qlumbar_rotation
degree
-10
10
-90
90
qarm_flex
degree
-40
40
-40
40
qarm_add
degree
-40
40
-40
40
qarm_rot
degree
-40
40
-40
40
qelbow_flex
degree
0
150
0
150
qpro_sup
degree
0
150
0
150
˙q
rads−1 or ms−1
-30
30
-30
30
s
-
0
5
0
5
a
-
0
5
0
5
ne
-
0
5
0
5
m
-
-5
5
-5
5
5/8
0
100
0
3.5
Translation in m
pelvis_tx
0
100
0.95
1.15
pelvis_ty
0
100
−0.1
0.1
pelvis_tz
0
100
−250
150
Moment in Nm
hip_flexion
0
100
−250
100
hip_adduction
0
100
−70
70
hip_rotation
0
100
−150
400
knee_angle
0
100
−300
100
Moment in Nm
ankle_angle
0
100
−40
120
subtalar_angle
0
100
−30
30
mtp_angle
0
100
−100
300
lumbar_extension
0
100
−250
150
Moment in Nm
lumbar_bending
0
100
−100
100
lumbar_rotation
0
100
−30
30
arm_flex
0
100
−30
30
Gait Cycle in %
arm_add
0
100
−30
30
Gait Cycle in %
Moment in Nm
arm_rot
0
100
−30
30
Gait Cycle in %
elbow_flex
0
100
−30
30
Gait Cycle in %
pro_sup
Simulated
Measured: Mean ± SD
Torso
Right side
Left side
Figure S1. Pelvis translation and joint moments of the straight running simulation. The degrees of freedom (DOFs) are
named according to their definition in the model file runMaD.osim. Black, red, and blue solid lines indicate the simulated
variables of the torso, the right side, and left side, respectively. Shaded areas show mean ± standard deviation (SD) of inverse
dynamics (ID) of the measured gait cycles of straight running.
6/8
0
100
−4
3
Translation in m
pelvis_tx
0
100
0.9
1.1
pelvis_ty
0
100
0
2
pelvis_tz
0
100
−250
150
Moment in Nm
hip_flexion
0
100
−250
100
hip_adduction
0
100
−70
70
hip_rotation
0
100
−150
400
knee_angle
0
100
−300
100
Moment in Nm
ankle_angle
0
100
−40
120
subtalar_angle
0
100
−30
30
mtp_angle
0
100
−100
300
lumbar_extension
0
100
−250
150
Moment in Nm
lumbar_bending
0
100
−100
100
lumbar_rotation
0
100
−30
30
arm_flex
0
100
−30
30
Gait Cycle in %
arm_add
0
100
−30
30
Gait Cycle in %
Moment in Nm
arm_rot
0
100
−30
30
Gait Cycle in %
elbow_flex
0
100
−30
30
Gait Cycle in %
pro_sup
Simulated
Measured: Mean ± SD
Torso
Right side
Left side
Figure S2. Pelvis translation and joint moments of the curved running simulation. The degrees of freedom (DOFs) are named
according to their definition in the model file runMaD.osim. Black, red, and blue solid lines indicate the simulated variables of
the torso, the right side, and left side, respectively. Shaded areas show mean ± standard deviation (SD) of inverse
dynamics (ID) of the measured gait cycles of curved running. The horizontal pelvis translation cannot be directly compared to
the measured data since the global frames were not aligned but rotated around the vertical axis.
7/8
References
1. Hamner, S., Seth, A. & Delp, S. L. Muscle contributions to propulsion and support during running. J. Biomech. 43,
2709–2716 (2010).
2. Baker, R. Pelvic angles: a mathematically rigorous definition which is consistent with a conventional clinical understanding
of the terms. Gait & Posture 13, 1–6 (2001).
3. Van den Bogert, A. J., Blana, D. & Heinrich, D. Implicit methods for efficient musculoskeletal simulation and optimal
control. Procedia IUTAM 2, 297–316 (2011).
4. Thelen, D. G. Adjustment of muscle mechanics model parameters to simulate dynamic contractions in older adults. J.
Biomech. Eng. 125, 70 (2003).
5. Chow, J. W. & Darling, W. G. The maximum shortening velocity of muscle should be scaled with activation. J. Appl.
Physiol. 86, 1025–1031 (1999).
6. Winters, J. M. An improved muscle-reflex actuator for use in large-scale neuromusculoskeletal models. Annals Biomed.
Eng. 23, 359–374 (1995).
7. McLean, S. G., Su, A. & Van den Bogert, A. J. Development and validation of a 3-D model to predict knee joint loading
during dynamic movement. J. biomechanical engineering 125, 864–874 (2003).
8. Van Soest, A. J. & Bobbert, M. F. The contribution of muscle properties in the control of explosive movements. Biol.
cybernetics 69, 195–204 (1993).
9. Miller, R. H. Hill-based muscle modeling. In Handbook of Human Motion, 373–394 (Springer, 2018).
10. Gerritsen, K. G., van den Bogert, A. J. & Nigg, B. M. Direct dynamics simulation of the impact phase in heel-toe running.
J. Biomech. 28, 661–668 (1995).
8/8
| Efficient trajectory optimization for curved running using a 3D musculoskeletal model with implicit dynamics. | 10-19-2020 | Nitschke, Marlies,Dorschky, Eva,Heinrich, Dieter,Schlarb, Heiko,Eskofier, Bjoern M,Koelewijn, Anne D,van den Bogert, Antonie J | eng |
PMC8307654 | International Journal of
Environmental Research
and Public Health
Article
Motion Analysis of Match Play in U14 Male Soccer Players and
the Influence of Position, Competitive Level and
Contextual Variables
Erling Algroy 1,*, Halvard Grendstad 2, Amund Riiser 3, Tone Nybakken 2, Atle Hole Saeterbakken 3
,
Vidar Andersen 3 and Hilde Stokvold Gundersen 2
Citation: Algroy, E.; Grendstad, H.;
Riiser, A.; Nybakken, T.;
Saeterbakken, A.H.; Andersen, V.;
Gundersen, H.S. Motion Analysis of
Match Play in U14 Male Soccer
Players and the Influence of Position,
Competitive Level and Contextual
Variables. Int. J. Environ. Res. Public
Health 2021, 18, 7287. https://
doi.org/10.3390/ijerph18147287
Academic Editors: Filipe
Manuel Clemente, Ana Filipa Silva
and Daniele Conte
Received: 4 June 2021
Accepted: 5 July 2021
Published: 7 July 2021
Publisher’s Note: MDPI stays neutral
with regard to jurisdictional claims in
published maps and institutional affil-
iations.
Copyright: © 2021 by the authors.
Licensee MDPI, Basel, Switzerland.
This article is an open access article
distributed
under
the
terms
and
conditions of the Creative Commons
Attribution (CC BY) license (https://
creativecommons.org/licenses/by/
4.0/).
1
Campus Bergen, NLA University College, 5812 Bergen, Norway
2
Department of Sport, Food and Natural Sciences, Campus Bergen, Western Norway University of
Applied Sciences, 5020 Bergen, Norway; [email protected] (H.G.); [email protected] (T.N.);
[email protected] (H.S.G.)
3
Department of Sport, Food and Natural Sciences, Campus Sogndal, Western Norway University of
Applied Sciences, 6851 Sogndal, Norway; [email protected] (A.R.); [email protected] (A.H.S.);
[email protected] (V.A.)
*
Correspondence: [email protected]
Abstract: This study aimed to investigate match running performance in U14 male soccer players
in Norway, and the influence of position, competitive level and contextual factors on running
performance. Locomotion was monitored in 64 different U14 players during 23 official matches.
Matches were played at two different competitive levels: U14 elite level (n = 7) and U14 sub-elite level
(n = 16). The inclusion criterion was completed match halves played in the same playing position.
The variables’ influence on match running performance was assessed using mixed-effect models,
pairwise comparisons with Bonferroni correction, and effect size. The results showed that the U14
players, on average, moved 7645 ± 840 m during a match, of which 1730 ± 681 m (22.6%) included
high-intensity running (HIR, 13.5–18.5 km·h−1) and sprinting (>18.5 km·h−1). Wide midfielders
(WM) and fullbacks (FB) covered the greatest sprint distance (569 ± 40 m) and, in addition to the
centre midfield position (CM), also covered the greatest total distance (TD) (8014 ± 140 m) and
HIR distance (1446 ± 64 m). Centre forwards (CF) performed significantly more accelerations
(49.5 ± 3.8) compared other positions. TD (7952 ± 120 m vs. 7590 ± 94 m) and HIR (1432 ± 57 m vs.
1236 ± 43 m) were greater in U14 elite-level matches compared with sub-elite matches. Greater TD
and sprint distances were performed in home matches, but, on the other hand, more accelerations
and decelerations were performed in matches played away or in neutral locations. Significantly
higher TD, HIR and sprinting distances were also found in lost or drawn matches. In conclusion,
physical performance during matches is highly related to playing position, and wide positions seem
to be the most physically demanding. Further, competitive level and contextual match variables are
associated with players’ running performance.
Keywords: match running performance; youth soccer; positional differences; competitive levels;
contextual variables
1. Introduction
In the last decade, several studies have described the physical demands of youth soccer [1].
Previous reports have shown that U14 players cover a total distance between 105–115 m·min−1
during a match and 11.5–14.5 m·min−1 at higher speed (13.1–19 km·h−1) [2,3] Additionally,
elite U14 players perform about 0.4 ± 0.2 sprints (speed above 25.2 km·h−1) [2] and
1.82 ± 0.33 accelerations per min during a match [2,4]. Although there is a growing body
of evidence related to the physical match characteristics of youth soccer players, very
few studies have examined running performance at the U14 level specifically in relation
to high-explosive actions like sprints, accelerations and decelerations [1,5]. Due to the
Int. J. Environ. Res. Public Health 2021, 18, 7287. https://doi.org/10.3390/ijerph18147287
https://www.mdpi.com/journal/ijerph
Int. J. Environ. Res. Public Health 2021, 18, 7287
2 of 9
increasing physical match demands with age, more knowledge about performance at
specific age levels is necessary to gain insights into the prerequisites of competing at U14
level. Additionally, the influence of competing standard and positional roles on players’
running performance seems to be an important factor in youth soccer [1] and must be
accounted for.
Although the scientific evidence is scarce, some studies have examined match running
characteristics among youth players. These have reported that U13–U18 players’ physical
match characteristics are affected by positional demands, as strikers and wide midfielders
demonstrated the highest peak game speeds and frequency of high-intensity actions [5,6].
Additionally, centre midfielders have been reported to cover the highest total distance [5,6],
whereas centre backs covered the lowest total distance [7] and the lowest amount of
high-intensity actions [6].
A current limitation in research on youth soccer is the lack of information regarding
the influence of contextual variables like match results, location and match status [1].
One previous study has shown that U14 elite level players outperform non-elite players
regarding match running performance [2], suggesting a higher external demand during
elite matches [7,8]. Studies from senior soccer have also reported greater match running
performance in games lost compared with games won [9].
To date, no study has examined the match performance characteristics of U14 soccer
players from Norway or other Scandinavian countries. Comparing data from different
countries and regions of the world would be important to enhance our understanding of
various approaches to and philosophies of talent development. Especially, more research
regarding accelerations and decelerations is necessary in youth soccer. These high-intensity
actions have been found to be crucial determinants for successful performance and to
discriminate high- and low-level adult players [10]. Accelerations and decelerations also
require high rates of force development and are therefore related to the total match load [11].
Additionally, studies on U14 players have, to the best of our knowledge, only been con-
ducted in the United Kingdom, Qatar and New Zealand [1]. Hence, the first aim of this
study was therefore to analyse match running performance in U14 male soccer players
in Norway. A second aim was to identify how playing position, competitive level and
contextual factors influenced running performance. We hypothesise that players’ playing
positions and competitive level influenced the amount of high-intensity actions in U14
soccer matches.
2. Materials and Methods
The current study is part of a longitudinal research project, examining factors related
to talent development in youth soccer [12]. The study design in the present study is
descriptive, and match running performance data were collected during the 2018 season,
from April to October. Collection of anthropometric data was performed during 2 different
days for each player in the middle of the 2018 season. Contextual factors investigated in
the present study were match results, match location and match halves.
2.1. Participants
Sixty-four U14 male outfield players (age: 14.0 ± 0.3 years, height: 166.3 ± 8.8 cm,
weight: 51.9 ± 9.7 kg, body fat: 8.7 ± 3.3%) from 4 clubs with youth soccer academies were
included. Three of the players were U13 players but were included as they played in the
U14 teams investigated. The Regional Committee for Medical and Health Research Ethics
approved the study (2017/1731), which was conducted in accordance with the Helsinki
Declaration. Since players were under the legal age of consent, both the players and their
parents gave written informed consent to participate. All results were treated anonymously.
2.2. Matches
Match running performance was obtained from 23 matches, 15 at the team’s home
location and 5 away and 3 on a neutral field. In 17 of the matches, the teams won; 1 match
Int. J. Environ. Res. Public Health 2021, 18, 7287
3 of 9
ended with a draw and 5 with a loss. All matches were official league or cup matches,
played outdoor on regular-sized synthetic grass soccer fields with 11 players per side.
Matches were played at 2 different competitive levels: U14 elite, national level (7 matches,
93 match halves), U14 sub-elite, the highest regional level (16 matches, 249 match halves).
Players playing in elite teams are recruited into playing at professional youth academies
and matches are played in the U14 national level league, in which youth academies from
professional Norwegian clubs participate. Some players played matches at more than
1 level.
Match playing time was 2 × 35 min in the U14 local series and 2 × 40 min for the U14
national series. All match performance outcomes were normalised to 35 min of playing
time before the analyses to allow direct comparison between matches at different competing
levels and to be comparable with other studies examining U14 players.
Because of the “rolling substitution” policy resulted in a large variation in playing
time between players, all match halves completed by 1 player in the same playing position
were included in the dataset. The dataset comprised 342 match halves, including 187 first
halves and 155 second halves; 278 of the included halves were from complete matches
played by 1 player.
2.3. GPS Tracking
During the matches, the players wore a portable and previously validated [13] GPS
device (Apex, STATSsport, Newry, UK) monitoring their motions and position with a
sampling frequency of 18 Hz. The players wore, in most cases, the same unit in every game
to limit inter-unit reliability issues, although the present GPS system has been shown to
have excellent inter-unit reliability [14]. The units were placed in vests located between the
players’ scapulae. The raw GPS data were synchronised to the start and end of each half
match period and exported for further analysis.
2.4. Match Running Categories
Match activities were divided into different speed categories: walking (0.1–4.5 km·h−1),
low-intensity running (LIR, 4.5–8.5 km·h−1), medium-intensity running (MIR, 8.6–13.5 km·h−1),
high-intensity running (HIR, 13.6–18.5 km·h−1), sprinting (>18.5 km·h−1) and maximal
speed. The speed thresholds were age-specific [3] and adapted from previous studies on
U14 soccer players [3,15] to compare running performance with previously reports on
U14 soccer players. In addition, sprint distance with a threshold adapted from studies on
senior soccer players (25.2 km·h−1 with a duration of at least 1 s) was included to compare
sprint distance between U14 players and senior elite players [16]. Total distance included
all locomotion during the match. Data are presented both in absolute (m) and relative
(m·min−1) distances.
The number of accelerations and decelerations was also examined, and were defined
as actions when speed was increased or decreased by more than 3 m·s−2 and lasted more
than 0.5 s. Only accelerations and decelerations at speeds above 13.5 km·h−1 were included.
2.5. Playing Position
Players were categorised as centre backs (CB, n = 70 match halves), fullbacks (FB,
n = 68 match halves), centre midfielders (CM, n = 100 match halves), wide midfielders
(WM, n = 65 match halves) or centre forwards (CF, n = 39 match halves). Players swapping
positions were excluded from the analysis. Goalkeepers were excluded from the study due
to their low running demands [16,17].
2.6. Statistical Analyses
Data are presented as means with the standard deviation (SD) or 95% confidence inter-
val (CI). Visual inspection confirmed that all data were normally distributed. Differences in
match running performance between halves were assessed by one-way ANOVA analysis.
Int. J. Environ. Res. Public Health 2021, 18, 7287
4 of 9
The fixed effects (E) of the independent variables on match running performance
were assessed using mixed-effect models with the player as a random effect to adjust for
multiple match observations by the same player. All independent variables were included
in the models, as they may influence match running performance in theory. The variables
analysed were playing position (5 positions), competitive level (elite vs. sub-elite), match
location (home vs. neutral/away), match results (win vs. draw/loss) and match half
(first vs. second half). The collinearity between the continuous independent variables
was inspected by Pearson’s product moment correlations coefficients and were included if
r < 0.6. Pairwise comparisons for competition level and playing position were performed
with Bonferroni correction. IBM SPSS Statistics (version 26, IBM, Armonk, NY, USA) was
used for all statistical analyses. Significance for all analyses was accepted at p ≤ 0.05. When
significant differences were detected, Cohen’s d effect size (ES) was calculated. An ES of
0–0.2 was considered trivial, >0.2 as small, >0.5 as medium and >0.8 as large [18].
3. Results
3.1. Match Running Performance
During a match, players covered, on average, 22% of the total distance by high-
intensity running (13.6–18.5 km·h−1) and sprinting (>18.5 km·h−1), 32% by medium-
intensity running (8.6–13.5 km·h−1), 32% by low-intensity running (4.6–8.5 km·h−1) and
16% by walking (0.1–4.5 km·h−1). More distance was covered in the first compared with
the second half, both in metres (ES = 0.71) and metres pr minute (ES = 0.29). Total distance
and high-intensity activities for each half are presented in Table 1.
Table 1. An overview of total match distance covered and high-intensity actions in Norwegian U14 players. Data are
presented as means ± SD. Match playing time was 2 × 35 min. One-way ANOVA was used when comparing match halves
independently of whether the same players played one or both halves.
First Half (n = 187)
Second Half (n = 155)
Full Match
p-Values
ES Values
Total distance
m
3883 ± 405
3762 ± 435
7645 ± 840
0.008
0.71
m·min−1
111.0 ± 11.6
107.5 ± 12.4
109.3 ± 12.0
0.008
0.29
HIR (13.6–18.5 km·h−1)
m
618 ± 185
600 ± 199
1218 ± 398
0.365
0.09
m·min−1
17.7 ± 5.3
17.1 ± 5.7
17.4 ± 5.5
0.365
0.11
Sprint distance
>18.5 km·h−1 (m) a
242 ± 105
229 ± 116
471 ± 221
0.260
0.12
>25.2 km·h−1 (m) b
19.8 ± 31.2
20.8 ± 31.1
40.6 ± 62.3
0.767
0.03
Maximal speed (km·h−1)
27.0 ± 2.2
26.6 ± 2.2
27.0 ± 2.2
0.061
0.18
Accelerations (n)
18.9 ± 9.4
19.7 ± 8.9
38.6 ± 18.3
0.419
0.09
Declarations (n)
23.1± 11.3
24.7 ± 10.9
47.8 ± 22.2
0.194
0.14
HIR: high-speed running. a threshold adapted from previous studies on U14 soccer players, b above 25.2 km·h−1 for more than 1 s
(threshold adapted from studies on senior soccer players). Accelerations/decelerations: actions (>13.6 km·h−1) when speed was increased
or decreased by more than 3 m·s−2 and lasted more than 0.5 s. Data were not corrected for multiple match observations by the same player.
p ≤ 0.05 is statistically significant. ES: 0–0.2 = trivial; >0.2 = small; >0.5 = medium; >0.8 = large.
3.2. Playing Position
Playing positions were associated with TD and all high-intensity parameters inves-
tigated (HIR, sprints, accelerations and decelerations). FB, WM and CM showed signifi-
cantly greater total distance and HIR distance compared with CB (ESTD = 0.62–1.05 and
ESHIR = 0.91–3.06) and CF (ESTD = 0.62–1.05 and ESHIR = 0.57–2.61). Wide playing posi-
tions (WM, FB) covered the most sprint distance (ES = 0.64–1.08), and CF also covered
greater sprint distance compared with CB positions (ES = 0.90). CF performed significantly
more accelerations compared with all other playing positions (ES = 0.48–0.96). WM and
CF had the highest number of decelerations (ES = 0.38–1.04), and CB performed signifi-
Int. J. Environ. Res. Public Health 2021, 18, 7287
5 of 9
cantly fewer decelerations compared with all other positions (ES = 0.53–1.18). Positional
differences according to running performance are described in Table 2.
Table 2. Positional differences (mean ± 95% CI) in match running performance during one match half (35 min).
Total Distance (m)
HIR (m) (13.6–18.5 km·h−1)
Sprints (m) (>18.5 km·h−1)
Accelerations (n)
Decelerations (n)
FB ‡
3987 [3838, 4136] (‡ > §, ∞) *
722 [653, 790] (‡ > §, ∞) *
281 [240, 322] (‡ > §, ¶) *
19.3 [16.3, 22.3]
25.2 [21.7, 28.6] (‡ > §) *
WM †
4015 [3883, 4148] († > §, ∞) *
712 [651, 774] († > §, ∞) *
288 [251, 326] († > §, ¶) *
20.0 [17.1, 22.9]
27.9 [24.5, 31.3] († > §, ¶) *
CB §
3673 [3506, 3839]
552 [475, 629]
187 [141, 233]
16.2 [13.1, 19.3]
18.0 [14.5, 21.5]
CM ¶
4020 [3887, 4153] (¶ > §, ∞) *
736 [674, 796] (¶ > §, ∞) *
217 [180, 254]
17.5 [14.9, 20.1]
23.6 [20.5, 26.6] (¶ > §) *
CF ∞
3734 [3551, 3917]
614 [529, 699]
257 [205, 309] (∞ > §) *
24.8 [20.1, 28.6]
(∞ > ‡, †, §, ¶) *
27.6 [23.2, 32.0] (∞ > §) *
HIR: high-intensity running; FB: fullback; WM: wide midfielder; CB: centre back; CM: centre midfielder; CF: centre forward. Accelera-
tions/decelerations: actions (>13.6 km·h−1) when speed was increased or decreased by more than 3 m·s−2 and lasted more than 0.5 s.
Data were corrected for multiple match observations by the same player and all independent variables. Pairwise comparisons for playing
position were performed with Bonferroni correction. * p ≤ 0.05.
3.3. Competitive Level
Players covered significantly greater total distance (ES = 0.35) and HIR distance
(ES = 0.43) (13.6–18.5 km·h−1) in matches played at U14 elite level compared with the local
U14 sub-elite level (Table 3). No differences between groups were found regarding sprints,
accelerations and decelerations.
Table 3. Match running performance (mean ± 95% CI) in one match half (35 min) in U14 players according to competitive
level.
Total Distance (m)
HIR (m) (13.6–18.5 km·h−1)
Sprints (m) (>18.5 km·h−1)
Accelerations (n)
Decelerations (n)
U14 elite matches
3898 [3779, 4016] *
671 [616, 727] *
247 [216, 279]
17.9 [15.1, 20.6]
22.9 [19.6, 26.2]
U14 sub-elite matches
3752 [3659, 3845]
596 [554, 639]
245 [222, 267]
20.4 [18.7, 22.2]
24.4 [22.2, 26.5]
U14 elite: national U14 elite level; U14 sub-elite: highest regional U14 level. HIR: high-intensity running. Accelerations/decelerations:
actions (>13.6 km·h−1) when speed was increased or decreased by more than 3 m·s−2 and lasted more than 0.5 s. Data were corrected for
multiple match observations by the same player. Pairwise comparisons for playing position were performed with Bonferroni correction.
* p ≤ 0.05.
3.4. Contextual Factors
Significant greater total distances (p < 0.001, ES = 0.36), HIR distances (p < 0.001,
ES = 0.32) and sprint distances (p = 0.016, ES = 0.11) were observed in matches lost or
drawn compared with matches won. Match location was also shown to have an impact on
most running variables investigated. Players performed greater TD (p = 0.043, ES = 0.13)
and sprint distance (p = 0.038, ES = 0.20) when playing at home compared with playing
away or at a neutral match location. On the other hand, the number of accelerations
(p < 0.001, ES = 0.56) and decelerations (p < 0.001, ES = 0.46) was higher when playing away
or at a neutral location compared with playing at home. Greater total distances (3883 ± 405
vs. 3762 ± 435, ES = 0.71) covered in one half were observed (p = 0.008), but we found no
other differences between match halves.
4. Discussion
This study aimed to analyse match running performance in U14 male soccer players
in Norway, and to identify how playing position, competitive level and contextual match
variables influenced running performance. Overall, 22% of the locomotion during matches
constituted high-speed running and sprinting, with the rest being performed at lower
running intensities. There were significant differences in physical performance between
positions, in addition to greater running demands when players competed at the highest
competitive standard. Furthermore, contextual factors were associated with most running
variables investigated.
The findings of the present study showed that Norwegian male U14 soccer players
run similar distances or more during a match compared with previous reports on U14
Int. J. Environ. Res. Public Health 2021, 18, 7287
6 of 9
players [1]. The group in our study covered a total distance of 109.2 m·min−1, which is
higher than that of 95.7 m·min−1 reported in New Zealand U14 male soccer players [3]
and 106.5 m·min−1 in English U14 soccer players [19]. Similarly, in our study, 22% of the
total match time was spent as high-intensity running actions (>13.6 km/h−1), whereas
reports from youth soccer in New Zealand [3], Italy [20] and England [19] have shown
total high-intensity running actions to constitute 7%, 16% and 17%, respectively, with a
similar speed threshold applied as in the present study. These results could be explained
by how the different countries approach and emphasise the game, as studies from senior
soccer highlighted that soccer philosophy, technical performance and physical performance
varied between different countries and leagues [21–23].
Across playing positions we found that CF had a greater number of accelerations
compared with all other playing positions. In contrast, a previous study on youth soccer
showed that attackers perform a high amount of accelerations, but that players in wide
positions (WM and FB) in general perform more accelerations than central players (CB, CM
and CF) [24]. However, Vigh-Larsen and colleagues [24] only examined 14 outfield players
from one single team, and their findings could reflect the team’s formation and tactical
dispositions, or the physical capacity of the individual players [25]. Indeed, discrepancies
have been found for high-intensity physical performance between different playing forma-
tions in elite senior soccer [26]. Regardless, our finding that CB perform fewer accelerations
than other positions is in line with previous findings [24,27], and also that CB positions in
general have a lower running performance compared with other playing positions [1]. Our
results showed that wide playing positions performed the highest amount of high-intensity
work and, in general, were the most physically demanding playing positions. This was also
supported by Pettersen and colleagues (2019), who investigated a group of Norwegian U17
soccer players [27]. In senior soccer, studies have demonstrated that evolving tactics over
the last decade have especially impacted the physical demands of wide players, as this
position has shown the greatest increase in high-intensity running [17,28]. Our findings
suggest that the high running demands of wide positions found at higher age groups and
in senior soccer also seems to be evident at U14 level.
Total distance and high-intensity running were significantly higher in matches at U14
elite level compared with sub-elite level, which is in line with findings from a previous
study [2]. However, the observed difference was small regarding effect size, and we found
no differences regarding sprint distance, or the number of accelerations and decelerations.
A possible explanation is that the sub-elite teams represented in our study were the
highest ranked sub-elite teams and included highly talented players also recruited in
the U14 national and regional teams. The observed difference in our study regarding
running performance and competitive level might been greater with a wider range of
teams included from the sub-elite group. Our finding is in contrast to what has previously
been reported in top-ranked U17 players, where the best teams outperformed players in
lower-ranked teams [29]. Contrarily, studies on senior players from teams that finished in
the top five at the end of the English Premier League season performed less sprinting than
those that finished outside of the top five [30].
Match results were associated with running performance. The players covered a
greater total distance, HIR distance and sprint distance in matches they lost or drew com-
pared with the matches they won. Previous studies in elite senior soccer have shown
conflicting results regarding the influence of match results, where Castellano et al. [9]
found no relationship with match results, whereas Lago et al. [31] found that players
ran less at higher intensities when winning. Playing against high-quality opposition has
been found to be associated with lower ball possession [32], and a possible explanation
could be that the lower-quality team has to cover a greater distance to regain possession
and defend important spaces. Our results also showed that match location influenced
match running performance, as greater TD and sprinting distance (>18.5 km·h−1) were
performed when playing at home, even though fewer accelerations and decelerations were
performed. Different tactical approaches to home matches versus away matches may be an
Int. J. Environ. Res. Public Health 2021, 18, 7287
7 of 9
explanation, but more research is necessary to better understand this. A limitation in the
present study according to contextual factors is the preponderance of matches ending with
a win compared with matches ending with a draw/loss. Our results showed a significant
association between contextual factors and several running performance variables; how-
ever, these findings had, in general, a small/medium effect size. More research is necessary
in order to better understand the link between match running performance and contextual
factors in youth soccer. The influence of these multivariate factors should not be considered
in isolation.
A challenge when comparing time motion data from different studies from youth
soccer is the different methodological approach to the rolling substitution policy, resulting
in a large variety in playing time between players. In the present study, all completed
match halves were included to increase the sample size and statistical power, in contrast
to most previous studies that only included completed matches. Excluding players who
did not play full matches may bias analyses by reducing the sample size and removing
variations in running performance. This is important to take into account when interpreting
positional differences, as offensive positions are commonly substituted [33]. We observed a
significantly lower relative total distance covered in the second half (3.9 m·min−1), but the
small difference observed between halves was argued to be of less practical meaning.
Overall, our results show that physical performance during matches is affected by
several factors and differs between leagues and countries. Running demands in soccer and
especially different positional running demands are evolving [28], and highlight the need
for updated research. This is the first study to assess match-related physical performance
and positional differences in Scandinavian U14 youth soccer. Our results suggest that
physical performance during matches is similar or higher than in reports from other parts
of the world and are highly related to playing position. Wide playing positions seem
to be the most physically demanding positions. In addition, centre forwards perform
more accelerations than all other positions. Given the high physical demands of high-
intensity work and the impact of these actions on post-match muscle damage [34], coaches
must consider different positional match loads, and individualise training according to
positional demands. Further, our data indicate that playing at the highest U14 level was
more physically demanding, suggesting a higher external demand during elite matches
compared with sub-elite matches. This reveals important information for practitioners,
as the physical training could be tailored to the game demands, and/or players could
be moved between levels to assist in talent development. Future research should seek to
improve our understanding of positional demands related to different tactics and team
formations, and to investigate training load according to positional demands. More research
is also necessary to better understand the influence of different contextual factors in
youth soccer.
5. Conclusions
Physical performance during matches in this group of Norwegian U14 male soccer
players was similar or higher than reports from other parts of the world and are highly
related to playing position. Total distance and high-intensity running were greater in
elite matches compared with sub-elite matches, but no differences were found regarding
sprints, accelerations and decelerations. Match result and match arena also influenced
running performance.
Author Contributions: Conceptualisation, E.A., H.G. and H.S.G.; formal analysis, E.A., A.R., V.A.
and H.S.G.; investigation, E.A, T.N. and H.S.G.; methodology, E.A., A.R. and H.S.G.; project adminis-
tration, H.S.G.; supervision, H.S.G.; visualisation, E.A.; writing—original draft, E.A., H.G. and H.S.G.;
writing—review and editing, E.A., H.G., A.R., T.N., A.H.S., V.A. and H.S.G. All authors have read
and agreed to the published version of the manuscript.
Funding: This research was given financial support by the University and College Network for
Western Norway.
Int. J. Environ. Res. Public Health 2021, 18, 7287
8 of 9
Institutional Review Board Statement: The study was conducted according to the guidelines of the
Declaration of Helsinki and approved by The Regional Committee for Medical and Health Research
Ethics, Grant number (720025).
Informed Consent Statement: Informed consent was obtained from all subjects involved in the study.
Data Availability Statement: The data presented in this study are available on request from the
corresponding author. The data are not publicly available due to ethical restrictions.
Acknowledgments: We would specifically like to thank all the participants, Knut Kvammen and
Håvard Wiersen for contributing to the data collection, and the University and College Network for
Western Norway for financial support.
Conflicts of Interest: The authors declare no conflict of interest.
References
1.
Vieira, L.H.P.; Carling, C.; Barbieri, F.A.; Aquino, R.; Santiago, P.R.P. Match running performance in young soccer players: A
systematic review. Sports Med. 2019, 49, 289–318. [CrossRef] [PubMed]
2.
Waldron, M.; Murphy, A. A comparison of physical abilities and match performance characteristics among elite and subelite
under-14 soccer players. Pediatr. Exerc. Sci. 2013, 25, 423–434. [CrossRef] [PubMed]
3.
Atan, S.A.; Foskett, A.; Ali, A. Motion Analysis of Match Play in New Zealand U13 to U15 Age-Group Soccer Players. J. Strength
Cond. Res. 2016, 30, 2416–2423. [CrossRef] [PubMed]
4.
Arruda, A.F.; Carling, C.; Zanetti, V.; Aoki, M.S.; Coutts, A.J.; Moreira, A. Effects of a very congested match schedule on body-load
impacts, accelerations, and running measures in youth soccer players. Int. J. Sports Physiol. Perform. 2015, 10, 248–252. [CrossRef]
5.
Al Haddad, H.; Simpson, B.M.; Buchheit, M.; Di Salvo, V.; Mendez-Villanueva, A. Peak match speed and maximal sprinting
speed in young soccer players: Effect of age and playing position. Int. J. Sports Physiol. Perform. 2015, 10, 888–896. [CrossRef]
6.
Buchheit, M.; Mendez-Villanueva, A.; Simpson, B.M.; Bourdon, P.C. Match running performance and fitness in youth soccer. Int.
J. Sports Med. 2010, 31, 818–825. [CrossRef]
7.
Mendez-Villanueva, A.; Buchheit, M.; Simpson, B.; Bourdon, P.C. Match play intensity distribution in youth soccer. Int. J. Sports
Med. 2013, 34, 101–110. [CrossRef]
8.
Saward, C.; Morris, J.; Nevill, M.; Nevill, A.M.; Sunderland, C. Longitudinal development of match-running performance in elite
male youth soccer players. Scand. J. Med. Sci. Sports 2016, 26, 933–942. [CrossRef]
9.
Castellano, J.; Blanco-Villaseñor, A.; Alvarez, D. Contextual variables and time-motion analysis in soccer. Int. J. Sports Med. 2011,
32, 415–421. [CrossRef]
10.
Mohr, M.; Krustrup, P.; Bangsbo, J. Match performance of high-standard soccer players with special reference to development of
fatigue. J. Sports Sci. 2003, 21, 519–528. [CrossRef]
11.
Nedelec, M.; McCall, A.; Carling, C.; Legall, F.; Berthoin, S.; Dupont, G. The influence of soccer playing actions on the recovery
kinetics after a soccer match. J. Strength Cond. Res. 2014, 28, 1517–1523. [CrossRef] [PubMed]
12.
Grendstad, H.; Nilsen, A.K.; Rygh, C.B.; Hafstad, A.; Kristoffersen, M.; Iversen, V.V.; Nybakken, T.; Vestbostad, M.; Algroy, E.A.;
Sandbakk, O.; et al. Physical capacity, not skeletal maturity, distinguishes competitive levels in male Norwegian U14 soccer
players. Scand. J. Med. Sci. Sports 2019, 30, 254–263. [CrossRef] [PubMed]
13.
Beato, M.; Coratella, G.; Stiff, A.; Iacono, A.D. The Validity and Between-Unit Variability of GNSS Units (STATSports Apex 10 and
18 Hz) for Measuring Distance and Peak Speed in Team Sports. Front. Physiol. 2018, 9, 1288. [CrossRef] [PubMed]
14.
Beato, M.; de Keijzer, K.L. The inter-unit and inter-model reliability of GNSS STATSports Apex and Viper units in measuring
peak speed over 5, 10, 15, 20 and 30 meters. Biol. Sport 2019, 36, 317–321. [CrossRef]
15.
Castagna, C.; D’Ottavio, S.; Abt, G. Activity profile of young soccer players during actual match play. J. Strength Cond. Res. 2003,
17, 775–780. [CrossRef]
16.
Ingebrigtsen, J.; Dalen, T.; Hjelde, G.H.; Drust, B.; Wisløff, U. Acceleration and sprint profiles of a professional elite football team
in match play. Eur. J. Sport Sci. 2015, 15, 101–110. [CrossRef]
17.
Bradley, P.S.; Sheldon, W.; Wooster, B.; Olsen, P.; Boanas, P.; Krustrup, P. High-intensity running in English FA Premier League
soccer matches. J. Sports Sci. 2009, 27, 159–168. [CrossRef]
18.
Cohen, J. Statistical Power Analysis for the Behavioral Sciences; Erlbaum: Hillsdle, MI, USA, 1988.
19.
Harley, J.A.; Barnes, C.A.; Portas, M.; Lovell, R.; Barrett, S.; Paul, D.; Weston, M. Motion analysis of match-play in elite U12 to U16
age-group soccer players. J. Sports Sci. 2010, 28, 1391–1397. [CrossRef]
20.
Castagna, C.; Impellizzeri, F.; Cecchini, E.; Rampinini, E.; Alvarez, J.C. Effects of intermittent-endurance fitness on match
performance in young male soccer players. J. Strength Cond. Res. 2009, 23, 1954–1959. [CrossRef]
21.
Yi, Q.; Groom, R.; Dai, C.; Liu, H.; Gómez Ruano, M.Á. Differences in Technical Performance of Players From ‘The Big Five’
European Football Leagues in the UEFA Champions League. Front. Psychol. 2019, 10, 2738. [CrossRef] [PubMed]
22.
Sarmento, H.; Pereira, A.; Matos, N.; Campaniço, J.; Anguera, T.M.; Leitão, J. English Premier League, Spai´ns La Liga and Italýs
Seriés A–What’s Different? Int. J. Perform. Anal. Sport 2013, 13, 773–789. [CrossRef]
Int. J. Environ. Res. Public Health 2021, 18, 7287
9 of 9
23.
Dellal, A.; Chamari, K.; Wong, D.P.; Ahmaidi, S.; Keller, D.; Barros, R.; Bisciotti, G.N.; Carling, C. Comparison of physical
and technical performance in European soccer match-play: FA Premier League and La Liga. Eur. J. Sport Sci. 2011, 11, 51–59.
[CrossRef]
24.
Vigh-Larsen, J.F.; Dalgas, U.; Andersen, T.B. Position-Specific Acceleration and Deceleration Profiles in Elite Youth and Senior
Soccer Players. J. Strength Cond. Res. 2018, 32, 1114–1122. [CrossRef] [PubMed]
25.
Buchheit, M.; Simpson, B.M.; Mendez-Villanueva, A. Repeated high-speed activities during youth soccer games in relation to
changes in maximal sprinting and aerobic speeds. Int. J. Sports Med. 2013, 34, 40–48. [CrossRef]
26.
Bradley, P.S.; Carling, C.; Archer, D.; Roberts, J.; Dodds, A.; Di Mascio, M.; Paul, D.; Diaz, A.G.; Peart, D.; Krustrup, P. The effect of
playing formation on high-intensity running and technical profiles in English FA Premier League soccer matches. J. Sports Sci.
2011, 29, 821–830. [CrossRef]
27.
Pettersen, S.A.; Brenn, T. Activity Profiles by Position in Youth Elite Soccer Players in Official Matches. Sports Med. Int. Open 2019,
3, E19–E24. [CrossRef]
28.
Bush, M.; Barnes, C.; Archer, D.T.; Hogg, B.; Bradley, P.S. Evolution of match performance parameters for various playing
positions in the English Premier League. Hum. Mov. Sci. 2015, 39, 1–11. [CrossRef]
29.
Varley, M.C.; Gregson, W.; McMillan, K.; Bonanno, D.; Stafford, K.; Modonutti, M.; Di Salvo, V. Physical and technical performance
of elite youth soccer players during international tournaments: Influence of playing position and team success and opponent
quality. Sci. Med. Footb. 2017, 1, 18–29. [CrossRef]
30.
Di Salvo, V.; Gregson, W.; Atkinson, G.; Tordoff, P.; Drust, B. Analysis of high intensity activity in Premier League soccer. Int. J.
Sports Med. 2009, 30, 205–212. [CrossRef]
31.
Lago, C.; Casais, L.; Dominguez, E.; Sampaio, J. The effects of situational variables on distance covered at various speeds in elite
soccer. Eur. J. Sport Sci. 2010, 10, 103–109. [CrossRef]
32.
Lago, C. The influence of match location, quality of opposition, and match status on possession strategies in professional
association football. J. Sports Sci. 2009, 27, 1463–1469. [CrossRef] [PubMed]
33.
Bradley, P.S.; Lago-Penas, C.; Rey, E.; Sampaio, J. The influence of situational variables on ball possession in the English Premier
League. J. Sports Sci. 2014, 32, 1867–1873. [CrossRef] [PubMed]
34.
De Hoyo, M.; Cohen, D.D.; Sañudo, B.; Carrasco, L.; Álvarez-Mesa, A.; Del Ojo, J.J.; Domínguez-Cobo, S.; Mañas, V.; Otero-
Esquina, C. Influence of football match time–motion parameters on recovery time course of muscle damage and jump ability. J.
Sports Sci. 2016, 34, 1363–1370. [CrossRef] [PubMed]
| Motion Analysis of Match Play in U14 Male Soccer Players and the Influence of Position, Competitive Level and Contextual Variables. | 07-07-2021 | Algroy, Erling,Grendstad, Halvard,Riiser, Amund,Nybakken, Tone,Saeterbakken, Atle Hole,Andersen, Vidar,Gundersen, Hilde Stokvold | eng |
PMC8296310 | International Journal of
Environmental Research
and Public Health
Review
Periodization and Programming for Individual 400 m
Medley Swimmers
Francisco Hermosilla 1,2
, José M. González-Rave 1,*
, José Antonio Del Castillo 3 and David B. Pyne 4
Citation: Hermosilla, F.; González-
Rave, J.M.; Del Castillo, J.A.; Pyne,
D.B. Periodization and Programming
for Individual 400 m Medley
Swimmers. Int. J. Environ. Res. Public
Health 2021, 18, 6474. https://
doi.org/10.3390/ijerph18126474
Academic Editors: Matteo Cortesi,
Sandro Bartolomei, Giorgio Gatta and
Tomohiro Gonjo
Received: 12 May 2021
Accepted: 12 June 2021
Published: 15 June 2021
Publisher’s Note: MDPI stays neutral
with regard to jurisdictional claims in
published maps and institutional affil-
iations.
Copyright: © 2021 by the authors.
Licensee MDPI, Basel, Switzerland.
This article is an open access article
distributed
under
the
terms
and
conditions of the Creative Commons
Attribution (CC BY) license (https://
creativecommons.org/licenses/by/
4.0/).
1
Sport Training Lab, University of Castilla-La Mancha, 45008 Toledo, Spain; [email protected]
2
Facultad de Ciencias de la Vida y la Naturaleza, Universidad Nebrija, 28248 Madrid, Spain
3
Catalonian Swimming Federation and High Performance Center, Alcalde Barnils, Av. 3-5,
Sant Cugat del Vallès, 08174 Barcelona, Spain; [email protected]
4
Research Institute for Sport and Exercise, Faculty of Health, University of Canberra,
Bruce, ACT 261, Australia; [email protected]
*
Correspondence: [email protected]; Tel.: +34-666160346
Abstract: Knowledge in the scientific domain of individual medley (IM) swimming training over
a competitive season is limited. The purpose of this study was to propose a detailed coaching
framework incorporating the key elements of a periodized training regimen for a 400 m IM swimmer.
This framework was based on the available coaching and scientific literature and the practical
experience and expertise of the collaborating authors. The season has been divided in two or three
macrocycles, further divided in three mesocycles each (six or nine mesocycles in total), in alignment
with the two or three main competitions in each macrocycle. The principal training contents to
develop during the season expressed in blood lactate zones are: aerobic training (~2 mmol·L−1),
lactate threshold pace (~4 mmol·L−1) and VO2max (maximum oxygen uptake) (~6 mmol·L−1).
Strength training should focus on maximum strength, power and speed endurance during the
season. Altitude training camps can be placed strategically within the training season to promote
physiological adaptation and improvements in performance. A well-constructed technical framework
will permit development of training strategies for the 400 m IM swimmer to improve both training
and competitive performance.
Keywords: swimming; individual medley; training; season
1. Introduction
Swimming competitions are performed in four major strokes (front crawl, backstroke,
breaststroke and butterfly). The individual medley events (IM) comprise all four swimming
strokes in the following order: butterfly, backstroke, breaststroke and freestyle. The two
variants of the IM are the 200 m IM (50 m of each stroke) and the 400 m IM (100 m of each
stroke). It is necessary to train all four strokes underpinning the widespread assertion
in the high performance swimming community that the IM events are the most complex
and challenging on swimming program [1]. Gonjo and Olstad [2] highlight the lack
of knowledge on comprehensive guidelines for the preparation of high-level 400 m IM
swimmers.
In the IM swimming, the energetic and biomechanics differences between the four
strokes yield a variable relative contribution of each stroke to the final performance. How-
ever, it is not clear which stroke(s) are more important in the final performance in IM events.
In the 400 m IM, breaststroke and freestyle seems to be the most relevant stroke in female
swimmers [3,4]. In contrast, for male swimmers, the backstroke and breaststroke appear
more important [4,5]. When planning training, coaches need to determine then prescribe
the relative proportion of training of each stroke for the 400 m IM event throughout the
season. To achieve the best performance in 400 m IM, coaches must ensure that middle-
Int. J. Environ. Res. Public Health 2021, 18, 6474. https://doi.org/10.3390/ijerph18126474
https://www.mdpi.com/journal/ijerph
Int. J. Environ. Res. Public Health 2021, 18, 6474
2 of 14
distance front crawl training is a priority given a positive association between that freestyle
and IM swimming [6].
To maximize the IM swimmers’ performance, it is important to establish a detailed
understanding of the training characteristics of both the 200 m and 400 m IM events
for effective planning and monitoring. A coach needs to consider the training volume,
frequency and intensity distribution for maximizing physical capacity and performance.
Periodization can be defined as the macromanagement of the delineated stages of training
process with respect to the time allocated toward various elements [7]. The key aspects that
underpin periodization are: (i) determining relevant dates (e.g., main and minor competi-
tions), (ii) determining the sequence of phases for each training cycle and (iii) managing
load dynamics with the intent of achieving peak or optimal performance at the critical
competitions [8]. There are few studies which have examined the key aspects related to the
best performance in 400 m IM events, including the training organization and periodization.
Therefore, the aim of this narrative review is to examine the evidence on periodization
related to the 400 m IM and identify key elements and best practices for this event. We ad-
dress key aspects, such as bioenergetics necessary to plan the periodization for 400 m IM
and examine traditional periodization following two or three peaks performance, training
methods and fitness phases for each training period in accordance with other narrative
reviews in individual sports [9]. A narrative review provides a historical account of the
development of theory and research on a topic (although the contribution to knowledge
will be relatively minor [10]. Here we address theoretical conceptualizations, training
constructs and relevant scientific literature, to propose a practical framework for preparing
400 m IM swimmers.
2. Literature Search Methodology
Electronic searches of PubMed/MEDLINE, SPORTDiscus, Scopus and Web of Science
were conducted. The search terms used were “individual medley swimming”, “middle
distance swimming training”, “swimming training periodization” and “swimming peri-
odization”. Relevant review articles were also examined to uncover studies which might
have been missed in the primary search. The reference list of selected manuscripts was
also examined for other potentially eligible manuscripts. No limits regarding the year
of publication were employed. Studies were included when (a) they were published in
English language (b); provided training zones, volumes and/or periodization details about
middle distance or IM events and (c) focused on swimming performance in IM events.
Exclusion criteria were: (a) swimmers with a current injury or disability and other aquatic
participants (e.g., water polo, diving, triathletes) and (b) studies focusing on pacing or
performance trends.
The initial database search identified 714 records that were relevant to the search
keywords. After removal of duplicates and elimination of papers based on title and
abstract screening, 15 manuscripts remained. Finally, four articles were included in this
review [11–14]. The 11 studies that did not match the eligibility criteria based on full-text
screening were discarded for one or more of the following reasons: not detailing training
intensity distributions (n = 6), conducted with master swimmers (n = 2), performance trends
in IM events across the years (n = 1) and pacing in 200 and 400 m IM (n = 1) (Figure 1).
Int. J. Environ. Res. Public Health 2021, 18, 6474
3 of 14
Int. J. Environ. Res. Public Health 2021, 18, x FOR PEER REVIEW
3 of 14
Figure 1. Flow chart summary of the study selection process.
3. Bioenergetics of Individual Medley Events
The 400 IM has a duration ranging from 4 to 4.30 min and is considered as a middle
distance swimming event [1]. Middle distance events are supported energetically by a
combination of phosphate energy; anaerobic glycolysis and aerobic combustion of carbo-
hydrate, fat and protein [15]. Competitive swimmers spend most of their training time
improving aerobic endurance, defined as the ability to sustain a high percentage of
VO2max for a long period, through careful and repeated interval-based training. This type
of training is important for performance in events around 4 min such as the 400 m IM [16].
The physiological preparation for a 400 m IM should address the key physiological factors
of the maximal aerobic power (rate of adenosine triphosphate resynthesis), capacity (total
amount of adenosine triphosphate resynthesis from available fuels) and VO2max (maxi-
mum oxygen uptake) [1]. The velocity associated with VO2max (vVO2max) is the single
best predictor of middle-distance swimming performance especially in 400 m events
[17,18]. From the data provided by 400 m front crawl swimmers [19,20], the estimated
velocity achieved during 400 m IM is ~100% of vVO2max. At these intensities, the attain-
ment of a VO2 steady state is delayed due to the emergence of a supplementary slowly
developing component of the VO2 response [21]. The VO2 fast component is stable at in-
tensities between 95, 100 and 105%; however, the kinetics of the VO2 slow component and
the corresponding metabolic profiles showed variations between this intensities [19].
Other important physiological factors include the lactate threshold (LT), the ability
to sustain a high percentage of VO2max during the competition and the energy cost of
locomotion [15,18,19,22]. The physiological adaptations should align with the periodiza-
tion of each swimmers’ training and competition calendar. These physiological adapta-
tions are usually prescribed with specific training sets and sessions in the pool and dry-
land training. One common approach in elite-level swimming to enhancing physiological
and performance adaptations is incorporation of altitude training (either real or simulated
to induce hypoxia).
4. Training Monitoring
External load monitoring is usually assessed by quantifying the weekly training vol-
ume [23]. The training volumes are usually classified into three or five intensities zones
[14]. The three training zone model is typically established using swimming velocity and
blood lactate concentrations as follows: z1 ≤ 2 mmol·L–1; z2 2–4 mmol·L–1, and z3 ≥ 4
mmol·L–1 [13]. However, [24,25] proposed a modification with z1 ≤ 3 mmol·L–1 and z2 be-
tween 3–4 mmol·L–1. In swimming, the most common model adopted in the sports science
literature comprises five zones: z1 ≤ 2 mmol·L–1, z2 2–4 mmol·L–1, z3 4–6 mmol·L–1, z4 6–
10 mmol·L–1 and z5 < 10 mmol·L–1 [12,13,26]. Training zones can be categorized according
Figure 1. Flow chart summary of the study selection process.
3. Bioenergetics of Individual Medley Events
The 400 IM has a duration ranging from 4 to 4.30 min and is considered as a mid-
dle distance swimming event [1]. Middle distance events are supported energetically
by a combination of phosphate energy; anaerobic glycolysis and aerobic combustion of
carbohydrate, fat and protein [15]. Competitive swimmers spend most of their training
time improving aerobic endurance, defined as the ability to sustain a high percentage of
VO2max for a long period, through careful and repeated interval-based training. This
type of training is important for performance in events around 4 min such as the 400 m
IM [16]. The physiological preparation for a 400 m IM should address the key physio-
logical factors of the maximal aerobic power (rate of adenosine triphosphate resynthesis),
capacity (total amount of adenosine triphosphate resynthesis from available fuels) and
VO2max (maximum oxygen uptake) [1]. The velocity associated with VO2max (vVO2max)
is the single best predictor of middle-distance swimming performance especially in 400 m
events [17,18]. From the data provided by 400 m front crawl swimmers [19,20], the es-
timated velocity achieved during 400 m IM is ~100% of vVO2max. At these intensities,
the attainment of a VO2 steady state is delayed due to the emergence of a supplementary
slowly developing component of the VO2 response [21]. The VO2 fast component is stable
at intensities between 95, 100 and 105%; however, the kinetics of the VO2 slow component
and the corresponding metabolic profiles showed variations between this intensities [19].
Other important physiological factors include the lactate threshold (LT), the ability
to sustain a high percentage of VO2max during the competition and the energy cost of
locomotion [15,18,19,22]. The physiological adaptations should align with the periodization
of each swimmers’ training and competition calendar. These physiological adaptations
are usually prescribed with specific training sets and sessions in the pool and dryland
training. One common approach in elite-level swimming to enhancing physiological and
performance adaptations is incorporation of altitude training (either real or simulated to
induce hypoxia).
4. Training Monitoring
External load monitoring is usually assessed by quantifying the weekly training
volume [23]. The training volumes are usually classified into three or five intensities
zones [14]. The three training zone model is typically established using swimming velocity
and blood lactate concentrations as follows: z1 ≤ 2 mmol·L−1; z2 2–4 mmol·L−1, and z3
≥ 4 mmol·L−1 [13]. However, [24,25] proposed a modification with z1 ≤ 3 mmol·L−1
and z2 between 3–4 mmol·L−1. In swimming, the most common model adopted in the
sports science literature comprises five zones: z1 ≤ 2 mmol·L−1, z2 2–4 mmol·L−1, z3
4–6 mmol·L−1, z4 6–10 mmol·L−1 and z5 < 10 mmol·L−1 [12,13,26]. Training zones can
be categorized according to the response in blood lactate concentration: Z1; Aerobic low
intensity (A1), z2; Aerobic maintenance (A2), z3; lactate threshold (LT), z4; VO2max and
Int. J. Environ. Res. Public Health 2021, 18, 6474
4 of 14
intensity above VO2max as 200 m race pace and z5 maximal swimming speed [26]. These
training zones need to be established and then checked periodically during a training
season
Monitoring of heart rate also has been used during training sessions to indicate
training intensities [27] but is subject to substantial biological and measurement error.
Nevertheless, blood lactate measurements are considered more useful in determining
the training intensity because they facilitate better monitoring of the effect of training
workloads on the muscle [23]. Thus, blood lactate is a good indicator of the muscles’
capacity for an athletic performance which allows coaches to identify the type and extent of
physiological disturbance and the degree of adaptation that has taken place over time [23].
An increase in blood lactate for the same training stimulus may, for example, point to in-
creased anaerobic metabolism, and therefore, higher levels of lactate at slower speeds may
be indicative of impending overtraining [28]. Nevertheless, the values of blood lactate are
associated with a high between-swimmer variability in swimming techniques, with a range
from <2 to >5 mmol·L−1 at lactate threshold intensity. Front crawl (3.3 mmol·L−1) and
breaststroke (2.9 mmol·L−1) present lower levels of blood lactate at the lactate threshold
intensity than butterfly (4.9 mmol·L−1) and backstroke (3.9 mmol·L−1) [29]. It is recom-
mended to schedule a blood lactate assessment test using a prescribed testing protocol
every few weeks [23].
The Rating of Perceived Exertion (RPE) is another commonly used method for assess-
ing the internal training load [30–33]. Studies have showed moderate to large correlations
between the heart rate and blood lactate concentrations [34,35]. Although, RPE is a valid
method for assessing the training stress in high-intensity exercises [36,37], it is important
to acknowledge that personal perceptions of physical efforts is a very complex interaction
of many factors [38]. Therefore, some investigators recommend to complement the RPE
with an objective assessment of internal training load such as blood lactate and/or heart
rate monitoring [39,40].
5. Training Periodization
Periodization is a process that serves as the macromanagement of the training program
in the context of the annual plan [7,41]. Various periodized models such as the reverse
linear [24,25,42], or block periodization [43], have been established, but the most common
periodized model in swimming is the so-called traditional periodization [14]. Over the
recent decades, many periodization approaches have evolved including traditional, blocks,
and reverse linear periodization, each offering a differing rationale and template for
sub-division of the program into sequential elements [14]. Some authors affirm that the
traditional model of periodization can take different forms (i.e., reverse) [44]. Reverse
linear periodization has been used in combination with a polarized intensity distribution
for improving sprint events. However, the small number of relevant studies did not report
any differences with the traditional model in 50 m performance, or a modest improvement
of 1% in 100 m performance [24,25]. Polarized training is not recommended for middle
distance swimmers; 400 m IM swimmers should benefit from specific periods of training
that employ a threshold-oriented training intensity distribution [13].
Training periodization involves the coordination of physical training, psychological
capacities training and skill acquisition, providing a comprehensive framework for optimal
preparation [45]. Periodization of training leads to a progressive enhancement in the critical
physiological and biomechanical factors required for swimming competitions [46]. On this
basis, a well-planned and effective periodized approach to training should be established,
monitored and refined for swimmers to achieve fitness and peak performance at the major
competition for the season [47,48]. In the same way, detailed monitoring of performance
and training during the season should be a fundamental aspect to maximize training
effectiveness and avoid excessive volume, intensity and/or training load which can cause
physiological disturbances (e.g., glycogen depletion, neuromuscular fatigue, decrements in
red cell volume and hemoglobin), injuries or illness [49].
Int. J. Environ. Res. Public Health 2021, 18, 6474
5 of 14
The traditional pyramidal model is most commonly used for swimmers characterized
by a sequential reduction in training volume moving from zone 1, to zones 2 and 3,
respectively. The majority (80%) of the volume is conducted in z1 and the remaining 20% in
z2 and z3 [50]. Issurin [51] conceptualized macrocycles as a training period which involved
a preparatory and competitive period, usually taking several months. A mesocycle is a
medium size training cycle consisting of a number of microcycles which usually involved
several weeks, while a microcycle is a small size training cycle consisting of a number of
days, frequently one week.
The season can been divided into three macrocycles [52], although the majority of
studies reported one or two macrocycles including a comprehensive evaluation conducted
on 127 elite swimmers and 20 competitive seasons [12]. This option seems to be the most
common used by coaches [53]. Our recommendation is to employ two or three macrocycles
divided into three mesocycles each (six or nine mesocycles in total), aligning with three
main competitions in each macrocycle to establish an annual plan. In many countries,
these competitions comprise in order: short course international championships, national
championships and main international competition for the calendar. The emergence of the
International Swimming League (ISL) in recent years may require more flexible planning
and periodization.
5.1. Altitude Training
Altitude training camps during a season can be useful in developing aerobic en-
durance in world-class endurance athletes [54]. Altitude training elicits an increases in
erythropoietic response [55] and hemoglobin mass [56,57]. However, there are also hypoxia-
induced non-hematological changes, such as mitochondrial gene expression and enhanced
muscle buffering capacity [58,59]. Altitude training can account for ~18–25% of annual
training volume in some world-class athletes [60] and is typically performed at altitudes
of ~1800–2300 m above sea level or higher [54,60,61]. The duration of an altitude training
camp depends on many factors; however, between three to four weeks is suggested for mid-
dle distance swimmers [59]. Endurance athletes can undertake altitude training to promote
specific training goals of the macrocycle [62]. For example, the early-season training camp
when training intensity is typically lower can focus on higher training volumes at low-to-
moderate intensities and capitalize on the hematological effects of the hypoxic stimulus [62].
Subsequent training camps should progress to a focus on lactate threshold, aerobic power
and VO2max training later in the macrocycle. However, a period of low-intensity training
during the first few days of altitude acclimatization is recommended [63].
Altitude training camps are generally placed before the main competitions. Six to
nine weeks prior to the main competition is typical time to undertake altitude training [59].
Nevertheless, other training camps could be carry out in the middle of the macrocycles
to emphasize aerobic training contents [59]. One key aspect that coaches and swimmers
must consider is the timing of return from altitude prior to competition. The timing of a
peak performance following altitude training is likely to be influenced by a combination
of altitude acclimatization and de-acclimatization responses, but more importantly are
the periodization of and responses to training and tapering conducted at and after alti-
tude [59]. Previous studies reported periods of between three and five weeks (usually three
weeks) [64,65] between altitude training camps and the main competition. This period
seems to be optimal timing of post-altitude performance peaking, but individualizing
training will be important for optimizing the time to compete after an altitude camp [65].
5.2. Preparatory and Main Competitions through the Season
In most individual sports, competitive athletes plan to optimize their performance at
the main competition no more than two or three times per year [66]. A retrospective study
divided the training season in two macrocycles, the first leading to the national selection
trials and the second macrocycle leading to the major international competition) [12]. Each
macrocycle should include at least two preparatory (minor) competitions before a major
Int. J. Environ. Res. Public Health 2021, 18, 6474
6 of 14
competition. Swimmers performance in the main competitions should faster given the
effects of the tapering phase [67] and extra motivation at the main competitions [66]. After a
2-week taper period, swimmers can show an improvement of ~3% in the main competition
in comparison with the preparatory competitions carried out three to six weeks before [68].
A common strategy adopted by IM coaches in minor competitions is to have swimmers
compete in one or more form stroke events (e.g., butterfly, backstroke or breaststroke) to
complement the specific IM events. At the major competitions, IM swimmers typically
concentrate on their main event (200 m or 400 m IM), but selection of events will depend
on qualifications, team selections and coach/swimmer preferences.
5.3. Training Intensity Distribution
Training load variables such as volume, frequency and intensity distribution play
an important role in maximizing physical capacity and performance [69]. Annual vol-
ume of kilometers for a middle-distance swimmer (400 m freestyle) ranged from 2055 to
2600 km [58]. Increasing the training volume is not be the only way of enhancing perfor-
mance and more objective and specific training sets are required to improve the quality of
the swimming training process [18]. Weekly volume and training intensity distribution are
used as a reference for determining a swimmers’ training load and prescribing training sets
and sessions. Middle distance swimmers show ranges of training volumes between 39,000
and 42,000 m depending on the type of macrocycles used [12]. However, some training
plans showed mean training volumes as high as 58,000 m [11]—peak volumes as high
as 70,000–80,000 m—have been reported anecdotally for some international IM and dis-
tance swimmers. On this basis, both training volume and intensity distribution should be
evaluated together for IM swimmers. A retrospective study showed that middle distance
swimmers follow a threshold model in which ~40–44% of the training was performed at
an intensity of < 2 mmol·L−1 (z1), and 44–46% at 2 to ≤ 4 mmol·L−1 (z2) and 9–14% at
>4 mmol·L−1 (z3) [12]. Threshold-oriented intensity distribution (z1 66%, z2 25%, z3 9%)
can improve crucial training contents for 400 m IM swimmers at the velocity at 4 mmol·L−1
and VO2max [13]. In summary, 400 m IM swimmers should benefit from specific periods
of training that employ a threshold-oriented intensity distribution. Coaches should also
consider that swimmers who train with a threshold intensity distribution might experience
additional fatigue induced by the cumulative impacts of threshold and high-intensity
training [13]
5.4. Macrocycle Distribution
Training cycles should be prescribed according to the principles of individualization
and progression [12]. Two or three distinctive peaks of high total load are suggested
in the overall training programs of elite swimmers across the year depending on the
number of major competitions scheduled for a particular season. Application of suitable
wave-like cycles in units such as a two or three week mesocycle is used to promote
physiological adaptations and skill acquisition. Swimmers can engage in mono-, bi- or
tri-cycled periodized programs depending on the sequencing of important competitions
within that year. An annual periodization composed of two to three macrocycles would be
appropriate for Olympic and World Championship seasons [14]. For example, three waves
of macrocycles could be planned as follows: the first cycle is conducted from September to
December, second from December to April and third from April to August. A two waves
macrocycle timeline could be planned as follows: the first cycle from September to April
and the second from April to August.
The main aim of the first macrocycle is to develop the general physical fitness and
foundation work for specific qualities oriented to the event. The goal of second and third
macrocycles is to develop the specific and competitive physiological qualities (VO2max,
race pace) required for the event, building from general to sport-specific qualities required
culminating in the taper at the end of the season. In the two-wave macrocycle, the objectives
and training contents of the first and second macrocycles are included in a single macrocycle
Int. J. Environ. Res. Public Health 2021, 18, 6474
7 of 14
with the same duration as the first and second cycles in the three wave cycle. Moreover,
the second macrocycle of the two-wave cycle has the same duration, training contents and
objectives that the third cycle in the three-wave cycle.
5.4.1. First Macrocycle
The beginning of a season in this macrocycle requires development of aerobic en-
durance up to the lactate threshold, given it is a priority objective on the endurance
training for 400 m IM swimmers [53]. The sessions could be performed in short course
and long course training depending on the characteristics of the swimmers and the sets
planned by coaches. Strength and conditioning (dry-land) training should focus on strength-
hypertrophy, maximal strength and strength-metabolic conditioning (e.g., circuit training)
with a duration ranging from 50–80 min [70]. Circuit training includes a cardiovascular
element in combination with dry-land resistance training using light loads (40–60% one
repetition maximum (1 RM)) and brief rest intervals with circuits performed a number of
times per session. This work yields metabolic adaptations including an athlete’s buffering
capacity [70,71]. Core training sessions can be used to develop stability and postural
control of the body position while swimming. Stabilizing muscles can form the basis for
generating more strength through the limbs [72]. The competition in this cycle is scheduled
in December such as an international short course (for three peaks of performance), but
in two peaks of performance, the first cycle is scheduled in April (national selection trial
(Figure 2)).
The main aim of the first macrocycle is to develop the general physical fitness and
foundation work for specific qualities oriented to the event. The goal of second and third
macrocycles is to develop the specific and competitive physiological qualities (VO2max,
race pace) required for the event, building from general to sport-specific qualities required
culminating in the taper at the end of the season. In the two-wave macrocycle, the objec-
tives and training contents of the first and second macrocycles are included in a single
macrocycle with the same duration as the first and second cycles in the three wave cycle.
Moreover, the second macrocycle of the two-wave cycle has the same duration, training
contents and objectives that the third cycle in the three-wave cycle.
5.4.1. First Macrocycle
The beginning of a season in this macrocycle requires development of aerobic endur-
ance up to the lactate threshold, given it is a priority objective on the endurance training
for 400 m IM swimmers [53]. The sessions could be performed in short course and long
course training depending on the characteristics of the swimmers and the sets planned by
coaches. Strength and conditioning (dry-land) training should focus on strength-hyper-
trophy, maximal strength and strength-metabolic conditioning (e.g., circuit training) with
a duration ranging from 50–80 min [70]. Circuit training includes a cardiovascular element
in combination with dry-land resistance training using light loads (40–60% one repetition
maximum (1 RM)) and brief rest intervals with circuits performed a number of times per
session. This work yields metabolic adaptations including an athlete’s buffering capacity
[70,71]. Core training sessions can be used to develop stability and postural control of the
body position while swimming. Stabilizing muscles can form the basis for generating
more strength through the limbs [72]. The competition in this cycle is scheduled in De-
cember such as an international short course (for three peaks of performance), but in two
peaks of performance, the first cycle is scheduled in April (national selection trial (Figure
2)).
Figure 2. Example of macrocycle and mesocycle distribution. Note: %TTL: Total training load per-
centage; GP: General Phase; SP: Specific Phase; CP: Competitive Phase.
Figure 2. Example of macrocycle and mesocycle distribution. Note: %TTL: Total training load
percentage; GP: General Phase; SP: Specific Phase; CP: Competitive Phase.
5.4.2. Second Macrocycle
As a general rule, national championships doubling as selection trials are the main
competition in this cycle. In the majority of countries, this competition is the qualifying
event for the next international competition. The second macrocycle should be character-
ized by high training volume and high amount of training at z2 (2–4 mmol·L−1) and z4
(6–10 mmol·L−1) [12]. The objective of these sessions is to improve the aerobic endurance,
up to the level of the lactate threshold pace and VO2max. In addition, sessions aimed
at lactate tolerance and speed could be prescribed as a continuation of the workloads
performed in the first macrocycle. The strength and conditioning training should focus on
Int. J. Environ. Res. Public Health 2021, 18, 6474
8 of 14
maximal strength, power and speed endurance with resistance exercise. Circuit training
is recommended performed in sessions ranging from 30–90 min. During this cycle, the
swimmers must continue with the core training sessions for improving their stability. In ad-
dition, progressively, the strength training could be transformed from strength-metabolic
conditioning to muscle endurance with exercises that simulate the time frame of the event,
using light and moderate weights in every exercise (30–50% 1 RM).
5.4.3. Third Macrocycle
The third macrocycle would be the last cycle in the season, featuring the main com-
petition of this cycle which is also the main competition of the season for international
swimmers. Clearly, the Olympic Games or the World Championship is the main major
competition for achieving the peak performance. The previous competitions should focus
on increasing the technical and physical exigency leading to peak performance at the end
of the cycle. It is crucial in this cycle to emphasize the technical aspects, especially during
the early stages of this cycle, including basic training contents (as aerobic endurance),
progressing to specific contents later (as VO2max) and, finally, the competitive contents
(as race pace). Thus, training in the third macrocycle should be characterized by high
amount of training at z2 and z4. Strength and conditioning training is similar to the second
macrocycle with a focus on maximal strength, power and speed endurance with resistance
exercise. These attributes can be maintained by circuit training to improve aerobic fitness;
however, this type of training should be reduced when the main competition approaches.
5.5. Mesocycle Distribution
Each cycle is divided in preparatory, competitive and transition following the Matveyev’s
proposal [73]. Bompa and Haff [74] reported the preparatory phase has 2 subphases: gen-
eral phase (GP) and specific phase (SP). The competitive phase (CP) is when the athletes
need to peak for a competition. For example, the mesocycle distribution in each macrocycle
could keep the following distribution. Three waves macrocycles timeline could be planned
as follows, the first cycle: GP: 6 weeks, SP: 10 weeks and CP: 2 weeks; second: GP: 4 weeks,
SP: 7 weeks and CP: 3 weeks; and finally the third cycle: GP: 3 weeks, SP: 10 weeks and CP:
3 weeks. Moreover, a two waves macrocycle timeline could be planned as follows, first
cycle: GP: 12 weeks, SP: 17 weeks and CP: 3 weeks; second: GP: 3 weeks, SP: 10 weeks and
CP: 3 weeks.
5.5.1. General Phase
The main objective of the general phase is to induce physiological, psychological,
and technical adaptations that serve as the foundation for competitive performances [75].
The development of aerobic or oxidative endurance should be the main objective in this
phase (see Table 1). Lactate threshold and VO2max sets could be included in the last few
weeks of these mesocycle. Front crawl would be the recommended stroke to perform this
type of session because of the large volume required in aerobic sets. However, coaches
should consider conducting mixed sets with other strokes. To improve the aerobic en-
durance, swimmers should train at hart rate 40 to 30 beats below maximum. The suggested
pace is half of personal best 200 m time plus 10 to 15 sec. The repeat distances to use when
training in this category are 200 to 1500 m [76]. Moreover, lactate threshold and VO2max
training could be placed in the final weeks of the general mesocycle as introductory sets
for the subsequent mesocycle. It would be recommended that a training volume between
55–65 km per week during this phase is appropriate for most 400 m IM swimmers. Dry-
land training during the general mesocycle is focused on the strength and hypertrophy
development. During the first cycle, strength and hypertrophy development are the main
objectives, whereas in the second and third cycles coaches should ensure an appropriated
maximal strength development.
Int. J. Environ. Res. Public Health 2021, 18, 6474
9 of 14
Table 1. Aerobic development and mixed-endurance training sessions and hypertrophy-maximum strength development
training set in the general mesocycle (final part) of the first cycle.
Swimming Training
Objective
Set
Volume (m)
Training
Intensity
Training
Zone
Stroke
Notes
Aerobic
development
and mixed-
endurance
training
1
2400
3 × (8 × 100)/
1:20–1:30–1:40
min
A2, LT, VO2max
Z2, Z3, Z4
FC
Performed as training testing,
speed control, HR, stroke,
frequency and [La-]
2
3600
36 × 100/
1:20–1:30–1:40
min
Sequence: 3 ×
100 A2 < 2 × 100
LT < 1 × 100
VO2max
Z2, Z3, Z4
FC
3 intensities simultaneously
with the [La-] of the previous
intensity in the next one
Strength Training
Objective
Cycle
Sets
Exercise
Repetitions
Intensity
Hypertrophy and
maximal strength
development
2
4
Short pull Hammer
8
65 < 75%RM
Row Hammer
6
80 < 85%RM
Pull over
8
70%RM
3
Pulls-Up (Eccentric)
3
Body weight
Pulls-UP (Supine grip)
3
6
Squat
8
85%RM
Hamstrings
8
75%RM
Note: Heart rate (HR); Blood lactate ([La-]); Front Crawl (FC); Aerobic maintenance (A2); LT: lactate threshold (LT); Maximum oxygen
uptake (VO2max); Repetition maximum (RM); Zone 1 (Z2); Zone 2 (Z2); Zone 3 (Z3); Zone 4 (Z4); Zone 5 (Z5). The recommendations made
in this table are based on both scientific and empirical data.
5.5.2. Specific Phase
As the swimmers progresses to the specific phase of training, it is important maintain
the level of physical development established during the preparatory phase. The aerobic
endurance training needs to be maintained through this phase. However, the training
should be focus on the lactate threshold and VO2max development (see Table 2). In the
case of the lactate threshold development, swimmers should train at heart rate from 30 to
20 beats below the maximum in repetitions of 50 to 400 m at seven to ten second plus the
half of personal best time in 200 m. For 400 m IM swimmers, it is recommended 3000 to
4500 m sets [53]. Finally, VO2max sets should be performed in sets of 300 to 500 m with 50
to 150 m repetitions. The suggested pace for VO2max training is half of personal best 200 m
time plus four to seven seconds [76]. Moreover, during the middle and the last part of this
period, the race pace training should be included. Race pace training can be carry out as
broken swims (a training repeat with more than one break) and splits (a training repeat
with one break) [53]. The specific sets should be performed in a mix of strokes, not only
in one stroke, and depending on the week, the training volume should oscillate between
65–90 km per week. During the training first cycle, the dryland training should be focus on
maximal strength development; nevertheless, during the second and third training cycles,
power and speed endurance development are the main objectives of the dryland training.
Maximum strength training could involve exercises as Hamstrings (training machine),
Leg press, Dumbbell Row, Chin ups, Lunges Back, Squat, Row Hammer, Bench Press and
Triceps overhead (pulley) with one or two sets with an intensity between 70–90% 1 RM
and four-six repetitions.
Int. J. Environ. Res. Public Health 2021, 18, 6474
10 of 14
Table 2. Aerobic and lactate threshold set in the specific mesocycle of the second cycle of the season.
Objective
Cycle
Set
Total
Volume
(m)
Training
Intensity
Training
Zone
Stroke
Notes
Aerobic
development
and high
intensity
(lactate
threshold)
training along
with aerobic
endurance
training
1
1
3600
1 × (300 m/3:45 min)
A1
Z1
FC
Intensity
progression in
each distance
1 × (400 m/5 min) +
1 × (500 m/6:30 min)
A2, LT
Z2-Z3
2
4000
1 × (400 m/5:15 min)
+ 8 × 50/50 s
A2, LT
Z2-Z3
BT
2 × (200 m) +
4 × 100/1:30 min)
A2, LT
Z2-Z3
BS
1 × (400 m/5 min) +
4 × (100 m/1:10 min)
A2, LT
Z2-Z3
FC
3
1000
4 × (25 m/1:30 min)
Max
Z5
UUS
4 × (50 m/2 min) +
1 × (100/2 min)
400 Race pace
Z4
FC
4 × (150 m/2 min)
A2 (last at LT)
Z2, Z3
FC
Note: Underwater undulatory swimming (UUS); Breaststroke (BK); Front crawl (FC); Butterfly (BT); Backstroke (BS); Aerobic low intensity
(A1); Aerobic maintenance (A2); LT: lactate threshold (LT); Zone 1 (Z2); Zone 2 (Z2); Zone 3 (Z3); Zone 4 (Z4); Zone 5 (Z5). The
recommendations made in this table are based on both scientific and empirical data.
5.5.3. Competitive Phase
Among the main tasks of the competitive phase is perfection of all training factors,
which enables the athlete to compete successfully in the main competition or champi-
onships targeted by the annual training plan [75]. A primary goal of the competitive and
tapering mesocycles is to remove fatigue to stimulate a supercompensation of performance.
During the competitive mesocycle, the most important training content is race pace training
(see Table 3) and dryland training where the focus on is power (explosiveness) development
(see Table 4). The tapering phase prior to international championships can include minor
or major competitions, moderate volume including more low intensity training [49,77].
The training volume is progressively decreased for performing a progressive sloped taper
phase, with the aim of achieving the best performance in the main competition. An optimal
taper duration appears to be between 8–21 days involving a decrease in training load of
~40–60% [75]. Mujika and Padilla [49] recommended a training load reduction between
60–90% and maintaining the training intensity to avoid detraining, provided reductions in
the other training variables allow (i.e., fewer training sessions per week or volume) for suffi-
cient recovery to optimize performance. The progressive reduction on the training volume
begins around 60 km per week and finishes around 20–30 km per week. The remarkable
decrease in volume took place in the second and third macrocycles of the season [77].
Int. J. Environ. Res. Public Health 2021, 18, 6474
11 of 14
Table 3. Race pace through mixed-endurance training set in the beginning of the competitive mesocycle of the second cycle.
Objective
Cycle
Set
Volume
(m)
Training
Intensity
Training
Zones
Stroke
Notes
Race pace
through mixed-
endurance
sessions
3
1
2200
2 × (8 × 50 m/1 min) +
1 × 400 m/5:30 min
400 m race
pace + A2
Z4-Z2
FC
Combination 400 m
speed, focusing on
frequency/speed
competitive and A2
1000
4 × (1 × 50 m/1:15
min) + 1 × 200/3 min
200 m race
pace + A2/LT
Z4-Z2-
Z3
FC
Intensity increase
looking for a [La-]
accumulation
800
4 × (150 m/2:15 min +
50 m/1 min)
LT + Max
speed
Z3-Z5
FC
The aim is to increase
the intensity,
focusing of AT and
50 m maximal speed
2
800
8 × (50 m/50 s) +
2 × (200/2:30 min)
200 m race
pace + A2
Z4-Z2
FC
The aim is to focus
on frequency/
competitive 400 m
speed
1000
10 × (100 m/
2–2:15 min)
400 m race
pace
Z4
1 × BT,
2 × BS,
3 × BK,
4 × FC
The aim is to
simulate the
competition as much
as possible
400
8 × (50 m/1 min)
Max speed
Z5
FC
The aim is to
simulate the last part
of the competition
800
4 × (200 m/2:30 min)
A2
Z2
FC
The aim is to remove
the lactate at AT
intensity
Note: Blood lactate ([La-]); Breaststroke (BK); Front crawl (FC); Butterfly (BT); Backstroke (BS); Aerobic low intensity (A1); Aerobic
maintenance (A2); LT: lactate threshold (LT); Maximum oxygen uptake (VO2 max); Zone 1 (Z2); Zone 2 (Z2); Zone 3 (Z3); Zone 4 (Z4);
Zone 5 (Z5). The recommendations made in this table are based on both scientific and empirical data.
Table 4. Power training set in the beginning of the competitive mesocycle of the second cycle.
Objective
Cycle
Sets
Exercise
Repetitions
Intensity
Power
development
2
3×
Push-ups (additional weight)
4
105–110% BW
Bench Press
Max
60% RM-0.9 m/s
5×
Chin-ups (Eccentric)
4
Body weight
Row Hammer
Max
60% RM-0.9 m/s
3×
Isometric Squat
20”
Body weight
Squat
Max
60% RM-0.9 m/s
Note: Body weight (BW); Repetition maximum (RM); Meters per second (m/s). The recommendations made in this table are based on both
scientific and empirical data.
6. Conclusions
Knowledge of the preparation and periodization of IM training over a season is limited
in both the coaching and sports science literature. Progressive development of the critical
energetic and biomechanics variables involves the design, implementation and evaluation
of an effective IM training plan. The training program we detailed here was organized with
a traditional periodization paradigm using two or three macrocycles for the season (with
two to three main competitions) incorporating a series of altitude camps. It is incumbent
upon the coach to adjust the programming based on individual responses and swimmers’
characteristics to optimize the training process for each swimmer. Future investigations
into IM training should determine the long-term effects of individual elite swimmers.
It would also be informative to investigate the effects of different periodization models and
training loads distribution on the performance in 400 IM using observation analyses of
elite swimmers and controlled studies with high-level emerging swimmers.
Int. J. Environ. Res. Public Health 2021, 18, 6474
12 of 14
Author Contributions: Conceptualization, J.M.G.-R., F.H. and J.A.D.C.; database search, F.H. and
J.A.D.C.; writing—original draft preparation, J.M.G.-R. and F.H; writing—review and editing, J.M.G.-
R., F.H. and D.B.P. All authors have read and agreed to the published version of the manuscript.
Funding: This research received no external funding.
Institutional Review Board Statement: Not applicable.
Informed Consent Statement: Not applicable.
Data Availability Statement: Not applicable.
Conflicts of Interest: The authors declare no conflict of interest.
References
1.
Pyne, D.; Sharp, R. Physical and energy requirements of competitive swimming events. Int. J. Sport Nutr. Exerc. Metab. 2014, 24,
351–359. [CrossRef] [PubMed]
2.
Gonjo, T.; Olstad, B.H. Race Analysis in Competitive Swimming: A Narrative Review. Int. J. Environ. Res. Public Health 2021, 18,
69. [CrossRef] [PubMed]
3.
What Can We Learn from Competition Analysis at the 1999 Pan Pacific Swimming Championships? Available online: https:
//ojs.ub.uni-konstanz.de/cpa/article/view/2155 (accessed on 14 June 2021).
4.
Saavedra, J.M.; Escalante, Y.; Garcia-Hermoso, A.; Arellano, R.; Navarro, F. A 12-year analysis of pacing strategies in 200-and
400-m individual medley in international swimming competitions. J. Strength Cond. Res. 2012, 26, 3289–3296. [CrossRef]
[PubMed]
5.
Robertson, E.Y.; Pyne, D.B.; Hopkins, W.G.; Anson, J.M. Analysis of lap times in international swimming competitions. J. Sports
Sci. 2009, 27, 387–395. [CrossRef]
6.
Del Castillo, J.A.; González-Ravé, J.M.; Hermosilla, F.; Santos del Cerro, J.; Pyne, B.D. The importance of previous season
performance on world-class 200- and 400-m individual medley swimming. Biol. Sport 2021, 39, 45–51.
7.
Cunanan, A.J.; DeWeese, B.H.; Wagle, J.P.; Carroll, K.M.; Sausaman, R.; Hornsby, W.G.; Haff, G.G.; Triplett, N.T.; Pierce, K.C.;
Stone, M.H. The general adaptation syndrome: A foundation for the concept of periodization. Sports Med. 2018, 48, 787–797.
[CrossRef]
8.
Afonso, J.; Clemente, F.M.; Ribeiro, J.; Ferreira, M.; Fernandes, R.J. Towards a de facto Nonlinear Periodization: Extending
Nonlinearity from Programming to Periodizing. Sports 2020, 8, 110. [CrossRef]
9.
Boullosa, D.; Esteve-Lanao, J.; Casado, A.; Peyré-Tartaruga, L.A.; da Rosa, R.G.; Del Coso, J. Factors Affecting Training and
Physical Performance in Recreational Endurance Runners. Sports 2020, 8, 35. [CrossRef]
10.
Baumeister, R.F.; Leary, M.R. Writing narrative literature reviews. Rev. Gen. Psychol. 1997, 1, 311–320. [CrossRef]
11.
Pollock, S.; Gaoua, N.; Johnston, M.J.; Cooke, K.; Girard, O.; Mileva, K.N. Training regimes and recovery monitoring practices of
elite British swimmers. J. Sports Sci. Med. 2019, 18, 577. [PubMed]
12.
Hellard, P.; Avalos-Fernandes, M.; Lefort, G.; Pla, R.; Mujika, I.; Toussaint, J.-F.; Pyne, D.B. Elite swimmers’ training patterns in
the 25 weeks prior to their season’s best performances: Insights into periodization from a 20-years cohort. Front. Physiol. 2019, 10,
363. [CrossRef]
13.
Pla, R.; Le Meur, Y.; Aubry, A.; Toussaint, J.; Hellard, P. Effects of a 6-week period of polarized or threshold training on performance
and fatigue in elite swimmers. Int. J. Sports Physiol. Perform. 2019, 14, 183–189. [CrossRef]
14.
González-Ravé, J.M.; Hermosilla, F.; González-Mohíno, F.; Casado, A.; Pyne, D.B. Training Intensity Distribution, Training
Volume, and Periodization Models in Elite Swimmers: A Systematic Review. Int. J. Sports Physiol. Perform. 2021, 1, 1–14. (in press).
[CrossRef]
15.
Capelli, C.; Pendergast, D.R.; Termin, B. Energetics of swimming at maximal speeds in humans. Eur. J. Appl. Physiol. Occup.
Physiol. 1998, 78, 385–393. [CrossRef]
16.
Toubekis, A.G.; Tokmakidis, S.P. Metabolic responses at various intensities relative to critical swimming velocity. J. Strength Cond.
Res. 2013, 27, 1731–1741. [CrossRef]
17.
Reis, J.F.; Alves, F.B.; Bruno, P.M.; Vleck, V.; Millet, G.P. Oxygen uptake kinetics and middle distance swimming performance.
J. Sci Med. Sport 2012, 15, 58–63. [CrossRef] [PubMed]
18.
Fernandes, R.J.; Vilas-Boas, J.P. Time to Exhaustion at the VO2max Velocity in Swimming: A Review. J. Hum. Kinet. 2012, 32,
121–134. [CrossRef] [PubMed]
19.
Fernandes, R.; Keskinen, K.; Colaço, P.; Querido, A.; Machado, L.; Morais, P.; Novais, D.; Marinho, D.; Vilas-Boas, J.P. Time Limit
at VO2max Velocity in Elite Crawl Swimmers. Int. J. Sports Med. 2008, 29, 145–150. [CrossRef] [PubMed]
20.
Sousa, A.C.; Vilas-Boas, J.P.; Fernandes, R.J. Kinetics and Metabolic Contributions Whilst Swimming at 95, 100, and 105% of the
Velocity at. Biomed. Res. 2014, 2014, 675363.
21.
Jones, A.M.; Grassi, B.; Christensen, P.M.; Krustrup, P.; Bangsbo, J.; Poole, D.C. Slow component of VO2 kinetics: Mechanistic
bases and practical applications. Med. Sci. Sports Exerc. 2011, 43, 2046–2062. [CrossRef]
Int. J. Environ. Res. Public Health 2021, 18, 6474
13 of 14
22.
Zamparo, P.; Capelli, C.; Cautero, M.; Di Nino, A. Energy cost of front-crawl swimming at supra-maximal speeds and underwater
torque in young swimmers. Eur. J. Appl. Physiol. 2000, 83, 487–491. [CrossRef]
23.
Feijen, S.; Tate, A.; Kuppens, K.; Barry, L.A.; Struyf, F. Monitoring the swimmer’s training load: A narrative review of monitoring
strategies applied in research. Scand. J. Med. Sci. Sports 2020, 30, 2037–2043. [CrossRef] [PubMed]
24.
Clemente-Suárez, V.; Fernandes, R.J.; Arroyo-Toledo, J.; Figueiredo, P.; González-Ravé, J.M.; Vilas-Boas, J. Autonomic adaptation
after traditional and reverse swimming training periodizations. Acta Physiol. Hung. 2015, 102, 105–113. [CrossRef] [PubMed]
25.
Clemente-Suárez, V.; Fernandes, R.J.; de Jesus, K.; Pelarigo, J.G.; Arroyo-Toledo, J.J.; Vilas-Boas, J.P. Do traditional and reverse
swimming training periodizations lead to similar aerobic performance improvements? J. Sports Med. Phys. Fit. 2018, 58, 761–767.
26.
Mujika, I.; Chatard, J.-C.; Busso, T.; Geyssant, A.; Barale, F.; Lacoste, L. Effects of training on performance in competitive
swimming. Can. J. Appl. Physiol. 1995, 20, 395–406. [CrossRef] [PubMed]
27.
Achten, J.; Jeukendrup, A.E. Heart rate monitoring. Sports Med. 2003, 33, 517–538. [CrossRef] [PubMed]
28.
Olbrecht, J. Determining Training Intensity and Content. In The Science of Winning: Planning, Periodizing and Optimizing Swim
Training; F & G Partners: Antwerp, Belgium, 2007.
29.
Carvalho, D.D.; Soares, S.; Zacca, R.; Sousa, J.; Marinho, D.A.; Silva, A.J.; Vilas-Boas, J.P.; Fernandes, R.J. Anaerobic threshold
biophysical characterisation of the four swimming techniques. Int. J. Sports Med. 2020, 41, 318–327. [CrossRef] [PubMed]
30.
Nicolas, M.; Vacher, P.; Martinent, G.; Mourot, L. Monitoring stress and recovery states: Structural and external stages of the short
version of the RESTQ sport in elite swimmers before championships. J. Sport Health Sci. 2019, 8, 77–88. [CrossRef]
31.
Elbe, A.-M.; Rasmussen, C.P.; Nielsen, G.; Nordsborg, N.B. High intensity and reduced volume training attenuates stress and
recovery levels in elite swimmers. Eur. J. Sport Sci. 2016, 16, 344–349. [CrossRef]
32.
García-Ramos, A.; Feriche, B.; Calderón, C.; Iglesias, X.; Barrero, A.; Chaverri, D.; Schuller, T.; Rodríguez, F.A. Training load
quantification in elite swimmers using a modified version of the training impulse method. Eur. J. Sport Sci. 2015, 15, 85–93.
[CrossRef]
33.
Foster, C.; Boullosa, D.; McGuigan, M.; Fusco, A.; Cortis, C.; Arney, B.E.; Orton, B.; Dodge, C.; Jaime, S.; Radtke, K.; et al. 25 Years
of Session Rating of Perceived Exertion: Historical Perspective and Development. Int. J. Sports Physiol. Perform. 2021, 16, 612–621.
[CrossRef]
34.
Psycharakis, S.G. A longitudinal analysis on the validity and reliability of ratings of perceived exertion for elite swimmers.
J. Strength Cond. Res. 2011, 25, 420–426. [CrossRef]
35.
Ueda, T.; Kurokawa, T. Relationships between perceived exertion and physiological variables during swimming. Int. J. Sports
Med. 1995, 16, 385–389. [CrossRef]
36.
Green, J.M.; McLester, J.R.; Crews, T.R.; Wickwire, P.J.; Pritchett, R.C.; Lomax, R.G. RPE association with lactate and heart rate
during high-intensity interval cycling. Med. Sci. Sports Exerc. 2006, 38, 167–172. [CrossRef] [PubMed]
37.
Foster, C.; Florhaug, J.A.; Franklin, J.; Gottschall, L.; Hrovatin, L.A.; Parker, S.; Doleshal, P.; Dodge, C. A new approach to
monitoring exercise training. J. Strength Cond. Res. 2001, 15, 109–115.
38.
Williams, J.G.; Eston, R.G. Determination of the intensity dimension in vigorous exercise programmes with particular reference to
the use of the rating of perceived exertion. Sports Med. 1989, 8, 177–189. [CrossRef] [PubMed]
39.
Foster, C. Monitoring training in athletes with reference to overtraining syndrome. Med. Sci. Sports Exerc. 1998, 30, 1164–1168.
[CrossRef] [PubMed]
40.
Impellizzeri, F.M.; Marcora, S.M.; Coutts, A.J. Internal and external training load: 15 years on. Int. J. Sports Physiol. Perform. 2019,
14, 270–273. [CrossRef] [PubMed]
41.
Kataoka, R.; Vasenina, E.; Loenneke, J.; Buckner, S.L. Periodization: Variation in the Definition and Discrepancies in Study Design.
Sports Med. 2021, 51, 625–651. [CrossRef]
42.
Arroyo-Toledo, J.J.; Clemente, V.J.; Gonzalez-Rave, J.M.; Ramos Campo, D.J.; Sortwell, A. Comparison between traditional and
reverse periodization: Swimming performance and specific strength values. Int. J. Swim. Kinet. 2013, 2, 87–96.
43.
Issurin, V.B. Biological background of block periodized endurance training: A review. Sports Med. 2019, 49, 31–39. [CrossRef]
44.
Stone, M.H.; Hornsby, G.; Haff, G.; Fry, A.; Suarez, D.; Liu, J.; Gonzalez-Rave, J.M.; Pierce, K.C. Periodization and Block
Periodization in Sports: Emphasis on strength-power training: A provocative and challenging narrative. J. Strength Cond. Res.
2021, (in press).
45.
Mujika, I.; Halson, S.; Burke, L.M.; Balagué, G.; Farrow, D. An integrated, multifactorial approach to periodization for optimal
performance in individual and team sports. Int. J. Sports Physiol. Perform. 2018, 13, 538–561. [CrossRef] [PubMed]
46.
Laffite, L.P.; Vilas-Boas, J.P.; Demarle, A.; Silva, J.; Fernandes, R.; Billat, V.L. Changes in physiological and stroke parameters
during a maximal 400-m free swimming test in elite swimmers. Can. J. Appl. Physiol. 2004, 29, S17–S31. [CrossRef] [PubMed]
47.
Pyne, D. The periodisation of swimming training at the Australian Institute of Sport. Sports Coach 1996, 18, 34–38.
48.
Aspenes, S.T.; Karlsen, T. Exercise-training intervention studies in competitive swimming. Sports Med. 2012, 42, 527–543.
[CrossRef] [PubMed]
49.
Mujika, I.; Padilla, S. Scientific bases for precompetition tapering strategies. Med. Sci. Sports Exerc. 2003, 35, 1182–1187. [CrossRef]
[PubMed]
50.
Seiler, S. What is best practice for training intensity and duration distribution in endurance athletes? Int. J. Sports Physiol. Perform.
2010, 5, 276–291. [CrossRef]
51.
Issurin, V. Block periodization versus traditional training theory: A review. J. Sports Med. Phys. Fit. 2008, 48, 65.
Int. J. Environ. Res. Public Health 2021, 18, 6474
14 of 14
52.
Turner, A.N.; Bishop, C.; Cree, J.; Carr, P.; McCann, A.; Bartholomew, B.; Halsted, L. Building a high-performance model for sport:
A human development-centered approach. J. Strength Cond. Res. 2019, 41, 100–107. [CrossRef]
53.
Sweetenham, B.; Atkinson, J. Championship Swim Training; Human Kinetics: Champaign, IL, USA, 2003; Volume 1.
54.
Pugliese, L.; Serpiello, F.R.; Millet, G.P.; La Torre, A. Training diaries during altitude training camp in two Olympic champions:
An observational case study. J. Sports Sci. Med. 2014, 13, 666.
55.
Semenza, G.L. HIF-1: Mediator of physiological and pathophysiological responses to hypoxia. J. Appl. Physiol. 2000, 88, 1474–1480.
[CrossRef]
56.
Millet, G.P.; Chapman, R.F.; Girard, O.; Brocherie, F. Is live high train low altitude training relevant for elite athletes? Flawed
analysis from inaccurate data. Br. J. Sports Med. 2019, 53, 923–925. [CrossRef]
57.
Gore, C.J.; Sharpe, K.; Garvican-Lewis, L.A.; Saunders, P.U.; Humberstone, C.E.; Robertson, E.Y.; Wachsmuth, N.B.; Clark, S.A.;
McLean, B.D.; Friedmann-Bette, B. Altitude training and haemoglobin mass from the optimised carbon monoxide rebreathing
method determined by a meta-analysis. Br. J. Sports Med. 2013, 47 (Suppl. 1), 31–39. [CrossRef] [PubMed]
58.
Gore, C.J.; Clark, S.A.; Saunders, P.U. Nonhematological mechanisms of improved sea-level performance after hypoxic exposure.
Med. Sci. Sports Exerc. 2007, 39, 1600–1609. [CrossRef] [PubMed]
59.
Mujika, I.; Sharma, A.P.; Stellingwerff, T. Contemporary periodization of altitude training for elite endurance athletes: A narrative
review. Sports Med. 2019, 49, 1651–1669. [CrossRef] [PubMed]
60.
Solli, G.S.; Tønnessen, E.; Sandbakk, Ø. The training characteristics of the world’s most successful female cross-country skier.
Front. Physiol. 2017, 8, 1069. [CrossRef] [PubMed]
61.
Bonne, T.C.; Lundby, C.; Jørgensen, S.; Johansen, L.; Mrgan, M.; Bech, S.R.; Sander, M.; Papoti, M.; Nordsborg, N.B. “Live
High–Train High” increases hemoglobin mass in Olympic swimmers. Eur. J. Appl. Physiol. 2014, 114, 1439–1449. [CrossRef]
62.
Saunders, P.U.; Pyne, D.B.; Gore, C.J. Endurance training at altitude. High. Alt. Med. Biol. 2009, 10, 135–148. [CrossRef]
63.
Millet, G.P.; Roels, B.; Schmitt, L.; Woorons, X.; Richalet, J.-P. Combining hypoxic methods for peak performance. Sports Med.
2010, 40, 1–25. [CrossRef]
64.
Wachsmuth, N.; Völzke, C.; Prommer, N.; Schmidt-Trucksäss, A.; Frese, F.; Spahl, O.; Eastwood, A.; Stray-Gundersen, J.; Schmidt,
W. The effects of classic altitude training on hemoglobin mass in swimmers. Eur. J. Appl. Physiol. 2013, 113, 1199–1211. [CrossRef]
65.
Chapman, R.F.; Stickford, A.S.L.; Lundby, C.; Levine, B.D. Timing of return from altitude training for optimal sea level performance.
J. Appl Physiol. 2014, 116, 837–843. [CrossRef]
66.
Bonifazi, M.; Sardella, F.; Lupo, C. Preparatory versus main competitions: Differences in performances, lactate responses and
pre-competition plasma cortisol concentrations in elite male swimmers. Eur. J. Appl. Physiol. 2000, 82, 368–373. [CrossRef]
[PubMed]
67.
Hellard, P.; Avalos, M.; Hausswirth, C.; Pyne, D.; Toussaint, J.F.; Mujika, I. Identifying Optimal Overload and Taper in Elite
Swimmers over Time. J. Sports Sci. Med. 2013, 12, 668–678.
68.
Costill, D.; Thomas, R.; Robergs, R.; Pascoe, D.; Lambert, C.; Barr, S.; Fink, W. Adaptations to swimming training: Influence of
training volume. Med. Sci. Sports Exerc. 1991, 23, 371–377. [CrossRef] [PubMed]
69.
Wenger, H.; Bell, G. The interactions of intensity, frequency and duration of exercise training in altering cardiorespiratory fitness.
Sports Med. 1986, 3, 346–356. [CrossRef] [PubMed]
70.
Crowley, E.; Harrison, A.J.; Lyons, M. Dry-land resistance training practices of elite swimming strength and conditioning coaches.
J. Strength Cond. Res. 2018, 32, 2592–2600. [CrossRef]
71.
Gotshalk, L.A.; Berger, R.A.; Kraemer, W.J. Cardiovascular responses to a high-volume continuous circuit resistance training
protocol. J. Strength Cond. Res. 2004, 18, 760–764.
72.
Willardson, J.M. Core stability training: Applications to sports conditioning programs. J. Strength Cond. Res. 2007, 21, 979–985.
[CrossRef]
73.
Matveyev, L.P. Periodization of Sports Training; Fiscultura I Sport: Moscow, Russia, 1966.
74.
Bompa, T.O.; Haff, G.G. Periodization: Theory and Methodology of Training; Human Kinetics: Champaign, IL, USA, 2009.
75.
Bompa, T.O.; Buzzichelli, C. Periodization: Theory and Methodology of Training; Human Kinetics: Champaign, IL, USA, 2018.
76.
Pyne, D. Training for positive outcomes. In ASCTA 1999 Conference Presentations; Australian Swimming Coaches and Teachers
Association: Lavington, Australia, 1999.
77.
Sokołowski, K.; Strzała, M.; Stanula, A. Different forms of swimmers’ final weeks pre-competition preparation. Balt. J. Health
Phys. Act. 2020, 12, 105–119. [CrossRef]
| Periodization and Programming for Individual 400 m Medley Swimmers. | 06-15-2021 | Hermosilla, Francisco,González-Rave, José M,Del Castillo, José Antonio,Pyne, David B | eng |
PMC10477198 | Physiological Reports. 2023;11:e15801.
| 1 of 12
https://doi.org/10.14814/phy2.15801
wileyonlinelibrary.com/journal/phy2
1 | INTRODUCTION
An important practical challenge in modern sports is to
obtain objective information on an athlete's physical per-
formance and fitness during the training process. One of
the most informative and popular methods for assessing
the functional status of athletes is to study the anaerobic
threshold (AT) (Ghosh, 2004; Solli et al., 2017). The AT
represents the transition to an anaerobic mechanism of
energy exchange when performing physical activity at
submaximal and maximum power. The production and
excretion of blood lactate are in equilibrium at the AT
(MLSS, maximal lactate steady state), and a significant
increase in the lactate level is observed (i.e., the lactate
threshold) then the athlete transitions through the AT
zone (Abreu et al., 2016; Gobatto et al., 2001). Despite
numerous studies, the metabolic basis of AT has not
been fully established. Studies of AT in humans are often
Received: 20 March 2023 | Revised: 10 August 2023 | Accepted: 14 August 2023
DOI: 10.14814/phy2.15801
O R I G I N A L A R T I C L E
Lactate thresholds and role of nitric oxide in male rats
performing a test with forced swimming to exhaustion
Natalya Potolitsyna
| Olga Parshukova | Nadezhda Vakhnina |
Nadezhda Alisultanova | Lubov Kalikova | Anastasia Tretyakova |
Alexey Chernykh | Vera Shadrina | Arina Duryagina | Evgeny Bojko
Institute of Physiology of Kоmi Science
Centre of the Ural Branch of the
Russian Academy of Sciences, FRC
Komi SC UB RAS, Syktyvkar, Russia
Correspondence
Evgeny Bojko, Institute of Physiology
of Kоmi Science Centre of the Ural
Branch of the Russian Academy of
Sciences, FRC Komi SC UB RAS,
Syktyvkar, Russia.
Email: [email protected]
Funding information
Institute of Physiology of Kоmi Science
Centre of the Ural Branch of the Russian
Academy of Sciences, FRC Komi SC UB
RAS, FUUU- 2022- 0063, Grant/Award
Number: 1021051201877- 3
Abstract
The present study assessed a complex of biochemical parameters at the anaerobic
threshold (AT) in untrained male Wistar rats with different times to exhaustion
(Tex) from swimming. The first group of rats was randomly divided into six sub-
groups and subjected to a swimming test to exhaustion without a load or with a
load of 2%– 10% of body weight (BW). In the first group, we established that for
untrained rats, the load of 4% BW in the swimming to exhaustion test was optimal
for endurance assessment in comparison with other loads. The second group of
rats went through a preliminary test with swimming to exhaustion at 4% BW and
was then divided into two subgroups: long swimming time (LST, Tex > 240 min)
and short swimming time (SST, Tex < 90 min). All rats of the second group per-
formed, for 6 days, an experimental training protocol: swimming for 20 min each
day with weight increasing each day. We established that the AT was 3% BW in
SST rats and 5% BW in LST rats. The AT shifted to the right on the lactate curve
in LST rats. Also, at the AT in the LST rats, we found significantly lower levels of
blood lactate, cortisol, and NO.
K E Y W O R D S
anaerobic threshold, exhaustion, lactate biochemical indices, rats, swimming test
This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided
the original work is properly cited.
© 2023 Institute of Physiology of Komi Science Center of the UB of the RAS. Physiological Reports published by Wiley Periodicals LLC on behalf of The Physiological
Society and the American Physiological Society.
2 of 12 |
POTOLITSYNA et al.
complicated by the infeasibility of complete control over
the experiments, and in- depth deep, long- term and high-
intensity studies of the AT phenomenon are not always
possible in humans (Cholewa et al., 2014). Therefore,
to understand many fundamental aspects of exercise
physiology, research using an adequate animal model is
needed. Rodents are among the most popular and eas-
ily available laboratory animals for use in such models.
Compared to human studies, working with animals al-
lows for greater control and regulation of environmental
conditions and food intake. These studies make it possible
to collect different types of tissues and perform a number
of experimental manipulations that cannot be performed
in humans (Cholewa et al., 2014). The results of studies
on rats are traditionally projected onto humans (Voltarelli
et al., 2002). There is evidence that rats adequately reflect
the human response to physical activity based on the main
biochemical parameters of blood (Goutianos et al., 2015).
However, rodents and humans may not have similar reac-
tions to physical exercise (Greek et al., 2012; Rice, 2012).
The AT in animals is determined by various changes in
blood biochemistry (Faude et al., 2009):
1. when animals achieve the reference lactate level in the
blood (4 mmol/L) during exercise (Heck et al., 1985);
2. when the first increase in blood lactate levels above the
baseline level is detected (onset of blood lactate accu-
mulation) (Farrell et al., 2018; Faude et al., 2009);
3. when a notable bend (sharp change in curvature) in
the lactate curve caused by massive lactate accumu-
lation during physical load is observed (Contarteze et
al., 2008; Gobatto et al., 2001);
4. when undergoing a short period of submaximal load to
induce hyperlactemia before starting the test with an
increasing load (lactate minimum test, LMT) (Voltarelli
et al., 2002).
In our opinion, fixed or precalculated lactate levels
might not take into account considerable interindividual
differences and differences between various lactate ana-
lyzers (Faude et al., 2009). Therefore, the most objective
AT assessment method is by observing a sharp change
(bend) in the lactate curve during exercise with an increas-
ing load. Then, the AT is determined as the physical load
equivalent to MLSS; in other words, the highest exercise
load at which the lactate levels in blood do not change sig-
nificantly (Contarteze et al., 2008; Faude et al., 2009; Heck
et al., 1985). In the case of individual lactate curves, visual
curve assessment is used. For groups, we believe a statisti-
cally significant increase in blood lactate levels in compar-
ison with the previous physical load is more appropriate.
It is also important to monitor other biochemical
indicators that characterize the level of physiological
adaptations of the body during exercise. Indicators in
sports physiology, such as cortisol, catecholamines, glu-
cose, urea, and other metabolites, are most often used (De
Araujo et al., 2016; Halson & Jeukendrup, 2004). Various
studies have also proposed other markers of AT, includ-
ing blood catecholamines (Davies et al., 1974) and saliva
amylase (Chicharro et al., 1999). We showed that elite ath-
letes (cross- country skiers) had a nitric oxide- dependent
(NO- dependent) mechanism for regulating lactate levels
during aerobic exercise, especially when working at the
AT. In our previous work, we revealed a positive relation-
ship between NO metabolites and blood lactate at the AT,
which was reversed at maximum load. This observation
suggests the existence of an adaptive mechanism for reg-
ulating the level of lactate on the AT in highly qualified
cross- country skiers (Parshukova et al., 2022).
The time of onset of the anaerobic (lactate) thresh-
old largely depends on the duration of the load (Pa-
padopoulos et al., 2006; Roecker et al., 1998; Weyand
et al., 1994), physical fitness level (Støren et al., 2014; Tanji
& Nabekura, 2019), intensity of movement (Wakayoshi
et al., 1993), testing methods (Contarteze et al., 2008; De
Araujo et al., 2016), and swimming patterns in swimmers
(Dos Reis et al., 2018). Perhaps due to the influence of a
large number of factors, research results are often contra-
dictory. Therefore, doubts are expressed that the AT (as an
indirect method for determining aerobic endurance) re-
flects the optimal intensity of training, especially for elite
athletes (Bosquet et al., 2002). Determination of the du-
ration of exercise to exhaustion is a direct method for de-
termining aerobic endurance and provides more complete
and reliable information (Beck et al., 2014). There are few
studies of the AT with measurements of the maximum du-
ration of physical activity to exhaustion in untrained rats.
Therefore, the hypothesis of the current study was that
the characteristics of the AT will be different in rats dis-
playing different times to exhaustion (Tex) while perform-
ing the same swimming tests in the same conditions. The
purpose of our study was to assess the biochemical param-
eters at the AT in untrained rats with different endurance
levels performing a swimming test to exhaustion.
2 | MATERIALS AND METHODS
2.1 | Experimental animals
Our study used male Wistar rats (n = 60), aged 8 weeks at
the beginning of the experiment, weighing 250– 300 g. Rats
were housed in a room with a temperature of 21 ± 1°C
and a controlled photoperiod (12 h of light/12 h of dark-
ness) on a standard vivarium diet, with access to water
ad libitum. The protocol of the study was reviewed for
| 3 of 12
POTOLITSYNA et al.
compliance with the “Rules of the European Convention
for the Protection of Vertebrates Used for Experimental
and Other Scientific Purposes” and approved by the local
Ethics Committee of the Institute of Physiology of the
Komi Research Center of the Ural Branch of the Russian
Academy of Sciences.
2.2 | Adaptation to water
Prior to experiments, all animals were adapted to water,
with a subsequent recovery for 14 days (Brito et al., 2015).
Adaptation consisted of keeping animals in shallow water
at a temperature of 31 ± 1°C for 30 min for 14 days. The
purpose of the adaptation period was to familiarize the
animals with testing conditions while avoiding physical
training adaptations.
2.3 | Experimental procedures
The swimming sessions were performed in cylindrical
water tanks (height 60 cm × diameter 45 cm) with a water
temperature of 30 ± 1°C and indoor air temperature of
22 ± 1°C. The rats swam in individual tanks with desatu-
rated water. After weighing the animals, a metal weight
of the necessary mass was affixed to the base of the tail
using elastic nontraumatic tape. A stopwatch was started
when the animals were placed in the water. Exhaustion
was defined as the animals being incapable of staying on
the water surface, the loss of symmetrical movements dur-
ing swimming, or the animals remaining underwater for
more than 10 s (Chimin et al., 2013). No deaths occurred
during or after exercise in any of the animal subgroups.
2.4 | Method for assessing physical
activity to exhaustion
Rats (n = 50) were randomly divided into six groups. Each
group performed a swimming test to exhaustion with one
of the following swim loads (SL): without load (SL0, n = 9)
or with a load of 2% (SL2, n = 8), 4% (SL4, n = 8), 6% (SL6,
n = 10), 8% (SL8, n = 8), and 10% (SL10, n = 7) of body
weight (BW). After the animals achieved exhaustion, they
were immediately removed from the water tank, dried,
anesthetized, and sacrificed via decapitation.
2.5 | Method of measuring the AT
To determine the AT, we used the method of Gobatto
et al. (2001) with modifications (Figure 1). A total of 10
rats were used in the study. All rats had previously gone
through the test to exhaustion with a load of 4% BW.
Based on this test, all rats were divided into two groups:
rats with a long swimming time (time to exhaustion,
Tex > 240 min, LST, n = 5) and rats with a short swimming
time (Tex < 90 min, SST, n = 5). After this test, all rats were
allowed to recover for 1 week. One rat from each group
was decapitated to assess the recovery of the blood bio-
chemistry parameters after exertion to exhaustion (recov-
ery control after swimming to exhaustion). The remaining
rats were subjected to physical activity to assess the AT.
FIGURE 1
The protocol for
anaerobic threshold evaluation in rats.
1st day
0% BM
swim 20 min
2nd day
2% BM
swim 20 min
3rd day
3% BM
swim 20 min
4th day
4% BM
swim 20 min
5th day
5% BM
swim 20 min
6th day
6% BM
swim 20 min
Swimming test to exhaustion with 4% of body weight
Recovery control after
swimming
to exhaustion
n=2
LST
Swimming, long time
Tex>240 min, n=4
Separation into groups by duration of the swimming
SST
Swimming, short time
Tex<90 min, n=4
Anaerobic threshold test
One week recovery
Sacrifce, blood sampling
4 of 12 |
POTOLITSYNA et al.
Each animal participated in six experimental tests over
6 days with a 24- h interval between tests. Each test con-
sisted of continuous swimming for 20 min with a load of
0%, 2%, 3%, 4%, 5%, or 6% BW in a tank filled with de-
saturated water at a temperature of 31 ± 1°C. Immediately
after performing the exercise, the rats were removed from
the water, and their tails were heated with warm water
and dried with a towel. Blood samples were taken from
the tail vein using a syringe and placed in heparinized Ep-
pendorf tubes (1.5 mL capacity). After performing the test
with last load (6% BW), the rats were sacrificed.
We determined the AT by observing a sharp increase
in blood lactate using the lactate curve obtained by per-
forming physical load tests (Faria et al., 2021; Faude
et al., 2009). The increase between the two consecutive
loads was required to be statistically significant. The AT
in this case was the lower of the two consecutive physi-
cal load intensities, the one after which the increase was
observed.
2.6 | Blood samples and analyses
The lactate levels in the blood samples collected from
the caudal veins were determined using a lactate ana-
lyzer (Accutrendplus, Roche Diagnostics GmbH). Blood
(mixed, arteriovenous) after decapitation was collected
in tubes containing heparin and centrifuged at 2400 rpm
for 10 min at 4°C. The samples were frozen and stored at
−40°C. Plasma levels of lactate (Sentinel Diagnostics),
urea, glucose, and cortisol (all from Human GmbH) were
measured using immunoenzyme assays (ChemWell 2900
biochemical analyzer). Levels of NO in the plasma were
measured using the Griess reaction by evaluating stable
metabolites of NO, including nitrites (NO2) and nitrates
(NO3), which were merged together as an index (NOx).
These methods were previously described (Parshukova
et al., 2020, 2022).
2.7 | Statistical analysis
All values are expressed as the means ± SD. Statistical
analyses were performed using Statistica 8.0 (Statsoft).
The statistical significance of differences between the SST
and LST groups was estimated using the Mann– Whitney
(U) test. For comparisons of multiple independent groups,
we used the Kruskal– Wallis test. For comparisons be-
tween workload groups within a corresponding group, we
used Friedman ANOVA and Kendall's coefficient of con-
cordance. When necessary, the Newman– Keuls post hoc
comparison test was used. The statistical significance level
was set at p < 0.05.
3 | RESULTS
The times of swimming to exhaustion and the parameters
of blood biochemistry in rats performing the Tex test with
various loads are presented in Table 1.
The total swimming time of rats performing the test to ex-
haustion expectedly decreased with increasing tail weights
and showed significant variation between individual rats.
The greatest variation between minimal and maximal
times of swimming was observed in the group swimming
with weights of 0%– 6% BW. All rats within each group were
clearly divided by the duration of swimming into LST and
SST groups (Table 2). The duration of swimming did not ex-
ceed 4 min in the SL8 and SL10 groups, and these values did
not differ significantly between the SL8 and SL10 groups.
The concentrations of glucose and lactate in the blood
of rats after exercise to exhaustion generally tended to
be higher as the weight of the attached load increased.
However, the picture became clearer when the rats were
separated within each group according to the duration of
swimming. Lactate and glucose levels in LST rats were sig-
nificantly lower than those in SST rats.
The average values of the other parameters did not show
regularities or trends that corresponded with differences in
load weights or the duration of swimming in groups with-
out separation. However, the differences became apparent
when the groups were divided into SST and LST groups.
The concentrations of cortisol were higher in SST rats than
in LST rats, and the concentrations of urea were lower.
Levels of nitric oxide metabolites were also notable. The
NOx index was significantly higher in LST rats than in SST
rats. The significant differences in NOx values primarily de-
pended on the levels of NO3, and the levels of NO2 showed
no significant differences between groups.
3.1 | Anaerobic threshold
Based on the time of swimming to exhaustion, all rats
were divided into rats that swam for more than 240 min
(LST, n = 4) and rats that swam for less than 90 min (SST,
n = 4). The dynamics of lactate and the level of the AT in
these groups are shown in Figure 2.
The dynamics of lactate in venous blood differed in
these two groups of rats. In the SST group, we observed a
sharper increase in this indicator, and starting at the load
of 3% BW, we registered a statistically significant differ-
ence from the first data point. However, the most signifi-
cant increase in lactate levels was detected at a load of 4%
BW. Further testing of rats at loads of 5%– 6% BW did not
show significant changes in lactate levels compared with a
load of 4% BW. Therefore, the AT in SST rats was assessed
at the level of 3% BW. The lactate curve in LST rats had a
| 5 of 12
POTOLITSYNA et al.
TABLE 1 Swimming time and biochemical parameters of blood in untrained male rats (all group) after performing swimming tests to exhaustion with different loads.
Work load groups
Swimming time,
min
Glucose,
mmol/L
Lactate,
mmol/L
Cortisol, ng/
mL
Urea, mmol/L
NOх, μmol/L
NO2, μmol/L
NO3, μmol/L
SL0
419.7 ± 229.1
4.9 ± 1.5
4.1 ± 2.5
21.0 ± 3.3
8.0 ± 3.5
37.0 ± 21.8
5.2 ± 2.3
31.8 ± 21.7
*SL6, SL8, SL10
*SL10
*SL8
*SL10
SL2
148.6 ± 130.2
5.2 ± 2.3
6.8 ± 4.4
19.8 ± 8.3
5.3 ± 1.8
27.7 ± 13.5
9.1 ± 2.1
18.5 ± 13.9
*SL4
*SL4
*SL4
*SL4
*SL4
SL4
89.1 ± 86.9
4.7 ± 2.4
5.7 ± 4.8
23.7 ± 5.5
5.1 ± 2.0
40.9 ± 8.2
3.9 ± 0.8
37.0 ± 7.8
*SL2, SL10
*SL2, SL8
*SL10
*SL2, SL8
*SL2
*SL2, SL6, SL8, SL10
SL6
59.9 ± 83.1
6.0 ± 2.8
11.0 ± 6.7
23.0 ± 5.0
6.3 ± 1.4
22.1 ± 13.0
6.4 ± 3.0
15.7 ± 15.1
*SL0
*SL4
SL8
3.0 ± 1.1
8.6 ± 2.9
14.2 ± 1.4
16.1 ± 2.4
5.0 ± 2.0
12.3 ± 2.5
5.6 ± 1.4
6.8 ± 2.7
*SL0
*SL0, SL4
*SL4
*SL4
SL10
2.2 ± 0.4
9.8 ± 1.8
13.3 ± 1.3
16.5 ± 3.8
2.8 ± 1.1
14.1 ± 2.2
6.6 ± 1.9
7.5 ± 1.7
*SL0
*SL0, SL4
*SL0, SL4
*SL4
Note: The values were expressed as means ± SD.
*The differences between workload groups are significant at <0.05 for the Kruskal– Wallis ANOVA test.
TABLE 2 Swimming time and biochemical parameters of blood in short swimming time (SST) and long swimming time (LST) male rats after performing swimming tests to exhaustion
with different loads.
Work load
groups
Swimming time, min
Glucose, mmol/L
Lactate, mmol/L
Cortisol, ng/mL
Urea, mmol/L
NOх, μmol/L
NO2, μmol/L
NO3, μmol/L
SST
LST
SST
LST
SST
LST
SST
LST
SST
LST
SST
LST
SST
LST
SST
LST
SL0
168.0 ± 24.0
545.5 ± 163.5
4.9 ± 2.3
5.0 ± 1.3
7.2 ± 1.6
2.5 ± 0.7
24.5 ± 2.2
19.2 ± 2.0
7.2 ± 5.5
8.5 ± 2.6
23.0 ± 18.7
42.0 ± 8.2
5.8 ± 2.1
5.2 ± 2.4
17.3 ± 19.9
36.8 ± 19.5
*SL6
#
*SL6
*SL6#
#
SL2
48.0 ± 14.2
282.7 ± 58.0
6.4 ± 2.4
3.5 ± 0.2
10.3 ± 0.5
2.1 ± 0.6
25.6 ± 5.5
15.4 ± 3.7
4.3 ± 0.7
6.6 ± 2.1
20.0 ± 3.5
37.9 ± 11.1
10.0 ± 1.1
8.0 ± 2.9
10.1 ± 2.4
29.9 ± 15.4
*SL6
#
#
#
#
#
SL4
29.2 ± 7.4
189.0 ± 48.8
5.8 ± 2.5
2.9 ± 0.5
8.3 ± 4.1
1.3 ± 0.2
26.0 ± 5.9
19.9 ± 1.0
4.7 ± 2.1
5.7 ± 2.0
37.0 ± 8.1
47.3 ± 2.2
3.6 ± 0.7
4.2 ± 1.0
33.4 ± 7.6
43.0 ± 3.1
*SL6#
*SL6
*SL6
*SL6 #
*SL6
*SL6
SL6
8.4 ± 6.8
180.0 ± 2.0
7.0 ± 7.4
3.0 ± 2.8
15.0 ± 1.7
1.5 ± 0.3
26.2 ± 2.9
19.1 ± 3.7
5.7 ± 1.3
7.7 ± 0.5
15.4 ± 8.7
37.8 ± 1.6
6.8 ± 3.6
5.5 ± 0.4
8.6 ± 12.0
32.3 ± 1.8
*SL0, SL2
*SL4#
#
*SL0, SL4
*SL4 #
#
*SL4
#
*SL4
#
Note: The values were expressed as means ± SD.
*The differences between workload groups (SST, LST accordingly) are significant at <0.05 for the Kruskal– Wallis ANOVA test.
#The differences between SST and LST groups are significant at <0.05 for the Mann– Whitney (U) test.
6 of 12 |
POTOLITSYNA et al.
flatter slope than the SST curve and was characterized by
a more gradual increase in the concentration of lactate in
the blood with the increasing weight of the attached loads.
The most significant increase in this indicator occurred
when the rats performed the test at a load of 6% BW. There-
fore, the AT for the LST group was set at a load of 5% BW.
The concentrations of lactate, glucose, and cortisol in
arteriovenous blood collected after decapitation of rats
were higher in SST rats than in LST rats (Table 3).
Urea levels in the SST and LST groups showed no signif-
icant differences. The NOx levels were significantly higher
in SST rats than in LST rats. The levels of NO2 and NO3
did not reveal significant differences in the arteriovenous
blood of either group of rats. However, we observed higher
values of both metabolites in SST rats than in LST rats.
4 | DISCUSSION
Only a few studies have described swimming Tex in rats
(Beck & Gobatto, 2013; Travassos et al., 2018; Venditti &
Di Meo, 1996). Some groups used arbitrary loads without
describing the reasoning for weight choice (Travassos
et al., 2018; Venditti & Di Meo, 1996). Other research-
ers used a method in which the load to exhaustion was
based on a preliminary calculation of AT. For example,
Beck et al. (2014) used the minimum lactate level on the
“U- shaped” lactate curve obtained after the blood LMT as
the AT. The loads that the rats were subjected to in this
test were then used as benchmarks for the swimming to
exhaustion experiment. However, the method of calculat-
ing AT significantly affects the interpretation of the re-
sults and makes it impossible to compare the results with
those of other studies. Therefore, we evaluated Tex in rats
swimming to exhaustion using various loads in the first
stage of our study (Table 1). Predictably, the increase in
the load weight affected the duration of the swim and re-
duced it from several hours to several minutes. However,
the swimming time also strongly depended on the endur-
ance of the rats. For most load weights, Tex was divisible
into two groups (Table 2): SST and LST. Loads heavier
than 8% BW were too heavy for untrained rats (Gobatto
FIGURE 2 Blood lactate levels in tail vein blood of rats from SST and LST groups. SST— swimming for short time, LST— swimming for
long time. The values are expressed as means ± SD. The statistical significance of differences between SST and LST groups was estimated
using Mann– Whitney (U) test; p- values were considered significant at *p < 0.05. The statistical significance of differences between workload
groups within corresponding group (SST or LST accordingly) was estimated using the Friedman ANOVA and Kendall coefficient of
concordance, and is shown in italics. When necessary, the Newman– Keuls post hoc comparison test was used. Statistical significance is
indicated in comparison with the specified load weight at #p < 0.05.
p=0.0021
p=0.0019
#0
#0,2,3
#0,2,3
#0,2,3
*
*
*
*
#0,2
#0,2,3,4
2
3
4
5
6
7
8
0
1
2
3
4
5
6
7
Lactate, mmol/l
Load weight (% of body mass)
SST
LST
TABLE 3 Biochemical parameters of arteriovenous blood of sacrificed rats after performing the last load (6% BW).
Groups
Lactate,
mmol/L
Glucose,
mmol/L
Cortisol, ng/
mL
Urea,
mmol/L
NOx,
μmol/L
NO2,
μmol/L
NO3,
μmol/L
SST
10.3 ± 4.4
9.5 ± 0.8
34.2 ± 9.3
4.0 ± 1.0
28.1 ± 1.8
9.5 ± 1.6
18.6 ± 3.1
LST
6.3 ± 0.7
8.5 ± 0.7
19.1 ± 3.2
4.5 ± 1.1
24.3 ± 1.8
8.0 ± 1.1
16.3 ± 2.9
p- valueSST- LST
0.049
0.126
0.049
0.512
0.049
0.126
0.512
Note: Data are presented as the means ± SD. p- valueSST- LST— The statistical significance of differences between short swimming time (SST) and long swimming
time (LST) groups was estimated using the Mann– Whitney (U) test. p < 0.05 are shown in bold.
| 7 of 12
POTOLITSYNA et al.
et al., 2001), and no differences were observed in this case.
Despite the lack of similar studies in the literature, there
are fragmentary results on rats swimming to exhaustion.
For example, Venditti and Di Meo (1996) showed a Tex
close to our results (294 ± 32 min) in untrained rats (swim-
ming to exhaustion) with a load of 2% of BW. Beck and
Gobatto (2013) reported that rats swam 108 ± 46 min with
a load weight of 5% BW, which is less than the load of
4%– 6% BW in our study. A greater variation in Tex was
described in Travassos et al. (2018).
Rats in the Travassos et al. (2018) study swam at a load
of 6% BW from 3 to 22 min and were divided into low per-
formance (Tex = 3– 12 min) and high performance (Tex = 12–
22 min) groups depending on the time of exhaustion.
Notably, the authors excluded all rats that swam longer than
22 min. The variability in the duration of swimming with
the load of 6% BW was larger in our study, and therefore, the
division into endurance groups was different. All rats that
swam less than 21 min with a load of 6% BW were included
in the SST group, and rats that swam approximately 180 min
were included in the LST group. Therefore, the study of un-
trained rats using various load weight protocols allows com-
parison of metabolic changes in LST and SST animals and
identifies the best load weight for the test.
We showed that the blood biochemistry changes in dif-
ferent load weights and the time of swimming to exhaus-
tion also significantly differed between the LST and SST
rats. The SST rats had higher levels of lactate, glucose and
cortisol, and the LST rats had higher levels of urea and
nitric oxide. Notably, these differences became more ob-
vious with increases in the load weight while swimming.
Despite all of the rats being of the same age, being
housed in the same conditions, being fed the same diet,
and having no previous training, some rats showed higher
inborn physical endurance. These results are fully con-
sistent with the hypothesis that genetics is an important
determinant of the response to physical activity (Koch
et al., 2005) and may affect the features of anatomy
(Britton & Koch, 2001), pulmonary function (Kirkton
et al., 2009), insulin response (Schwarzer et al., 2021), and
the predominant type of skeletal muscle fibers (Abernethy
et al., 1990). It was expected that the metabolism in rats
with different physical endurance would be different. A
significant increase in blood lactate at the lower loads in
SST rats reflects lower aerobic capacities, and hypoxia oc-
curs faster in these rats under high- intensity physical exer-
cise. Howlett et al. (2009) showed that SST rats had VO2max
and oxygen transfer in skeletal muscles that was 50% lower
than those in higher endurance rats, despite having higher
absolute muscle mass. The maintenance of glucose levels
in hypoxia is provided primarily by glycolysis and glycog-
enolysis (Brooks & Mercier, 1994; Emhoff et al., 2013).
With sufficient oxygen supply during prolonged physical
exercise, there is a higher fat utilization. The increased
contribution of lipids to energy metabolism makes it pos-
sible to significantly increase endurance during physical
exercise (Brooks & Mercier, 1994; Nosaka et al., 2009).
There are more data on a more complex system of regu-
lation of lipid metabolism depending on the intensity of
exercise (Lyudinina et al., 2018; Romijn et al., 1993).
Physical exercises also stimulate increases in cortisol
levels. This hormone plays a significant role in acceler-
ating lipolysis, ketogenesis, and proteolysis (Del Corral
et al., 1998). The level of cortisol increases in proportion to
the intensity of exercise, but the final level depends on the
total duration of exercise. Moderate- and high- intensity
exercises increase the levels of circulating cortisol. In con-
trast, low- intensity exercise does not lead to an increase
in cortisol levels (Del Corral et al., 1998; Hill et al., 2008).
The levels of cortisol did not show significant differ-
ences with respect to Tex in our study. However, cortisol
levels were higher in SST rats than in LST rats, especially
at loads of 2%– 4% BW. Perhaps, this result occurred be-
cause of the different behaviors of rats when performing
the test and the levels of individual stress. Glucocorticoids
in rodents are often used as biomarkers of stress, with
cortisol reacting faster during severe acute stress, unlike
corticosterone, which is associated more with adaptation
during chronic stress (Gong et al., 2015).
NO is another metabolite that allows adaptation to
significant physical exercise (Oral, 2021). It is a signaling
molecule with a wide variety of effects in mammals, the
most well- known of which is the regulation of local vaso-
motor tone and resistance to microvascular flow (Baskurt
et al., 2011). Skeletal muscles of rodents contain unusu-
ally high concentrations of nitrates compared to blood and
other tissues, which indicates the high importance of ni-
tric oxide for their body (Piknova et al., 2015). Nitric oxide
has an extremely short half- life of only a few milliseconds
in biological tissues, and it is important that it is con-
stantly produced at its sites of effect (Jones et al., 2021).
Experimental data indicate that physical exercises lead to
an increase in the enzymatic synthesis of nitric oxide and
activation of the associated vascular control mechanisms
(Baskurt et al., 2011). We previously found a positive cor-
relation between nitrogen oxide and lactate at the AT and
a negative correlation at maximum load in elite cross-
country skiers possessing high endurance (Parshukova
et al., 2020). The higher level of NOx we obtained in LST
rats, but not SST rats, is consistent with these findings. It
characterizes a more adequate response of the vascular
bed in response to physical exercise and allows better con-
trol of vascular tone for a longer time. The increase in NOx
levels in LST rats was observed primarily due to the NO3
fraction. Under conditions of normal and increased oxy-
gen consumption by tissues, NO is formed enzymatically
8 of 12 |
POTOLITSYNA et al.
via the oxidation of L- arginine, and the final metabolite of
this process is primarily NO3 (Cubrilo et al., 2011).
Our study showed that the use of a load weight of 4%
BW was the most informative for studying the level of
physical endurance in untrained rats. At this load weight
and a background of a wide Tex range, the rats showed
significant changes in most of the biochemical indices as-
sessed in our study, which included the most informative
indices in relation to the problem under discussion.
4.1 | Anaerobic threshold
There are a large number of methods for studying the
lactate threshold in rats (Faude et al., 2009). The choice
of method depends on the research goals. However, it
is also important to consider the natural abilities of rats
and our capacity to project the results on humans in the
future. The optimal method for determining the AT in
rats is swimming with increasing load. Several studies
of the lactate threshold in swimming rats found that
loads of 4%– 6% BW were more often used (Contarteze
et al., 2008; Gobatto et al., 2001; Voltarelli et al., 2002).
For example, Gobatto et al. (2001) showed that the AT
corresponded to 6% BW at a blood lactate concentra-
tion of 5.5 mmol/L. Another study established the AT
at a weight of 4.0% BW and a lactate level of 5.2 mmol/L
(Abreu et al., 2016). Similar lactate values were shown
at an AT with a load of 4.5% BW (Zhouab et al., 2018).
However, these studies do not mention individual en-
durance variation in rats. The results of our study
showed that this characteristic of laboratory animals
may significantly shift the AT to the left or the right on
the lactate curve. The ATs in SST and LST rats were 3%
BW and 5% BW, respectively. The lactate curve of SST
was less flat than that of LST. For increasing endur-
ance, it is generally recognized that a shift of the lactate
curve to the right is interpreted as an increase in physi-
cal performance, and a shift to the left is considered a
deterioration in endurance (Abreu et al., 2016; Faude
et al., 2009). The lactate concentration at the AT was
also different and higher in SST rats than in LST rats
(5.8 mmol/L vs. 5.2 mmol/L). A lower lactate level at the
end of physical exercise in LST rats may be associated
with a lower rate of lactate accumulation and/or a lower
metabolic clearance of lactate (Donovan & Brooks, 1983;
Yang et al., 2020). Higher endurance augments capaci-
ties for lactate production, disposal, and clearance (Mes-
sonnier et al., 2013). Our data are generally consistent
with the results of other studies, although no data on the
AT when swimming in rats with a load below 4% were
found. However, rats with higher endurance were likely
included for various reasons in the described studies
(Abreu et al., 2016; Contarteze et al., 2008; Gobatto
et al., 2001; Voltarelli et al., 2002; Zhouab et al., 2018).
Biochemical data from the arteriovenous rat blood as-
says (Table 3) also demonstrated significant differences
between the SST and LST rats. Because the collection of
arteriovenous blood occurred within 3– 5 min after the last
collection of blood from the caudal vein, the data on lactate
from arteriovenous blood showed higher values relative to
lactate from the caudal vein. The most significant increase
was observed in SST rats, which reflected their lower re-
covery abilities compared to LST animals. Glucose, corti-
sol, and NOx levels were also significantly higher in the
SST rats. This pattern of blood biochemistry generally
characterizes more significant rearrangements and higher
stress levels in SST rats than in LST rats at a similar level
of physical exercise. Notably, the increase in NOx levels
in this test occurred due to an increase in nitrites (NO2),
unlike in swimming to exhaustion. Under hypoxic con-
ditions, NO2 is an alternative source of nitric oxide syn-
thesis (Gladwin et al., 2000; Schulman & Hare, 2012) and
participates in adaptation to hypoxia caused, for example,
by physical exertion (Gladwin et al., 2000). The current
understanding of nitrite- dependent mechanisms of adap-
tation to hypoxia is based on data on the reduction of NO2
by oxygen- dependent and hypoxic nitrite reductase (Glad-
win & Kim- Shapiro, 2008).
NO is a mediator of skeletal muscle function and af-
fects cellular respiration and contractility. In working skel-
etal muscle, inhibition of NOS improves the economy of
muscle contraction, decreases the outflow of lactate from
the muscles, and reduces the oxygen cost (Krause & Van
Etten, 2005). Thiol groups, reactive metal ions in the pro-
teins' active centers, can interact with NO, which leads to
various responses and further biological events in skeletal
muscles. NO- mediated reactions inhibit heme- containing
proteins, such as cytochrome C oxidase, thus interfering
with the function of cytochrome C oxidase in cell respi-
ration (Borutaite & Brown, 1996). Inhibition of this en-
zyme and of the sarcoplasmic reticulum Ca2+- ATPase in
fast- twitch and slow- twitch skeletal muscle fibers by NOS-
generated NO may also lead to inhibition of mitochondrial
respiration in skeletal muscle (Klebl et al., 1998). More-
over, aconitase and respiratory chain complex I can also
be targeted by NO (Clementi et al., 1998). NO is crucial
for the activation and inhibition of ryanodine receptors
(RyRs) (Stamler & Meissner, 2001), which play a decisive
role in the release of Ca2+ into the cytosol and therefore
in muscle excitation and contraction (Mazzone & Carme-
liet, 2008). In our experimental work, we have shown that
elite athletes (cross- country skiers) have an NO- dependent
mechanism for regulating lactate levels during aerobic ex-
ercise, especially when working at the AT. In particular, at
the AT, we have revealed a positive relationship between
| 9 of 12
POTOLITSYNA et al.
NOx (nitric oxide metabolites) and blood lactate, with that
relationship being reversed at maximum load. This obser-
vation suggests the existence of an adaptive mechanism
for regulating lactate levels at the AT in highly qualified
cross- country skiers (Parshukova et al., 2022).
Therefore, our data provide a new understanding of
the role of NO- dependent mechanisms in the phenome-
non of AT.
5 | CONCLUSION
We found that the level of individual endurance signifi-
cantly affected the AT in untrained rats. The AT in SST
rats and 5% BW in LST rats. These groups also had dif-
ferent blood biochemistry profiles at the AT and after
swimming to exhaustion. There was a shift in the AT to
the right side on the lactate curve in the zone of the AT
in LST rats compared to SST rats, and the levels of lactate,
glucose, cortisol, and NOx were lower. At the end of the
exercise to exhaustion, SST rats had higher blood levels
of lactate, glucose, and cortisol, and LST rats had higher
levels of urea and NOx.
AUTHOR CONTRIBUTIONS
Evgeny Bojko, Nadezhda Vakhnina, Natalya Potolit-
syna conceives and designed the experiments; Natalya
Potolitsyna, Nadezhda Vakhnina, Nadezhda Alisyl-
tanova, Lubov Kalikova, Anastasia Tretyakova, Alexey
Chernykh, Vera Shadrina, Arina Duryagina performed
experiments; Natalya Potolitsyna, Olga Parshukova,
Lubov Kalikova analysed data; Natalya Potolitsyna,
Olga Parshukova, Lubov Kalikova, Nadezhda Vakhnina
interpreted results of experiments; Natalya Potolitsyna
prepared figures; Natalya Potolitsyna, Alexey Chernykh
drafted
manuscript;
Natalya
Potolitsyba,
Evgeny
Bojko, Olga Parshukova, Nadezhda Vakhnina, Alexey
Chernykh edited and revised manuscript; Natalya Po-
tolitsyna, Evgeny Bojko, Olga Parshukova, Nadezhda
Vakhnina, Alexey Chernykh approved final version of
manuscript.
ACKNOWLEDGMENTS
The study was conducted within the framework of the
research work of the Institute of Physiology of Kоmi Sci-
ence Centre of the Ural Branch of the Russian Academy of
Sciences, FRC Komi SC UB RAS, FUUU- 2022- 0063 (No.
1021051201877- 3).
FUNDING INFORMATION
This research did not receive any specific grant from fund-
ing agencies in the public, commercial, or not- for- profit
sectors.
CONFLICT OF INTEREST STATEMENT
No conflicts of interest, financial or otherwise, are de-
clared by the authors.
ETHICS STATEMENT
The Ethics Committee of the Institute of Physiology of the
Russian Academy of Sciences, Syktyvkar approved the
experimental design and protocol of the study. The study
was performed in accordance with the ethical standards
as laid down in the 1964 Declaration of Helsinki and its
later amendments.
ORCID
Natalya Potolitsyna
https://orcid.
org/0000-0003-4804-6908
Evgeny Bojko
https://orcid.org/0000-0002-8027-898X
REFERENCES
Abernethy, P., Thayer, R., & Taylor, A. W. (1990). Acute and chronic
responses of skeletal- muscle to endurance and sprint exer-
cise— A review. Sports Medicine, 10(6), 365– 389. https://doi.
org/10.2165/00007 256- 19901 0060- 00004
Abreu, P., Mendes, S. V. D., Leal- Cardoso, J. H., & Ceccatto, V. M.
(2016). Anaerobic threshold employed on exercise training
prescription and performance assessment for laboratory ro-
dents: A short review. Life Sciences, 15(151), 1– 6. https://doi.
org/10.1016/j.lfs.2016.02.016
Baskurt, O. K., Ulker, P., & Meiselman, H. J. (2011). Nitric oxide,
erythrocytes
and
exercise.
Clinical Hemorheology and
Microcirculation, 49(1– 4), 175– 181. https://doi.org/10.3233/
ch- 2011- 1467
Beck, W., & Gobatto, C. (2013). Effects of maximum intensity aer-
obic swimming exercise until exhaustion at different times of
day on the hematological parameters in rats. Acta Physiologica
Hungarica, 100(4), 427– 434. https://doi.org/10.1556/aphys
iol.100.2013.013
Beck, W. R., De Araujo, G. G., Scariot, P. P. M., dos Reis, I. G. M., &
Gobatto, C. A. (2014). Time to exhaustion at anaerobic threshold
in swimming rats: Metabolic investigation. Bratislavské Lekárske
Listy, 115(10), 617– 621. https://doi.org/10.4149/bll_2014_119
Borutaite, V., & Brown, G. C. (1996). Rapid reduction of nitric oxide
by mitochondria, and reversible inhibition of mitochondrial
respiration by nitric oxide. The Biochemical Journal, 1, 295– 299.
https://doi.org/10.1042/bj315 0295
Bosquet, L., Léger, L., & Legros, P. (2002). Methods to determine
aerobic endurance. Sports Medicine, 32, 675– 700. https://doi.
org/10.2165/00007 256- 20023 2110- 00002
Brito, A. F., Silva, A. S., Souza, I. L. L., Pereira, J. C., Martins, I. R. R.,
& Silva, B. A. (2015). Intensity of swimming exercise influences
tracheal reactivity in rats. Journal of Smooth Muscle Research,
51, 70– 81. https://doi.org/10.1540/jsmr.51.70
Britton, S. L., & Koch, L. G. (2001). Animal genetic models for com-
plex traits of physical capacity. Exercise and Sport Sciences
Reviews, 29(1), 7– 14. https://doi.org/10.1097/00003 677- 20010
1000- 00003
Brooks, G. A., & Mercier, J. (1994). Balance of carbohydrate and lipid
utilization during exercise: The “crossover” concept. Journal of
10 of 12 |
POTOLITSYNA et al.
Applied Physiology, 76(6), 2253– 2261. https://doi.org/10.1152/
jappl.1994.76.6.2253
Chicharro, J. L., Pérez, M., Carvajal, A., Bandrés, F., & Lucía, A.
(1999). The salivary amylase, lactate and electromyographic
response to exercise. The Japanese Journal of Physiology, 49(6),
551– 554. https://doi.org/10.2170/jjphy siol.49.551
Chimin, P., Almeida, F. N., Okuno, N. M., Franzói de Moraes, S. M.,
Gobatto, C. A., & Nakamura, F. Y. (2013). Critical load forced-
swim test with Wistar rats does not properly estimate anaerobic
threshold: The relationship with morphophysiological factors
and performance indices. Science & Sports, 28(3), e51– e57.
https://doi.org/10.1016/j.scispo.2012.08.002
Cholewa, J., Guimarães- Ferreira, L., da Silva, T. T., et al. (2014). Basic
models modeling resistance training: An update for basic sci-
entists interested in study skeletal muscle hypertrophy. Journal
of Cellular Physiology, 229, 1148– 1156. https://doi.org/10.1002/
jcp.24542
Clementi, E., Brown, G. C., Feelisch, M., & Moncada, S. (1998).
Persistent inhibition of cell respiration by nitric oxide: Crucial
role of snitrosylation of mitochondrial complex I and protective
action of glutathione. Proceedings of the National Academy of
Sciences of the United States of America, 95, 7631– 7636. https://
doi.org/10.1073/pnas.95.13.7631
Contarteze, R. V. L., Manchado, F. B., Gobatto, C. A., & De
Mello, M. A. R. (2008). Stress biomarkers in rats submitted
to swimming and treadmill running exercises. Comparative
Biochemistry and Physiology. Part A: Molecular & Integrative
Physiology,
151(3),
415– 422.
https://doi.org/10.1016/j.
cbpa.2007.03.005
Cubrilo, D., Djordjevic, D., Zivkovic, V., Djuric, D., Blagojevic, D.,
Spasic, M., & Jakovljevic, V. (2011). Oxidative stress and nitrite
dynamics under maximal load in elite athletes: Relation to
sport type. Molecular and Cellular Biochemistry, 355(1– 2), 273–
279. https://doi.org/10.1007/s1101 0- 011- 0864- 8
Davies, C. T. M., Few, J., Foster, K. G., & Sargeant, A. J. (1974).
Plasma catecholamine concentration during dynamic exercise
involving different muscle groups. European Journal of Applied
Physiology and Occupational Physiology, 32(3), 195– 206. https://
doi.org/10.1007/bf004 23215
De Araujo, G. G., Papoti, M., dos Reis, I. G. M., De Mello, M. A. R.,
& Gobatto, C. A. (2016). Short and long term effects of high-
intensity interval training on hormones, metabolites, antioxi-
dant system, glycogen concentration, and aerobic performance
adaptations in rats. Frontiers in Physiology, 7, 505. https://doi.
org/10.3389/fphys.2016.00505
Del Corral, P., Howley, E. T., Hartsell, M., & Ashraf, M. (1998).
Metabolic effects of low cortisol during exercise in humans.
Journal of Applied Physiology, 84(3), 939– 947. https://doi.
org/10.1152/jappl.1998.84.3.939
Donovan, C. M., & Brooks, G. A. (1983). Endurance training affects
lactate clearance, not lactate production. The American Journal
of Physiology, 244(1), E83– E92. https://doi.org/10.1152/ajpen
do.1983.244.1.E83
Dos Reis, I. G. M., Martins, L. E. B., De Araujo, G. G., & Gobatto, C.
A. (2018). Forced swim reliability for exercise testing in rats by a
tethered swimming apparatus. Frontiers in Physiology, 9, 1839.
https://doi.org/10.3389/fphys.2018.01839
Emhoff, C. A. W., Messonnier, L. A., Horning, M. A., Fattor, J. A.,
Carlson, T. J., & Brooks, G. A. (2013). Gluconeogenesis and
hepatic glycogenolysis during exercise at the lactate threshold.
Journal of Applied Physiology, 114(3), 297– 306. https://doi.
org/10.1152/jappl physi ol.01202.2012
Faria, V. S., Pejon, T. M. M., Gobatto, C. A., de Araujo, G. G.,
Cornachione, A. S., & Beck, W. R. (2021). Acute melatonin
administration improves exercise tolerance and the metabolic
recovery after exhaustive effort. Scientific Reports, 11(1), 19228.
https://doi.org/10.1038/s4159 8- 021- 97364 - 7
Farrell, J. W., 3rd, Lantis, D. J., Ade, C. J., Cantrell, G. S., & Larson,
R. D. (2018). Aerobic exercise supplemented with muscular en-
durance training improves onset of blood lactate accumulation.
Journal of Strength and Conditioning Research, 32(5), 1376–
1382. https://doi.org/10.1519/JSC.00000 00000 001981
Faude, O., Kindermann, W., & Meyer, T. (2009). Lactate threshold
concepts: How valid are they? Sports Medicine, 39(6), 469– 490.
https://doi.org/10.2165/00007 256- 20093 9060- 00003
Ghosh, A. K. (2004). Anaerobic threshold: Its concept and role in en-
durance sport. Malaysian Journal of Medical Sciences, 11(1), 24–
36. https://www.ncbi.nlm.nih.gov/pmc/artic les/PMC34 38148/
Gladwin, M., Shelhamer, J., Schechter, A., et al. (2000). Role of cir-
culating nitrite and S- nitrosohemoglobin in the regulation of
the regional blood flow in humans. Proceedings of the National
Academy of Sciences of the United States of America, 97, 11482–
11486. https://doi.org/10.1073/pnas.97.21.11482
Gladwin, M. T., & Kim- Shapiro, D. B. (2008). The functional nitrite
reductase activity of the heme- globins. Blood, 112, 2636– 2647.
https://doi.org/10.1182/blood - 2008- 01- 115261
Gobatto, C. A., de Mello, M. A., Sibuya, C. Y., De Azevedo, J. R., Dos
Santos, L. A., & Kokubun, E. (2001). Maximal lactate steady
state in rats submitted to swimming exercise. Comparative
Biochemistry and Physiology. Part A: Molecular & Integrative
Physiology,
130(1),
21– 27.
https://doi.org/10.1016/s1095
- 6433(01)00362 - 2
Gong, S., Miao, Y. L., Jiao, G.- Z., Sun, M.- J., LiH, L. J., Luo, M.- J.,
& Tan, J. H. (2015). Dynamics and correlation of serum corti-
sol and corticosterone under different physiological or stress-
ful conditions in mice. PLoS One, 10(2), e0117503. https://doi.
org/10.1371/journ al.pone.0117503
Goutianos, G., Tzioura, A., Kyparos, A., Paschalis, V., Margaritelis,
N. V., Veskoukis, A. S., Zafeiridis, A., Dipla, K., Nikolaidis, M.
G., & Vrabas, I. S. (2015). The rat adequately reflects human
responses to exercise in blood biochemical profile: A compar-
ative study. Physiological Reports, 3(2), e12293. https://doi.
org/10.14814/ phy2.12293
Greek, R., Menache, A., & Rice, M. J. (2012). Animal models in an
age of personalized medicine. Personalized Medicine, 9, 47– 64.
https://doi.org/10.2217/pme.11.89
Halson, S. L., & Jeukendrup, A. E. (2004). Does overtraining exist?
An analysis of overreaching and overtraining research. Sports
Medicine, 34(14), 967– 981. https://doi.org/10.2165/00007 256-
20043 4140- 00003
Heck, H., Mader, A., Hess, G., Mück, E. S., Müller, R., & Hollmann,
W. (1985). Justification of the 4- mmol/L lactate threshold.
International Journal of Sports Medicine, 6, 117– 130. https://
doi.org/10.1055/s- 2008- 1025824
Hill, E. E., Zack, E., Battaglini, C., Viru, M., Viru, A., & Hackney,
A. C. (2008). Exercise and circulating cortisol levels: The inten-
sity threshold effect. Journal of Endocrinological Investigation,
31(7), 587– 591. https://doi.org/10.1007/BF033 45606
Howlett, R. A., Kirkton, S. D., Gonzalez, N. C., Wagner, H. E.,
Britton, S. L., Koch, L. G., & Wagner, P. D. (2009). Peripheral
| 11 of 12
POTOLITSYNA et al.
oxygen transport and utilization in rats following continued
selective breeding for endurance running capacity. Journal of
Applied Physiology, 106(6), 1819– 1825. https://doi.org/10.1152/
jappl physi ol.00914.2007
Jones, A. M., Vanhatalo, A., Seals, D. R., Rossman, M. J., Piknova,
B., & Jonvik, K. L. (2021). Dietary nitrate and nitric oxide me-
tabolism: Mouth, circulation, skeletal muscle, and exercise per-
formance. Medicine and Science in Sports and Exercise, 53(2),
280– 294. https://doi.org/10.1249/MSS.00000 00000 002470
Kirkton, S. D., Howlett, R. A., Gonzalez, N. C., Giuliano, P. G., Britton,
S. L., Koch, L. G., Lauren, G., & Wagner, P. D. (2009). Continued
artificial selection for running endurance in rats is associated with
improved lung function. Journal of Applied Physiology, 106(6),
1810– 1818. https://doi.org/10.1152/jappl physi ol.90419.2008
Klebl, B. M., Ayoub, A. T., & Pette, D. (1998). Protein oxidation,
tyrosine nitration, and inactivation of sarcoplasmic reticu-
lum Ca21- ATPase in low- frequency stimulated rabbit muscle.
FEBS Letters, 422, 381– 384. https://doi.org/10.1016/S0014
- 5793(98)00053 - 2
Koch, L. G., Green, C. L., & Britton, S. L. (2005). Test of the prin-
ciple of initial value in rat genetic models of exercise capac-
ity. American Journal of Physiology. Regulatory, Integrative
and Comparative Physiology, 288(2), 466– R472. https://doi.
org/10.1152/ajpre gu.00621.2004
Krause, D. S., & Van Etten, R. A. (2005). Tyrosine kinases as targets
for cancer therapy. The New England Journal of Medicine, 353,
172– 187. https://doi.org/10.1056/NEJMr a044389
Lyudinina, A. Y., Ivankova, G. E., & Bojko, E. R. (2018). Priority use
of medium- chain fatty acids during high- intensity exercise in
cross- country skiers. Journal of the International Society of Sports
Nutrition, 15(1), 57. https://doi.org/10.1186/s1297 0- 018- 0265- 4
Mazzone, M., & Carmeliet, P. (2008). Drug discovery: A lifeline
for suffocating tissues. Nature, 453, 1194– 1195. https://doi.
org/10.1038/4531194a
Messonnier, L. A., Emhoff, C. A. W., Fattor, J. A., Horning, M. A.,
Carlson, T. J., & Brooks, G. A. (2013). Lactate kinetics at the lac-
tate threshold in trained and untrained men. Journal of Applied
Physiology, 114(11), 1593– 1602. https://doi.org/10.1152/jappl
physi ol.00043.2013
Nosaka, N., Suzuki, Y., Nagatoishi, A., Kasai, M., Wu, J., & Taguchi,
M. (2009). Effect of ingestion of medium- chain triacylglycerols
on moderate- and high- intensity exercise in recreational ath-
letes. Journal of Nutritional Science and Vitaminology, 55(2),
120– 125. https://doi.org/10.3177/jnsv.55.120
Oral, O. (2021). Nitric oxide and its role in exercise physiology. The
Journal of Sports Medicine and Physical Fitness, 61(9), 1208–
1211. https://doi.org/10.23736/ S0022 - 4707.21.11640 - 8
Papadopoulos, C., Doyle, J. A., & La Budde, B. D. (2006). Relationship
between running velocity of 2 distances and various lac-
tate parameters. International Journal of Sports Physiology
and Performance, 1(3), 270– 283. https://doi.org/10.1123/
ijspp.1.3.270
Parshukova, O. I., Varlamova, N. G., & Bojko, E. R. (2020). Nitric
oxide production in professional skiers during physical activity
at maximum load. Frontiers in Cardiovascular Medicine, 7, 1– 8.
https://doi.org/10.3389/fcvm.2020.582021
Parshukova, O. I., Varlamova, N. G., Potolitsyna, N. N., Lyudinina,
A. Y., & Bojko, E. R. (2022). Features of metabolic support of
physical performance in highly trained cross- country skiers of
different qualifications during physical activity at maximum
load. Cell, 11, 39. https://doi.org/10.3390/cells 11010039
Piknova, B., Park, J. W., Swanson, K. M., Dey, S., Noguchi, C. T., &
Schechter, A. N. (2015). Skeletal muscle as an endogenous ni-
trate reservoir. Nitric Oxide: Biology and Chemistry, 47, 10– 16.
https://doi.org/10.1016/j.niox.2015.02.145
Rice, J. (2012). Animal models: Not close enough. Nature, 484(7393),
9. https://doi.org/10.1038/natur e11102
Roecker, K., Schotte, O., Niess, A. M., Horstmann, T., & Dickhuth,
H. H. (1998). Predicting competition performance in long-
distance running by means of a treadmill test. Medicine and
Science in Sports and Exercise, 30(10), 1552– 1557. https://doi.
org/10.1097/00005 768- 19981 0000- 00014
Romijn, J. A., Coyle, E. F., Sidossis, L. S., Gastaldelli, A., Horowitz,
J. F., Endert, E., & Wolfe, R. R. (1993). Regulation of endoge-
nous fat and carbohydrate metabolism in relation to exercise
intensity and duration. The American Journal of Physiology,
265(3 Pt 1), E380– E391. https://doi.org/10.1152/ajpen
do.1993.265.3.E380
Schulman, I. H., & Hare, J. M. (2012). Regulation cardiovascular pro-
cesses by S- nitrosylation. Biochimica et Biophysica Acta, 1820,
752– 762. https://doi.org/10.1016/j.bbagen.2011.04.002
Schwarzer, M., Molis, A., Schenkl, C., Schrepper, A., Britton, S. L.,
Koch, L. G., & Doenst, T. (2021). Genetically determined ex-
ercise capacity affects systemic glucose response to insulin
in rats. Physiological Genomics, 53(9), 395– 405. https://doi.
org/10.1152/physi olgen omics.00014.2021
Solli, G. S., Tønnessen, E., & Sandbakk, Ø. (2017). The training char-
acteristics of the world's most successful female cross- country
skier. Frontiers in Physiology, 8, 1069. https://doi.org/10.3389/
fphys.2017.01069
Stamler, J. S., & Meissner, G. (2001). Physiology of nitric oxide in
skeletal muscle. Physiological Reviews, 81, 209– 237. https://doi.
org/10.1152/physr ev.2001.81.1.209
Støren, Ø., Rønnestad, B. R., Sunde, A., Hansen, J., Ellefsen, S., &
Helgerud, J. (2014). A time- saving method to assess power
output at lactate threshold in well- trained and elite cyclists.
Journal of Strength and Conditioning Research, 28, 622– 629.
https://doi.org/10.1519/JSC.0b013 e3182 a73e70
Tanji, F., & Nabekura, Y. (2019). Oxygen uptake and respiratory ex-
change ratio relative to the lactate threshold running in well-
trained distance runners. The Journal of Sports Medicine and
Physical Fitness, 59(6), 895– 901. https://doi.org/10.23736/
S0022 - 4707.18.08828 - X
Travassos, P. B., Godoy, G., De Souza, H. M., Curi, R., & Bazotte,
R. B. (2018). Performance during a strenuous swimming ses-
sion is associated with high blood lactate: Pyruvate ratio and
hypoglycemia in fasted rats. Brazilian Journal of Medical and
Biological Research, 51(5), 7057. https://doi.org/10.1590/1414-
431X2 0187057
Venditti, P., & Di Meo, S. (1996). Antioxidants, tissue damage, and
endurance in trained and untrained young male rats. Archives
of Biochemistry and Biophysics, 331(1), 63– 68. https://doi.
org/10.1006/abbi.1996.0283
Voltarelli, F. A., Gobatto, C. A., & De Mello, M. A. R. (2002).
Determination of anaerobic threshold in rats using the lac-
tate minimum test. Brazilian Journal of Medical and Biological
Research, 35(11), 1389– 1394. https://doi.org/10.1590/S0100
- 879X2 00200 1100018
12 of 12 |
POTOLITSYNA et al.
Wakayoshi, K., Yoshida, T., Udo, M., Harada, T., Moritani, T., Mutoh,
Y., & Miyashita, M. (1993). Does critical swimming velocity
represent exercise intensity at maximal lactate steady- state.
European Journal of Applied Physiology and Occupational
Physiology, 66(1), 90– 95. https://doi.org/10.1007/BF008 63406
Weyand, P., Cureton, K., Conley, D., Sloniger, M., & Liu, Y. (1994).
Peak oxygen deficit predicts sprint and middle- distance track
performance. Medicine and Science in Sports and Exercise,
26(9), 1174– 1180.
Yang, W. H., Park, H., Grau, M., & Heine, O. (2020). Decreased blood
glucose and lactate: Is a useful indicator of recovery ability in ath-
letes? International Journal of Environmental Research and Public
Health, 17(15), 5470. https://doi.org/10.3390/ijerp h1715 5470
Zhouab, Q., Huangc, Z.- G., Zhuab, X.- J., Xieb, Z.- H., Yaob, T.- F.,
Wangb, Y.- H., & Liab, J.- H. (2018). Effects of nicotinamide N-
methyltransferase (NNMT) inhibition on the aerobicand the
anaerobic endurance exercise capacity. Science & Sports, 33(4),
e159– e165. https://doi.org/10.1016/j.scispo.2018.02.006
How to cite this article: Potolitsyna, N.,
Parshukova, O., Vakhnina, N., Alisultanova, N.,
Kalikova, L., Tretyakova, A., Chernykh, A.,
Shadrina, V., Duryagina, A., & Bojko, E. (2023).
Lactate thresholds and role of nitric oxide in male
rats performing a test with forced swimming to
exhaustion. Physiological Reports, 11, e15801.
https://doi.org/10.14814/phy2.15801
| Lactate thresholds and role of nitric oxide in male rats performing a test with forced swimming to exhaustion. | [] | Potolitsyna, Natalya,Parshukova, Olga,Vakhnina, Nadezhda,Alisultanova, Nadezhda,Kalikova, Lubov,Tretyakova, Anastasia,Chernykh, Alexey,Shadrina, Vera,Duryagina, Arina,Bojko, Evgeny | eng |
PMC7036986 | International Journal of
Environmental Research
and Public Health
Article
Alterations in Running Biomechanics after 12 Week
Gait Retraining with Minimalist Shoes
Yang Yang 1, Xini Zhang 1, Zhen Luo 1, Xi Wang 1, Dongqiang Ye 1 and Weijie Fu 1,2,*
1
School of Kinesiology, Shanghai University of Sport, Shanghai 200438, China;
[email protected] (Y.Y.); [email protected] (X.Z.); [email protected] (Z.L.);
[email protected] (X.W.); [email protected] (D.Y.)
2
Key Laboratory of Exercise and Health Sciences of Ministry of Education, Shanghai University of Sport,
Shanghai 200438, China
*
Correspondence: [email protected] or [email protected]; Tel.: +86-21-65507368; Fax: +86-21-51253242
Received: 20 December 2019; Accepted: 25 January 2020; Published: 28 January 2020
Abstract: Purpose: The intervention of 12 week gait retraining with minimalist shoes was established
to examine its effect on impact forces, joint mechanics, and vertical stiffness during running. Methods:
Thirty male recreational runners were randomly assigned to the gait retraining + minimalist shoe
(n = 15, GR) and minimalist shoe (n = 15, MIN) groups. The ground reaction force and marker
trajectories were collected before and after intervention at a speed of 3.33 ± 5% m/s. Results: A total of
17 participants (9 in the GR group and 8 in the MIN group) completed the training. After training,
(1) the loading rate of both groups decreased significantly, and the loading rate of the GR group was
lower than that of the MIN group. (2) The foot strike angle of the GR group decreased significantly
after training, and the plantarflexion angle and hip joint angular extension velocity increased in both
groups. (3) The moment of ankle joint increased in the GR group, and the stiffness of lower limbs was
significantly improved in both groups. Conclusion: The 12 week gait retraining with minimalist shoes
converted rearfoot strikers into forefoot strikers with a rate of 78% (7/9). More importantly, such a
combined program, compared to the training with only minimalist shoes, can avoid the peak impact
force and decrease the loading rate more effectively, thus providing a potential means of reducing
risk of running injury caused by impact forces. Moreover, the increased vertical stiffness of lower
extremity after gait retraining may improve running economy and corresponding energy utilization.
However, these observations also suggest that the sole use of minimalist footwear may have limited
effects on reducing running-related impacts.
Keywords: gait retraining; running biomechanics; strike pattern; minimalist shoe
1. Introduction
As one of the most popular sports in the world, running is attracting increasing attention
nowadays [1]. However, a high injury rate (19–79%) in running has been reported [2]. The impact load
is two to three times of the body weight at touchdown, which is considered to be the main risk factor
for causing damage such as stress fracture/fracture, patellofemoral joint pain syndrome, and plantar
fasciitis [2–5]. Thus, how to reduce the impact and risk of running injury has always been a hot issue
in the biomechanics, sports medicine, rehabilitation, and related industries [6,7].
In the past 50 years, the injury rate of running has not changed much despite the development of
running shoes [8]. Studies show that the cushioning function of running shoes cannot be utilized in
actively landing [6,9,10]. Hence, researchers have considered different shoe designs and the postural
control in lower limbs whilst running. As a result, gait retraining and minimalist shoe training derived
Int. J. Environ. Res. Public Health 2020, 17, 818; doi:10.3390/ijerph17030818
www.mdpi.com/journal/ijerph
Int. J. Environ. Res. Public Health 2020, 17, 818
2 of 13
from barefoot running theory have been applied to rehabilitation, medical treatment, and sports
fields [11–14].
Minimalist footwear are shoes with a lighter mass, greater flexibility, and lower heel-to-toe
drop than conventional running shoes [15]. Runners who use this type of footwear likely adopt a
non-rearfoot strike pattern [16,17], which can reduce impact forces [11,16]. McCarthy et al. [13] showed
that after a 12 week simulated barefoot training, in which the participants were free to adopt their
own running pattern, 100% used non-rearfoot strike patterns. Latorre-Roman et al. [18] found that
a 12 week barefoot training program causes significant changes in the foot strike pattern, with a
tendency towards midfoot or forefoot strikes. However, not all runners who are used to wearing
conventional shoes can switch to non-rearfoot strike when wearing minimalist shoes [18]. Without
the cushioning of conventional shoes, the risk of high impact-related injuries likely increases [19,20].
Therefore, combining gait retraining and minimalist shoes may be more effective and secure than
adopting the two separately.
Gait retraining, an active training program with instruction or feedback, differs from the minimalist
shoe training, which is a passive adaptive process of special shoe conditions (e.g., minimalist shoes,
barefoot shoes). Promoting a forefoot strike pattern, which is similar to the barefoot movement in the
literature, is considered as a possible way forward [17,21]. In addition to promoting non-rearfoot strike
patterns, gait retraining encourages forefoot/midfoot strike for a high frequency, light stride and an
upright posture [21,22]. Gait retraining reduces loading rate and impact peak by increasing the stride
frequency and adopting a non-rearfoot strike pattern [16,21,23]. Warne et al. [12] showed that a 6 week
combination program of gait retraining with minimalist shoes causes more significant changes than
that of gait retraining with conventional shoes. In light of the above information, the combination
of gait retraining and minimalist shoes can reduce the loading rate and peak impact force by using
a non-rearfoot strike pattern [13,16,23]. The implication is that the shoe condition should match the
running posture. However, the long-term impact on the running posture of such a combination
program remains unclear. The combined training program with a long incremental load may be
effective and safe.
The purpose of this study was to establish a combined intervention mode of 12 week gait retraining
with minimalist shoes and examine its effect on factors related to the risks of running injury and
performance, i.e., impact forces, joint mechanics, and vertical stiffness. The hypothesis was that the
participants received 12 week gait retraining with minimalist shoes would have a lower loading rate
and a decreased foot-strike angle compared to that of those who only used minimalist shoes.
2. Methods
2.1. Participants
Thirty recreational male runners (age: 30.0 ± 6.4 years; height: 175.0 ± 5.2 cm; body mass: 71.9 ±
9.4 kg; weekly running volume: 27.4 ± 8.7 km) were recruited. Inclusion criteria are as follows: (1) they
ran at least 3 days per week with a minimum of 20 km/week for at least 3 months prior to the study
and (2) they were used to running with rearfoot strike in cushioned shoes and had no experience of
barefoot running or special sneakers (e.g., five-finger shoes, minimalist shoes, and racing spikes). Prior
to this experiment, participants completed a basic information questionnaire and signed an informed
consent form to ensure that they had no musculoskeletal injuries for the past 6 months. This study was
approved by the Institutional Review Board of the Shanghai University of Sport (no. 2017007).
2.2. Experimental Design
A parallel randomized control design was used in this study. Thirty participants were randomly
(random number sort) divided into gait retraining + minimalist shoe (GR) and minimalist shoe (MIN)
groups. The two groups underwent the same testing process but with different interventions (Figure 1).
Int. J. Environ. Res. Public Health 2020, 17, 818
3 of 13
Foot size was measured, and participants in both groups were provided with a pair of minimalist
footwear (type INOV-8 Bare-XF 210 V2: 3 mm outsole, no midsole, 0 mm heel-toe drop, 227 g weight).
Figure 1. Flow diagram of this study.
2.3. Testing Procedure
A 10-camera motion capture system (100 Hz, T40, Vicon Motion Inc., Oxford, United Kingdom)
was used to collect kinematic data including hip, knee, and ankle joints (Figure 2). Two 90 × 60 ×
10 cm Kistler 3D force platforms (9287B, Kistler Corporation, Winterthur, Switzerland) were used
to collect ground reaction force (GRF) data at a sampling rate of 1000 Hz. Before the over-ground
test, the participants performed a 5 min warm-up on a treadmill at optional running speed with the
minimalist shoes, followed by 1 min 3.33 m/s experimental speed adaptation. During the over-ground
test, three successful right foot contacts on the force plate were required. Its presence was not mentioned
to avoid targeting problems. The speed during the over ground test was monitored to ensure the
participants ran at 3.33 m/s using a Witty-Manual grating timing system (Witty wireless training timer,
Microgate Corp., Bolzano, Italy) with a 5% acceptable variance. Both the GR and the MIN groups were
tested before and after the intervention.
Int. J. Environ. Res. Public Health 2020, 17, 818
4 of 13
Figure 2. The marker-set and the experimental setup.
2.4. Intervention
GR group: The participants were required to run at a medium-intensity self-selected speed with
minimalist shoes and strike with forefoot. A pressure sensitive insole (Podoon) was applied to GR
runners to distinguish foot strike patterns. The sensors were located at the metatarsophalangeal joint
and heel. Sound feedback could be obtained from a mobile application if participants struck with the
heel. The gait retraining program lasted 12 weeks and was three times a week. The duration of the
training gradually increased from 5 min in the 1st week to 48 min in the 12th week (Table 1) [12,13].
Weekly group training was also provided to ensure the quality of retraining and to minimize the
dropout rates. After each training, the experimenter will remind the participants who do not meet the
requirements or mismatch with the data in the cloud.
Table 1. The 12 week gait retraining intervention.
Week
1
2
3
4
5
6
7
8
9
10
11
12
Duration (min)
5
10
15
20
25
30
35
40
42
44
46
48
Times per week
3
3
3
3
3
3
3
3
3
3
3
3
MIN group: During the running training, the participants were required to run at a medium-
intensity self-selected speed wearing minimalist shoes without any instructions for the strike pattern.
A pressure-sensitive insole (Podoon) was also applied to MIN runners for matching the same insole
condition of the GR group, but they did not receive the mobile application that provided sound
feedback. The schedule was the same as that of the GR group.
The intervention training was only an alternative part of the training [12,13]. The total running
distance per week was unchanged. Participants kept record training logs, including the time training
start/stop, location, and distance. During training, they were told that any discomfort or injury needed
to be reported to the experimenter. The researchers checked the training logs stored in the cloud.
Int. J. Environ. Res. Public Health 2020, 17, 818
5 of 13
The participants in both groups were allowed to wear habitual running shoes when out of training.
During training sessions, the two groups were prevented from interacting with one another.
Inclusion criteria: (1) completed all tests, (2) no more than three absences, and (3) completed the
last 3 weeks’ training with no more than six absences. Those who satisfied any condition were included.
During training, participants were allowed to delay or withdraw due to injury or personal reasons.
2.5. Data Processing
Kinematic data and GRF were analyzed via the gait analysis software Visual 3D (v5, C-Motion,
Inc., Germantown, MD, USA) using inverse dynamics. The GRF was filtered with a cut-off frequency
of 100 Hz. Marker trajectories were filtered with a cut-off frequency of 7 Hz [10] via a fourth-order
Butterworth low-pass filter. The hip, knee, and ankle angles of the lower limb were defined on the
basis of our previous model [24], and the kinematic features of each joint were calculated.
Impact variables included peak impact forces and maximum loading rates. The maximum loading
rate (LR) is equivalent to a slope of 20–80% of first peak (FP). If FP is non-existent, then LR is calculated
by using 13% of the gait cycle as a representative value [25,26].
Kinematic variables included (1) ground contact time (CT), which represents the duration between
touchdown to off-ground; (2) strike angle (θf) which refers to the angle between the foot and ground at
initial contact; (3) angles of the hip, knee, and ankle joints when contacting the ground (θ0) and the
maximum joint angle (θmax); and (4) joint angular velocities including the angular velocity at initial
contact (ω0) and the maximum angular velocity of hip, knee, and ankle joints (ωp). The angle of ankle
joint was 0◦ during standing, negative for extension/plantarflexion, and positive for flexion/dorsiflexion
(Figure 3).
Kinetic variables included (1) joint moment determined by the net moment generated by the
muscles of the hips, knees, and ankles of the lower limbs using the inverse dynamics in Visual 3D
biomechanical analysis software; (2) peak extension joint power (p), which is the product of the net
moment (M) and joint angular velocities (ω), and (3) vertical stiffness (k = GRFi/∆y) [27]. For the joint
moment, the maximum extension moment (Mmax) of each joint was selected. GRFi represents the
vertical GRF when the center of gravity (CoG) was lowest, and ∆y represents the vertical displacement
of CoG during centrifugation.
Figure 3. Angles of lower extremity joints.
Int. J. Environ. Res. Public Health 2020, 17, 818
6 of 13
2.6. Statistics
The mean and standard deviation for each variable were calculated. The results of each group
of pre/post were tested for normality. The original value was used in all tables and figures for
comparison. A two-way repeated measure ANOVA was used to examine the effects of retraining (pre-
and post-training) and groups (GR and MIN) on each variable (Version 22.0; SPSS, Inc., Chicago, IL,
USA). Independent t-tests and paired t-tests were used as post-hoc tests when a significant interaction
was detected. The significance level was set as α = 0.05.
3. Results
3.1. Dropout Rate
Seventeen participants completed intervention and met the inclusion criteria (nine in the GR
group, eight in the MIN group) (Table 2). Specifically, an FFS runner in the GR group was excluded
after pre-test. During intervention, two participants (one in GR, one in MIN) were excluded due to
injuries caused by non-training related events, i.e., walked downstairs carelessly. Two participants
(one in GR, one in MIN) were excluded due to mismatch of the cloud data, and they could not provide
reliable evidence, such as app or smart watch data. Three participants (one in GR, two in MIN) who
lost contact during the training were excluded. Five participants (two in GR, three in MIN) who
quit or missed too much training were also excluded. No significant difference was observed in the
average running volumes between the GR and MIN groups (GR: 28.3 ± 11.2 km/week, MIN: 26.9 ±
10.7 km/week).
Table 2. Information of the participants who completed training. GR: gait retraining + minimalist shoe,
MIN: minimalist shoe.
Age (years)
Height (cm)
Body Mass (kg)
km per Week (km)
GR (n = 9)
32.4 ± 6.1
174.8 ± 5.3
70.2 ± 6.0
28.3 ± 11.2
MIN (n = 8)
27.6 ± 5.2
173.9 ± 7.0
75.4 ± 11.7
26.9 ± 10.7
t-test
p = 0.104
p = 0.773
p = 0.262
p = 0.787
3.2. Impact Forces
A significantly main effect of time on the loading rate was observed (Figure 4; Table 3), which was
significantly reduced by 22.6% (GR) and 17.2% (MIN) after training (p < 0.001, p = 0.017). The loading
rate of the GR group was lower than that of the MIN group after training (p = 0.015). No interaction
effect was noted between time × group for any other GRF parameters in this study.
Figure 4. Comparison of loading rate between two groups before and after training.
Int. J. Environ. Res. Public Health 2020, 17, 818
7 of 13
Table 3. Contrast of ground reaction force (GRF) in different shoe conditions before and after training.
Parameter
GR
MIN
Pre
Post
Pre
Post
FP (BW)
1.78 ± 0.20
N/A
1.83 ± 0.22
2.05 ± 0.47
TFP (ms)
25.50 ± 5.31
N/A
26.20 ± 4.71
28.97 ± 15.75
LR (BW·s−1)
71.62 ± 13.66 #
55.44 ± 25.21 *
74.00 ± 21.42
61.30 ± 32.90
CT (ms)
233.58 ± 20.44
226.35 ± 11.90
243.27 ± 26.65
240.02 ± 26.26
FP: the first peak of the touch-down phase; TFP: the instant reaching the FP; BW: body weight; LR: loading rate;
CT: ground contact time; #: significant difference from pre- to post-tests. * significant difference between groups at
time point, p < 0.005.
3.3. Kinematics
A significant main effect of time was observed on the foot-strike angle, ankle angle (Figure 5),
and angular velocity of hip (p = 0.026, p = 0.011, p = 0.032, respectively) (Table 4). In addition,
a significant interaction effect between time × group on the foot-strike angle was observed (p = 0.013).
After the post-hoc test, the foot-strike angle of the GR group decreased by 10.3◦ after training (p = 0.015),
but no difference was noted in the MIN group (p = 0.753) (Figure 5). The foot-strike angle of the GR
group was significantly different from that of the MIN group in the post-test (p = 0.017). After training,
the ankle angle significantly decreased by 4.6◦ (GR) and 2.5◦ (MIN) at touchdown, and the maximum
angular velocity of hip joint increased by 15.2% (GR) and 25.2% (MIN). There was no interaction effect
between time × group for any other kinematic parameters.
Figure 5. Comparison of foot-strike angle (left) and ankle angle (right) between two groups before
and after training. * significant difference from pre- to post-tests in GR group; #: significant difference
between groups at time point, p < 0.005; &: significant difference from pre- to post-tests in MIN group.
Table 4. Kinematics changes of hip, knee, and ankle before and after training.
Joints
Parameter
GR
MIN
Pre
Post
Pre
Post
Foot/Ankle
θf (deg)
8.07 ± 4.64 *,&
−2.21 ± 3.09 #
8.73 ± 6.68
9.27 ± 8.9
θ0 (deg)
−0.13 ± 4.29 *
−4.73 ± 4.79 #
0.61 ± 3.76 *
−1.89 ± 5.27
θmax (deg)
17.48 ± 4.76
16.92 ± 4.78
15.91 ± 2.51
16.58 ± 3.38
ω0 (deg·s−1)
289.57 ± 85.81
332.77 ± 103.51
259.22 ± 34.13
267.83 ± 67.03
ωp (deg·s−1)
−269.34 ± 90.64
−245.00 ± 60.65
−231.46 ± 66.38
−213.37 ± 44.05
Knee
θ0 (deg)
−13.56 ± 5.60
−14.24 ± 5.11
−14.17 ± 4.01
−11.80 ± 3.76
θmax (deg)
−34.46 ± 2.08
−35.41 ± 4.75
−35.47 ± 2.93
−36.15 ± 4.62
ω0 (deg·s−1)
−96.83 ± 35.33
−85.96 ± 51.95
−85.80 ± 39.66
−83.61 ± 57.09
ωp (deg·s−1)
103.52 ± 34.96
109.21 ± 26.37
90.68 ± 40.83
74.21 ± 22.57
Int. J. Environ. Res. Public Health 2020, 17, 818
8 of 13
Table 4. Cont.
Hip
θ0 (deg)
25.50 ± 5.04
27.36 ± 8.23
29.36 ± 5.36
28.25 ± 7.28
θmax (deg)
−11.06 ± 6.59
−10.74 ± 7.30
−7.92 ± 6.03
−9.73 ± 6.71
ω0 (deg·s−1)
−82.08 ± 29.88
−60.34 ± 15.15
−64.37 ± 43.64
−62.22 ± 21.88
ωp (deg·s−1)
95.49 ± 39.13 *
109.99 ± 26.54
100.45 ± 26.82 *
125.73 ± 28.51
θf: the angle between foot and ground at initial contact; θ0: the angle at initial contact; θmax: maximum angle;
ω0: the peak angular velocity at initial contact; ωp: the peak angular velocity of extension; * significant difference
from pre- to post-tests; # significant difference between groups at time point, p < 0.005; &: interaction effect between
time × group, p < 0.005.
3.4. Kinetics
A significant main effect was observed on the peak ankle extension moment (p < 0.001), peak knee
extension moment (p = 0.004), and peak power of the hip (p < 0.001) (Figure 6). The peak moment of the
knee for the GR and MIN groups significantly decreased by 13.4% (GR) and 12.8% (MIN), respectively,
after training, and the peak power of the hip for both groups was significantly decreased by 38.6%
(GR) and 38.2% (MIN). In addition, a significant interaction effect was noted on the peak moment of
ankle (p = 0.024). Specifically, the peak moment was increased by 17.8% after training in the GR group
(p = 0.001), but no difference was observed in the MIN group.
Figure 6. Comparison of peak moment (left column) and peak power (right column) of hip, knee, and
ankle between two groups before and after training. * significant difference from pre- to post-tests.
Int. J. Environ. Res. Public Health 2020, 17, 818
9 of 13
3.5. Vertical Stiffness
A significant main effect of time was observed on vertical stiffness (p = 0.035). After training,
the vertical stiffness improved in the GR and MIN groups by 17.2% and 7.1%, respectively (Figure 7).
However, no significant main effect or interaction was observed on the vertical displacement of CoG
and the vertical GRF (Table 5).
Figure 7. Comparison of lower limbs stiffness between two groups before and after training. * significant
difference from pre- to post-tests.
Table 5. Vertical GRF when the center of gravity (CoG) was lowest, and the vertical displacement
of CoG. GRFi represents the vertical GRF when the CoG was lowest, and ∆y represents the vertical
displacement of CoG during centrifugation.
Parameter
GR
MIN
Pre
Post
Pre
Post
GRFi (BW)
2.61 ± 0.30
2.71 ± 0.31
2.55 ± 0.28
2.60 ± 0.27
∆y (cm)
5.96 ± 0.90
5.42 ± 0.99
6.21 ± 1.43
5.67 ± 1.47
4. Discussion
The purpose of this study was to examine the effect of gait retraining with minimalist shoes on
impact forces, joint mechanics, and vertical stiffness. Significant reduction was found in foot-strike
angle and LR. The kinematics and kinetics of the lower extremity joints changed after 12 weeks of gait
retraining with the minimalist shoes. Compared with the MIN group, more significant changes were
noted in the GR group, especially in the strike patterns and kinematics and kinetics characteristics of
the ankle joint. These results supported the hypothesis that the participants who received 12 week gait
retraining with minimalist shoes would have a lower loading rate and more lower foot strike angle
compared to that of those who only used minimalist shoes.
4.1. Impact Forces
Gait retraining can significantly reduce the impact force, which is considered to be the main cause
of lower-limb injuries, and this result supports the findings of previous studies [2–4]. Seven out of
nine participants of the GR group changed to forefoot strike after gait retraining. Previous studies
demonstrate that the impact force mainly depends on the effective mass of the lower limbs [11].
The forefoot strike can obviously reduce the effective mass by adjusting the angle between the foot and
the ground at initial contact, thus avoiding the high impact force caused by rearfoot strike.
The LR, the change in force per unit time, is considered to be a sensitive index to detect the
variations in the amount of impact forces during running. In this study, the LR for both groups decreased
after training. For the GR group, the change of the strike pattern avoided the peak of impact force,
Int. J. Environ. Res. Public Health 2020, 17, 818
10 of 13
thereby reducing the LR. This outcome is similar to the findings of other related studies [2–4,12,25,26].
By contrast, the MIN showed a significantly reduced LR without the change of strike pattern, which may
be related to the adaptability of the body after the change of shoe condition [11]. The LR of the GR
group was lower than that of the MIN group after training, indicating that the effect of gait retraining
may be more significant. Gait retraining as an intervention for actively changing the strike pattern may
be better matched with minimalist shoes to reduce the LR and avoid the peak of the impact force.
4.2. Kinematics
The participants in the GR group preferred to strike with the forefoot after training, as shown
by the foot-strike angle reduced by approximately 10.3◦. This preference of forefoot strike in the GR
group was the embodiment of the gait retraining [11]. After 12 weeks of training, the participants in
the GR group ran similarly to runners who run barefoot.
In the MIN group, no change was noted in strike pattern after training. In the study of McCarthy
et al. [13], all participants who preferred rearfoot strike became non-rearfoot strike after training
with minimalist shoes. Latorre-Roman et al. [18] revealed that a 12 week barefoot running program
causes significant changes in foot strike patterns with a tendency towards a non-rearfoot strike in
long-distance runners. In the study of Hollander et al. [28], habitually shod participants who actively
changed from shod to barefoot increased the foot strike index. However, the results in the current
study did not show this result, which might have been due to the difference in training program or the
gender, age, and other attributes of the participants.
Although only runners in the GR group showed a significant change in foot-strike pattern, the two
groups exhibited more plantarflexion after training. For the GR group, the larger plantarflexion angle
indicated a trend towards forefoot strike pattern [11,29]. In this study, the variation of the plantarflexion
angle of the GR group was not larger than the foot-strike angle (10.3◦ vs. 4.8◦). This outcome indicates
that the runners in GR may not achieve forefoot strike by simply increasing the plantarflexion angle.
No significant change was observed in knee and hip angles when touching the ground. Apart from
increasing the plantarflexion angle of ankle, the participants in the GR group might have achieved the
forefoot strike pattern by adjusting the position of the body and the forward of the trunk after training
instead of simply “running on the toe” [11].
4.3. Kinetics and Vertical Stiffness
The peak ankle moment increased significantly in the GR group after training, but no significant
difference was observed in the MIN group. This outcome may have been due to increasing arm of
force changed by the foot strike pattern. Warne et al. reported no significant change in the stiffness of
ankle after gait retraining [12]. Although the peak ankle moment increased, runners maintained the
stiffness by increasing the range of ankle upon touchdown. In the current study, this phenomenon may
have appeared as the increased plantarflexion angle. The peak knee extension moment for both groups
decreased significantly after training. When wearing minimalist shoes, the reduced stiffness of the
knee may result in reduced moments or increased knee excursion compared with wearing traditional
shoes [12], suggesting that the knee is more inclined to soft landing in the case of minimalist shoes.
In this study, the peak power of hip of both groups decreased significantly. In the study by
Williams et al. [30], runners wearing shoes showed a significant decrease in the power of hip when
using forefoot strike pattern or barefoot running compared with rearfoot strike pattern, and this effect
continued after training. Similarly, in the present study, the vertical stiffness, which is considered to
be related to running economy and energy utilization, increased significantly after training in both
groups [31]. Gait retraining or using minimalist shoes may improve running performance. The results
also showed that the vertical GRF and the displacement of CoG did not change significantly after
training. Hence, vertical stiffness seems better for evaluating the effectiveness of the training.
During the intervention, two participants were injured due to non-training factors and they were
excluded from the statistics of the proportion of injuries. In other related studies, the proportion of
Int. J. Environ. Res. Public Health 2020, 17, 818
11 of 13
injuries in McCarthy’s study was 20–26% [13], and the proportion was 17% for two runners who
were injured in the study of Warne et al. [12]. The number of injured in the present study was also
two. This outcome suggests that gait retraining is important when running barefoot or wearing
minimalist shoes.
4.4. Limitations
Firstly, the sample size was relatively small due to the long training period and participant
dropout. A large sample size could have increased the statistical power in such a way that additional
variables achieved significance, especially the variable trends in kinematics but without significance.
For future research on recreational runners, additional attention should be paid to the training
control of the participants due to work travel and other reasons, which are difficult but useful for the
sample preservation. In addition, this study focuses on the contrast between two different training
programs (GR and MIN). Thus, no barefoot and control groups are set up. However, a complete
four-group study (combined, control, minimalist, and barefoot groups) is necessary for future research.
Secondly, individual differences in movement learning ability might have led to different training
effects, especially in the GR group. Moreover, the long-term retention effects caused by retraining
changes are unknown. Finally, future investigations, including EMG assessment accompanied with
neuro-musculoskeletal adaptations after gait retraining, are warranted.
5. Conclusions
The 12 week gait retraining with minimalist shoes converted rearfoot strikers into forefoot
strikers. Seven of out nine participants transformed into forefoot strike patterns with a rate of 78%.
More importantly, such a combined program, compared to the training with only minimalist shoes,
can avoid the peak impact force and decrease the loading rate more effectively, thus providing a potential
means of reducing risk of running injury caused by impact forces. Moreover, the increased vertical
stiffness of lower extremity after gait retraining may improve running economy and corresponding
energy utilization. However, these observations also suggest that the sole use of minimalist footwear
may have limited effects on reducing running-related impacts.
Author Contributions: Conceptualization, W.F.; methodology, Y.Y.; formal analysis, Y.Y., X.Z., Z.L., X.W. and D.Y.;
investigation, Y.Y., X.Z., Z.L., X.W. and W.F.; resources, W.F.; data curation, Y.Y.; writing—original draft preparation,
Y.Y.; writing—review and editing, W.F.; project administration, W.F.; funding acquisition, W.F. All authors have
read and agreed to the published version of the manuscript.
Funding: This work was supported by the National Natural Science Foundation of China (11772201, 11572202);
Talent Development Fund of Shanghai Municipal (2018107); the National Key Technology Research and
Development Program of the Ministry of Science and Technology of China (2019YFF0302100); and the “Dawn”
Program of Shanghai Education Commission, China (19SG47).
Conflicts of Interest: The authors declare no conflict of interest.
References
1.
Bramble, D.M.; Lieberman, D.E. Endurance running and the evolution of Homo. Nature 2004, 432, 345–352.
[CrossRef] [PubMed]
2.
Van Gent, R.N.; Siem, D.; van Middelkoop, M.; van Os, A.G.; Bierma-Zeinstra, S.M.; Koes, B.W. Incidence
and determinants of lower extremity running injuries in long distance runners: A systematic review. Br. J.
Sports Med. 2007, 41, 469–480, discussion 480. [CrossRef] [PubMed]
3.
Milner, C.E.; Ferber, R.; Pollard, C.D.; Hamill, J.; Davis, I.S. Biomechanical factors associated with tibial stress
fracture in female runners. Med. Sci. Sports Exerc. 2006, 38, 323–328. [CrossRef]
4.
Pohl, M.B.; Hamill, J.; Davis, I.S. Biomechanical and anatomic factors associated with a history of plantar
fasciitis in female runners. Clin. J. Sport. Med. 2009, 19, 372–376. [CrossRef]
5.
Wang, H.; Kia, M.; Dickin, D.C. Influences of load carriage and physical activity history on tibia bone strain.
J. Sport Health Sci. 2019, 8, 478–485. [CrossRef]
Int. J. Environ. Res. Public Health 2020, 17, 818
12 of 13
6.
Fu, W.; Liu, Y.; Zhang, S. Effects of footwear on impact forces and soft tissue vibrations during drop jumps
and unanticipated drop landings. Int. J. Sports Med. 2013, 34, 477–483. [CrossRef]
7.
Fu, W.; Wang, X.; Liu, Y. Impact-induced soft-tissue vibrations associate with muscle activation in human
landing movements: An accelerometry and EMG evaluation. Technol. Health Care 2015, 23 (Suppl. 2),
S179–S187. [CrossRef]
8.
Nigg, B.M. The role of impact forces and foot pronation: A new paradigm. Clin. J. Sport. Med. 2001, 11, 2–9.
[CrossRef]
9.
Wang, X.; Zhang, S.; Fu, W. Changes in Impact Signals and Muscle Activity in Response to Different Shoe
and Landing Conditions. J. Hum. Kinet. 2017, 56, 5–18. [CrossRef] [PubMed]
10.
Fu, W.; Fang, Y.; Gu, Y.; Huang, L.; Liu, Y. Shoe Cushioning Reduces Impact and Muscle Activation during
Landings from Unexpected, but not Self-initiated, Drops. J. Sci. Med. Sport 2017, 20, 915–920. [CrossRef]
[PubMed]
11.
Lieberman, D.E.; Venkadesan, M.; Werbel, W.A.; Daoud, A.I.; D’Andrea, S.; Davis, I.S.; Mang’eni, R.O.;
Pitsiladis, Y. Foot strike patterns and collision forces in habitually barefoot versus shod runners. Nature 2010,
463, 531–535. [CrossRef] [PubMed]
12.
Warne, J.P.; Smyth, B.P.; Fagan, J.O.; Hone, M.E.; Richter, C.; Nevill, A.M.; Moran, K.A.; Warrington, G.D.
Kinetic changes during a six-week minimal footwear and gait-retraining intervention in runners. J. Sports
Sci. 2017, 35, 1538–1546. [CrossRef] [PubMed]
13.
McCarthy, C.; Fleming, N.; Donne, B.; Blanksby, B. 12 weeks of simulated barefoot running changes foot-strike
patterns in female runners. Int. J. Sports Med. 2014, 35, 443–450. [CrossRef] [PubMed]
14.
Daoud, A.I.; Geissler, G.J.; Wang, F.; Saretsky, J.; Daoud, Y.A.; Lieberman, D.E. Foot strike and injury rates in
endurance runners: A retrospective study. Med. Sci. Sports Exerc. 2012, 44, 1325–1334. [CrossRef]
15.
Lussiana, T.; Hébert-Losier, K.; Mourot, L. Effect of minimal shoes and slope on vertical and leg stiffness
during running. J. Sport Health Sci. 2015, 4, 93–100. [CrossRef]
16.
Altman, A.R.; Davis, I.S. A kinematic method for footstrike pattern detection in barefoot and shod runners.
Gait Posture 2012, 35, 298–300. [CrossRef]
17.
Giandolini, M.; Arnal, P.J.; Millet, G.Y.; Peyrot, N.; Samozino, P.; Dubois, B.; Morin, J.-B.t. Impact reduction
during running: Efficiency of simple acute interventions in recreational runners. Eur. J. Appl. Physiol. 2013,
113, 599–609. [CrossRef]
18.
Latorre-Roman, P.A.; Garcia-Pinillos, F.; Soto-Hermoso, V.M.; Munoz-Jimenez, M. Effects of 12 weeks of
barefoot running on foot strike patterns, inversion-eversion and foot rotation in long-distance runners.
J. Sport Health Sci. 2019, 8, 579–584. [CrossRef]
19.
Ryan, M.; Elashi, M.; Newsham-West, R.; Taunton, J. Examining injury risk and pain perception in runners
using minimalist footwear. Br. J. Sports Med. 2014, 48, 1257–1262. [CrossRef]
20.
Salzler, M.J.; Bluman, E.M.; Noonan, S.; Chiodo, C.P.; de Asla, R.J. Injuries Observed in Minimalist Runners.
Foot Ankle Int. 2012, 33, 262–266. [CrossRef]
21.
Goss, D.L.; Gross, M.T. A review of mechanics and injury trends among various running styles. US Army
Med. Dep. J. 2012, 62–71.
22.
Dallam, G.M.; Wilber, R.L.; Jadelis, K.; Fletcher, G.; Romanov, N. Effect of a global alteration of running
technique on kinematics and economy. J. Sports Sci. 2005, 23, 757–764. [CrossRef] [PubMed]
23.
Crowell, H.P.; Davis, I.S. Gait retraining to reduce lower extremity loading in runners. Clin. Biomech. 2011,
26, 78–83. [CrossRef] [PubMed]
24.
Xia, R.; Zhang, X.; Wang, X.; Sun, X.; Fu, W. Effects of Two Fatigue Protocols on Impact Forces and Lower
Extremity Kinematics during Drop Landings: Implications for Noncontact Anterior Cruciate Ligament
Injury. J. Healthc. Eng. 2017, 5690519. [CrossRef] [PubMed]
25.
Blackmore, T.; Willy, R.W.; Creaby, M.W. The high frequency component of the vertical ground reaction force
is a valid surrogate measure of the impact peak. J. Biomech. 2016, 49, 479–483. [CrossRef]
26.
Samaan, C.D.; Rainbow, M.J.; Davis, I.S. Reduction in ground reaction force variables with instructed barefoot
running. J. Sport Health Sci. 2014, 3, 143–151. [CrossRef]
27.
Liu, Y.; Peng, C.H.; Wei, S.H.; Chi, J.C.; Tsai, F.R.; Chen, J.Y. Active leg stiffness and energy stored in the
muscles during maximal counter movement jump in the aged. J. Electromyogr. Kinesiol. 2006, 16, 342–351.
[CrossRef]
Int. J. Environ. Res. Public Health 2020, 17, 818
13 of 13
28.
Hollander, K.; Liebl, D.; Meining, S.; Mattes, K.; Willwacher, S.; Zech, A. Adaptation of Running Biomechanics
to Repeated Barefoot Running: A Randomized Controlled Study. Am. J. Sports Med. 2019, 47, 1975–1983.
[CrossRef]
29.
Bonacci, J.; Saunders, P.U.; Hicks, A.; Rantalainen, T.; Vicenzino, B.G.; Spratford, W. Running in a minimalist
and lightweight shoe is not the same as running barefoot: A biomechanical study. Br. J. Sports Med. 2013, 47,
387–392. [CrossRef]
30.
Williams, D.S.B.; Green, D.H.; Wurzinger, B. Changes in lower extremity movement and power absorption
during forefoot striking and barefoot running. Int. J. Sports Phys. Ther. 2012, 7, 525–532.
31.
Perl, D.P.; Daoud, A.I.; Lieberman, D.E. Effects of footwear and strike type on running economy. Med. Sci.
Sports Exerc. 2012, 44, 1335–1343. [CrossRef] [PubMed]
© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access
article distributed under the terms and conditions of the Creative Commons Attribution
(CC BY) license (http://creativecommons.org/licenses/by/4.0/).
| Alterations in Running Biomechanics after 12 Week Gait Retraining with Minimalist Shoes. | 01-28-2020 | Yang, Yang,Zhang, Xini,Luo, Zhen,Wang, Xi,Ye, Dongqiang,Fu, Weijie | eng |
PMC10651037 | PLOS ONE
Dose response of running on blood biomarkers of wellness in generally healthy
individuals
--Manuscript Draft--
Manuscript Number:
PONE-D-23-25168R1
Article Type:
Research Article
Full Title:
Dose response of running on blood biomarkers of wellness in generally healthy
individuals
Short Title:
Biomarker signature of runners
Corresponding Author:
Bartosz Nogal
InsideTracker
Cambridge, MA UNITED STATES
Keywords:
physical activity, exercise, blood biomarkers, running, generally healthy, mendelian
randomization
Abstract:
Exercise is effective toward delaying or preventing chronic disease, with a large body
of evidence supporting its effectiveness. However, less is known about the specific
healthspan-promoting effects of exercise on blood biomarkers in the disease-free
population. In this work, we examine 23,237 generally healthy individuals who self-
report varying weekly running volumes and compare them to 4,428 generally healthy
sedentary individuals, as well as 82 professional endurance athletes. We estimate the
significance of differences among blood biomarkers for groups of increasing running
levels using analysis of variance (ANOVA), adjusting for age, gender, and BMI. We
attempt and add insight to our observational dataset analysis via two-sample
Mendelian randomization (2S-MR) using large independent datasets. We find that
self-reported running volume associates with biomarker signatures of improved
wellness, with some serum markers apparently being principally modified by BMI,
whereas others show a dose-effect with respect to running volume. We further detect
hints of sexually dimorphic serum responses in oxygen transport and hormonal traits,
and we also observe a tendency toward pronounced modifications in magnesium
status in professional endurance athletes. Thus, our results further characterize blood
biomarkers of exercise and metabolic health, particularly regarding dose-effect
relationships, and better inform personalized advice for training and performance.
Order of Authors:
Bartosz Nogal
Svetlana Vinogradova
Gil Blander
Milena Jorge
Paul Fabian
Ali Torkamani
Response to Reviewers:
Editor comments:
1. When submitting your revision, we need you to address these additional
requirements. Please ensure that your manuscript meets PLOS ONE's style
requirements, including those for file naming.
Thank you. We believed we now correctly formatted the manuscript to reflect PLOS
formatting requirements for publishing.
2.We note that the grant information you provided in the ‘Funding Information’ and
‘Financial Disclosure’ sections do not match.
Thank you - we believe that we now addressed this. We removed all funding
information from the manuscript text, amended it to reflect the funding institutions
involvement, and moved it to the cover letter, as requested. Please let us know if
Powered by Editorial Manager® and ProduXion Manager® from Aries Systems Corporation
anything needs further attention.
3. In your Data Availability statement, you have not specified where the minimal data
set underlying the results described in your manuscript can be found.
We updated our Data Availability statement via the submission system to reflect our
opennes to share the minimal dataset upon request from the corresponding author, as
well as the url to the public repository where the gwas summary statistics can be found.
4. We note that you have included the phrase “data not shown” in your manuscript....
Thank you. We now addressed this within the text of the manuscript and the “data not
shown” no longer appears
5. Please include captions for your Supporting Information files at the end of your
manuscript, and update any in-text citations to match accordingly.
Captions for Supporting Information is now included at the end of the manuscript.
6. When submitting your revision, we need you to address these additional
requirements. Please ensure that your manuscript meets PLOS ONE's style
requirements, including those for file naming.
We believe we have now addressed the style and formatting requirements.
Reviewer comments:
Reviewer #1: The majority parts of the articles are technically sound. Moreover, the
purpose of the study is very sound since it focused on healthy active population.
Among the few drawbacks of the study the way the study subjects categorized into
groups based on the duration of the activity (>10hours per week and less that 10hours
per week is not appropriate. Moreover, the reliability/validity of the information sources
in relation to the biomarker tests and lifestyle habits of the study subjects didn't
consider the immediate effects of medical services and medication conditions of the
respondents at the time of reporting the volume of exercise and biomarker test results.
Medical services and lifestyle habits specially all the habits in addition to
exercises/running are very important to reach informative decision in this research. So,
the above two points need further explanation or modification.
Response to Reviewer #1: We appreciate the reviewer's feedback and are pleased
that they find the majority of our study technically sound and recognize the importance
of our focus on a healthy, active population. We also appreciate the reviewer pointing
out an opportunity to improve the clarity around our experimental design as it pertains
subject groupings.
Regarding the categorization of study subjects, we want to clarify that we actually
categorized them into five groups. These groups include professional endurance
runners, high volume amateur runners (>10 hours per week), medium volume amateur
runners (3-10 hours per week), low volume amateur runners (<3 hours per week), and
the sedentary. We now added a sentence starting on line 125 the explicitly states this
categorization (“The cohort was divided into five groups:…”). These groupings were
determined based on the respondents' self-reported data.
We acknowledge the potential influence of medication use on our analysis, and we
now address it starting on line 461 (“These factors, such as diet, sleep, and/or
medications were not readily ascertained in this free-living cohort…”) and in the Study
Limitations section (line 595). We noted that unmeasured confounders such as
medications, nutritional supplements, and unreported health conditions may exist.
However, given the nature of our cohort, which primarily consists of self-selected,
generally healthy individuals, the impact of significant medication use is expected to be
limited. We believe that the observed trends in healthier biomarker levels with
increased reported running volume support this assertion.
Powered by Editorial Manager® and ProduXion Manager® from Aries Systems Corporation
Furthermore, we recognize the importance of lifestyle habits beyond exercise in
influencing our results. To address this, we employed statistical genomics, specifically
two-sample Mendelian randomization with physical activity as the exposure. This
analysis allowed us to explore other potential habits and behaviors contributing to
improved biomarker signatures in physically active runners within our cohort. We kindly
refer the reviewer to the "Vigorous physical activity associates with healthier behaviors"
section in the results for a detailed examination of this aspect. Notably, our entire
cohort is composed of health-conscious individuals within the same health advisory
platform, with the primary differentiator being self-reported running activity. We also
controlled for key variables such as age, sex, and BMI in our ANOVA analyses.
We hope these explanations clarify our approach and address the reviewer's concerns
adequately.
Reviewer #2: How your data is reliable by using A cross-sectional study design? &
How again the Data is reliable by using self-reported running. I understand that
Biomarkers are objective measure, but do you think that Self-report is trustworthy?
Thank you
Response to Reviewer #2: We appreciate the reviewer's questions and concerns
regarding the reliability of our runners data, which is largely derived from self-reported
exercise habits. Cross-sectional studies inherently have limitations when it comes to
establishing causality, and we acknowledge this challenge. To address potential
confounding factors, we conducted additional causal analyses, specifically
investigating the effects of BMI on the biomarkers under examination to begin to
disentangle the relative contributions of known factors. Furthermore, we performed
secondary Mendelian randomization (MR) analyses to identify and account for
potential confounders in our findings. We kindly invite the reviewer to explore the
"Vigorous physical activity associates with healthier behaviors" section in the results for
a comprehensive exploration of these confounding aspects.
Regarding the reliability of self-reported running activity, we recognize that self-reports
can be subject to biases, and individuals may tend to overestimate their exercise
commitment. To address this drawback, we added language addressing these
limitations in the “Study limitations” section (Line 579: “First, it is generally known that
subjects tend to overestimate their commitment to exercise …”). We do note that our
study cohort comprises self-selected individuals who are health-conscious and
possibly less prone to over-report their running volume. Additionally, the robust
increasing trend in baseline levels of muscle damage biomarkers (CK, AST), which are
known to be associated with participation in sports and exercise, provides indirect
evidence that the different running groups in our study were indeed engaging in
increasing volumes of strenuous physical activity.
While self-reporting has its limitations, it remains a valuable method for capturing
individuals' exercise behaviors in large-scale observational studies. We took measures
to mitigate potential biases, and our findings align with established trends in biomarker
responses to physical activity.
Reviewer #3: Upon a meticulous review of the article in question, I wish to commend
the authors for crafting a piece that not only carries immense scientific weight but is
also articulated with great clarity. Such insightful work surely merits publication in your
distinguished journal. It's admirable how the authors have navigated through a myriad
of physiological and biochemical variables (blood biomarkers) across five distinct
participant categories and presented their results with lucidity. The experimental
framework is robust, the statistical evaluations are apt, and the narrative progresses
seamlessly. The references provided are both relevant and adequate. Nevertheless, I'd
like to offer a few observations and suggestions:
Response: We appreciate the reviewer's positive feedback and kind words about our
manuscript. We eagerly await their observations and suggestions should they see
further opportunities to improve our work based on our responses to the current
suggestions.
Powered by Editorial Manager® and ProduXion Manager® from Aries Systems Corporation
Original Title: “Dose response of running on blood biomarkers of wellness in the
generally healthy.”
Proposed Title: “Dose-response relationship between running and blood biomarkers of
wellness in generally healthy individuals.”
Response: Thank you – title has been changed.
Page 2, Line 8: The mention of “exposure to sunlight” seems somewhat out of context.
Could the authors clarify its relevance or indicate if it has been discussed elsewhere in
the article?
Response: Thank you for the suggestion, we removed this as we agree it was not
relevant in this manuscript.
Page 17, Lines 17-18: The text reads: "These observations suggest that elite
endurance runners………to their magnesium status."
Comments: It would be helpful to clarify whether the professional athletes (PRO)
participating in this study are specifically elite endurance runners. Kindly integrate this
distinction into the main text if accurate.
Response: Thank you for the clarifying suggestion. We included the pro/elite
endurance runners clarification within the abstract as well as a section heading (lines 7
and 425)
Page 19, Lines 1-2: The assertion: “Indeed whether exercise………..is inconclusive,”
needs to be substantiated with a relevant citation.
Response: Thank you – citations have been added.
Table 1: Please include standard deviation (SD) values. I also recommend expressing
exercise duration in terms of "h/week" instead of "hr".
Response: Thank you for the catch – units changed to “h/week” and SDs added to
Table 1.
We are grateful for your valuable feedback, which has contributed to improving the
clarity and accuracy of our manuscript.
Additional Information:
Question
Response
Financial Disclosure
Enter a financial disclosure statement that
describes the sources of funding for the
work included in this submission. Review
the submission guidelines for detailed
requirements. View published research
articles from PLOS ONE for specific
examples.
This statement is required for submission
and will appear in the published article if
the submission is accepted. Please make
sure it is accurate.
InsideTracker was the sole funding source. The funder provided support in the form of
salaries for authors B.N., S.V., P.F., M.J., and G.B., and was involved in the decision to
publish, but did not have an impact on the experimental design, data analysis, and
conclusions.
Powered by Editorial Manager® and ProduXion Manager® from Aries Systems Corporation
Unfunded studies
Enter: The author(s) received no specific
funding for this work.
Funded studies
Enter a statement with the following details:
Initials of the authors who received each
award
•
Grant numbers awarded to each author
•
The full name of each funder
•
URL of each funder website
•
Did the sponsors or funders play any role in
the study design, data collection and
analysis, decision to publish, or preparation
of the manuscript?
•
NO - Include this sentence at the end of
your statement: The funders had no role in
study design, data collection and analysis,
decision to publish, or preparation of the
manuscript.
•
YES - Specify the role(s) played.
•
* typeset
Competing Interests
Use the instructions below to enter a
competing interest statement for this
submission. On behalf of all authors,
disclose any competing interests that
could be perceived to bias this
work—acknowledging all financial support
and any other relevant financial or non-
financial competing interests.
This statement is required for submission
and will appear in the published article if
the submission is accepted. Please make
sure it is accurate and that any funding
sources listed in your Funding Information
later in the submission form are also
declared in your Financial Disclosure
statement.
View published research articles from
PLOS ONE for specific examples.
B.N., S.V., P.F., and G.B. are employees of InsideTracker. This does not alter our
adherence to PLOS ONE policies
on sharing data and materials.
Powered by Editorial Manager® and ProduXion Manager® from Aries Systems Corporation
NO authors have competing interests
Enter: The authors have declared that no
competing interests exist.
Authors with competing interests
Enter competing interest details beginning
with this statement:
I have read the journal's policy and the
authors of this manuscript have the following
competing interests: [insert competing
interests here]
* typeset
Ethics Statement
Enter an ethics statement for this
submission. This statement is required if
the study involved:
Human participants
•
Human specimens or tissue
•
Vertebrate animals or cephalopods
•
Vertebrate embryos or tissues
•
Field research
•
Write "N/A" if the submission does not
require an ethics statement.
General guidance is provided below.
Consult the submission guidelines for
detailed instructions. Make sure that all
information entered here is included in the
Methods section of the manuscript.
BRANY IRB File # 22-12-501-1095
BRANY IRB has determined this research is exempt from IRB review under
category(ies) # (4)(ii), as detailed in 45 CFR 46.104(d) and BRANY’s
Standard Operating Procedures (category excerpted below).
(4) Secondary research for which consent is not required: Secondary research uses of
identifiable private information or identifiable biospecimens, with the following criterion
met:
(ii) Information, which may include information about biospecimens, is recorded by the
investigator in such a manner that the identity of the human subjects cannot readily be
ascertained directly or through identifiers linked to the subjects, the investigator does
not
contact the subjects, and the investigator will not re-identify subjects
Powered by Editorial Manager® and ProduXion Manager® from Aries Systems Corporation
Format for specific study types
Human Subject Research (involving human
participants and/or tissue)
Give the name of the institutional review
board or ethics committee that approved the
study
•
Include the approval number and/or a
statement indicating approval of this
research
•
Indicate the form of consent obtained
(written/oral) or the reason that consent was
not obtained (e.g. the data were analyzed
anonymously)
•
Animal Research (involving vertebrate
animals, embryos or tissues)
Provide the name of the Institutional Animal
Care and Use Committee (IACUC) or other
relevant ethics board that reviewed the
study protocol, and indicate whether they
approved this research or granted a formal
waiver of ethical approval
•
Include an approval number if one was
obtained
•
If the study involved non-human primates,
add additional details about animal welfare
and steps taken to ameliorate suffering
•
If anesthesia, euthanasia, or any kind of
animal sacrifice is part of the study, include
briefly which substances and/or methods
were applied
•
Field Research
Include the following details if this study
involves the collection of plant, animal, or
other materials from a natural setting:
Field permit number
•
Name of the institution or relevant body that
granted permission
•
Data Availability
Authors are required to make all data
underlying the findings described fully
available, without restriction, and from the
time of publication. PLOS allows rare
exceptions to address legal and ethical
concerns. See the PLOS Data Policy and
FAQ for detailed information.
Yes - all data are fully available without restriction
Powered by Editorial Manager® and ProduXion Manager® from Aries Systems Corporation
A Data Availability Statement describing
where the data can be found is required at
submission. Your answers to this question
constitute the Data Availability Statement
and will be published in the article, if
accepted.
Important: Stating ‘data available on request
from the author’ is not sufficient. If your data
are only available upon request, select ‘No’ for
the first question and explain your exceptional
situation in the text box.
Do the authors confirm that all data
underlying the findings described in their
manuscript are fully available without
restriction?
Describe where the data may be found in
full sentences. If you are copying our
sample text, replace any instances of XXX
with the appropriate details.
If the data are held or will be held in a
public repository, include URLs,
accession numbers or DOIs. If this
information will only be available after
acceptance, indicate this by ticking the
box below. For example: All XXX files
are available from the XXX database
(accession number(s) XXX, XXX.).
•
If the data are all contained within the
manuscript and/or Supporting
Information files, enter the following:
All relevant data are within the
manuscript and its Supporting
Information files.
•
If neither of these applies but you are
able to provide details of access
elsewhere, with or without limitations,
please do so. For example:
Data cannot be shared publicly because
of [XXX]. Data are available from the
XXX Institutional Data Access / Ethics
Committee (contact via XXX) for
researchers who meet the criteria for
access to confidential data.
The data underlying the results
presented in the study are available
from (include the name of the third party
•
All Mendelian Randomization data required replicate the causal analysis can be freely
accessed at the following url: https://gwas.mrcieu.ac.uk/. No special access is required
and all the datasets used can be further freely accessed via the free "TwoSampleMR"
R package as described in the methods (biomarker dataset codes are shared in
Supplementary tables as well). The minimal dataset required to replicate the blood
biomarker results has been uploaded as Supporting Information.
Powered by Editorial Manager® and ProduXion Manager® from Aries Systems Corporation
and contact information or URL).
This text is appropriate if the data are
owned by a third party and authors do
not have permission to share the data.
•
* typeset
Additional data availability information:
Powered by Editorial Manager® and ProduXion Manager® from Aries Systems Corporation
1
Dose response of running on blood biomarkers of wellness in
generally healthy individuals
Bartek Nogal1¶ , Svetlana Vinogradova1¶, Milena Jorge1, Ali Torkamani2,3, Paul Fabian1,
Gil Blander1*
1InsideTracker, Cambridge, Massachusetts, United States of America
2The Scripps Translational Science Institute, The Scripps Research Institute, La Jolla, CA,
United States of America
3Department of Integrative Structural and Computational Biology, The Scripps Research
Institute, La Jolla, CA, United States of America
* Corresponding author
E-mail: [email protected] (GB)
¶These authors contributed equally to this work
Manuscript
Click here to access/download;Manuscript;Manuscript.docx
1
Abstract
1
Exercise is effective toward delaying or preventing chronic disease, with a large body of evidence
2
supporting its effectiveness. However, less is known about the specific healthspan-promoting
3
effects of exercise on blood biomarkers in the disease-free population. In this work, we examine
4
23,237 generally healthy individuals who self-report varying weekly running volumes and
5
compare them to 4,428 generally healthy sedentary individuals, as well as 82 professional
6
endurance runners. We estimate the significance of differences among blood biomarkers for
7
groups of increasing running levels using analysis of variance (ANOVA), adjusting for age,
8
gender, and BMI. We attempt and add insight to our observational dataset analysis via two-sample
9
Mendelian randomization (2S-MR) using large independent datasets. We find that self-reported
10
running volume associates with biomarker signatures of improved wellness, with some serum
11
markers apparently being principally modified by BMI, whereas others show a dose-effect with
12
respect to running volume. We further detect hints of sexually dimorphic serum responses in
13
oxygen transport and hormonal traits, and we also observe a tendency toward pronounced
14
modifications in magnesium status in professional endurance athletes. Thus, our results further
15
characterize blood biomarkers of exercise and metabolic health, particularly regarding dose-effect
16
relationships, and better inform personalized advice for training and performance.
17
Introduction
18
Physical inactivity is one of the leading modifiable behavioral causes of death in the US [1].
19
Worldwide, physical inactivity is estimated to account for about 8.3% of premature mortality, an
20
effect size that is on the same order as smoking and obesity [2]. At the same time, the potent health
21
benefits of exercise have been proven time and time again, with results so consistent across a wide
22
variety of chronic diseases that some posit it can be considered a medical intervention [3-5].
23
However, since most investigators report the effects of exercise in either diseased populations or
24
athletes [6, 7], there exists a significant gap in knowledge as to the measurable effects of exercise
25
in the generally healthy population who exercise for the purpose of improving their healthspan,
26
which can be projected via established measures such as blood biomarkers [8-11].
27
It is well established that routine laboratory biomarkers are validated proxies of the state of an
28
individual’s overall metabolic health and other healthpan-related parameters [12]. A large body
29
of evidence supports the effectiveness of exercise in modifying blood biomarkers toward disease
30
mitigation in clinical cohorts as well as athletes, where the effect sizes may be larger [6, 13].
31
Indeed, it’s been shown that more favorable changes in response to exercise training occur usually
32
in those with more pronounced dyslipidemia [13]. In professional athletes, the sheer volume
33
and/or intensity of physical activity may drive large effects in various hematological, lipid,
34
immune, and endocrine variables [6]. Our aim is to help fill the gap in understanding of the effects
35
of exercise on blood biomarkers in the generally healthy, free-living population. Toward this end,
36
we endeavored to explore the effects of vigorous exercise such as running in apparently healthy,
37
mostly non-athletic cohort to better understand the landscape of blood biomarker modifications
38
expected in the individual who partakes in recreational physical activity for the purpose of
39
maintaining good health.
40
For this purpose, we leveraged the InsideTracker dataset that includes information on self-reported
41
exercise habits combined with blood biomarker and genomics data. We have previously reported
42
on the results of a longitudinal analysis on blood biomarker data from 1032 generally healthy
43
individuals who used our automated, web-based personalized nutrition and lifestyle platform [14].
44
3
For the purpose of this investigation, we focused on running as the exercise of choice as it is one
45
of the most common (purposeful) physical activity modalities practiced globally by generally
46
healthy individuals and would thus be relevant. Moreover, since this was a cross-sectional study
47
based on self-reported exercise habits, we attempted to increase our capacity to infer intervention
48
effects, as well as tease out potential confounders, by performing 2S-MR in large independent
49
cohorts.
50
Materials and methods
51
Dataset
52
We conducted an observational analysis of data from InsideTracker users. InsideTracker is a
53
direct-to-consumer (DTC) company established in 2009 that markets and sells InsideTracker
54
(insidetracker.com), a personalized lifestyle recommendation platform. The platform provides
55
serum biomarker and genomics testing, and performs integrative analysis of these datasets,
56
combined with activity/sleep tracker data toward biomarker and healthspan optimization (of note,
57
at the time of this analysis, we did not have sufficient users with activity/sleep tracker data to
58
include this data stream in the current study). New users were continuously added to the
59
InsideTracker database from January 2011 to March 2022.
60
61
Recruitment of participants
62
Recruitment of participants aged between 18 and 65 and residing in North America was conducted
63
through company marketing and outreach. Participants were subscribing members to the
64
InsideTracker platform and provided informed consent to have their blood test data and self-
65
reported information used in an anonymized fashion for research purposes. Research was
66
conducted according to guidelines for observational research in tissue samples from human
67
subjects. Eligible participants completed a questionnaire that included age, ethnicity, sex, dietary
68
preferences, physical activity, and other variables. This study employed data from 23,237
69
participants that met our analysis inclusion requirements, namely absence of any chronic disease
70
as determined by questionnaire and metabolic blood biomarkers within normal clinical reference
71
ranges. The platform is not a medical service and does not diagnose or treat medical conditions,
72
so medical history and medication use were not collected. The Institutional Review Board (IRB)
73
determine this work was not subject to a review based on category 4 exemption (“secondary
74
research” with de-identified subjects).
75
Biomarker collection and analysis
76
Blood samples were collected and analyzed by Clinical Laboratory Improvement Amendments
77
(CLIA)–approved, third-party clinical labs (primarily Quest Diagnostics and LabCorp).
78
Participants were instructed to fast for 12 hours prior to the phlebotomy, with the exception of
79
water consumption. Results from the blood analysis were then uploaded to the platform via
80
electronic integration with the CLIA-approved lab. Participants chose a specific blood panel from
81
7 possible offerings, each comprising some subset of the biomarkers available. Due to the variation
82
in blood panels offered, the participant sample size per biomarker is not uniform.
83
Biomarker dataset preparation
84
5
In our raw dataset, occasional outlier values were observed that were deemed implausible (e.g.
85
fasting glucose < 65 mg/dL). To remove anomalous outliers in a systematic way, we used the
86
Interquartile Range (IQR) method of identifying outliers, removing data points which fell below
87
Q1 – 1.5 IQR or above Q3 + 1.5 IQR. The cohort was divided into five groups: professional
88
endurance runners (PRO), high volume amateur (>10 h/week, HVAM ), medium volume amateur
89
(3-10 h/week, MVAM), low volume amateur (<3 h/week, LVAM), and sedentary (SED).
90
Calculation of polygenic scores
91
The variants (SNPs) comprising the polygenic risk scores were derived from publicly available
92
GWAS summary statistics (https://www.ebi.ac.uk/gwas/). Scores were calculated across users by
93
summing the product of effect allele doses weighted by the beta coefficient for each SNP, as
94
reported in the GWAS summary statistics. Variant p-value thresholds were generally chosen based
95
on optimization of respective PGS-blood biomarker correlation in the entire InsideTracker cohort
96
with both blood and genomics datasets (~1000-1500 depending on the blood biomarker at the time
97
of analysis). Genotyping data was derived from a combination of a custom InsideTracker array
98
and third party arrays such as 23andMe and Ancestry. Not all variants for any particular PGS were
99
genotyped on every array; proxies for missing SNPs were extracted via the “LDlinkR” package
100
using the Utah Residents (CEPH) with Northern and Western European ancestry (CEU) population
101
(R2 > 0.8 cut-off). Only results PGSs for which there was sufficient biomarker-genotyping dataset
102
overlap were reported (note that none of the blood biomarker PGSs met this requirement).
103
Blood biomarker analysis with respect to running volume and
104
polygenic scores
105
To estimate significance of differences for blood biomarkers levels among exercise groups, we
106
performed 3-way analysis of variance (ANOVA) analysis adjusting for age, gender, and BMI
107
(type-II analysis-of-variance tables function ANOVA from ‘car’ R package, version 3.0-12).
108
When estimating the effort of reported training volume on biomarkers, we assigned numerical
109
values corresponding to 4 levels of running and performed ANOVA analysis with those levels
110
treating it as an independent variable. P-values were adjusted using the Benjamini & Hochberg
111
method [15]. P-values for interaction plots were calculated with ANOVA including interaction
112
between exercise group and polygenic scores category. When comparing runners (PRO and
113
HVAM combined) versus sedentary individuals, we used propensity score matching method to
114
account for existing covariates (age and gender): we identified 745 sedentary individuals with
115
similar to runners’ age distributions among both males and females. We used ‘MatchIt’ R package
116
(version 4.3.3) implementing nearest neighbor method for matching [16].
117
Mendelian randomization
118
We attempted to add insight around the causality of exercise vs. BMI differences with respect to
119
serum marker improvement by performing MR analyses on a subset of biomarker observations
120
where BMI featured as a strong covariate and was thus used as the IV in the 2S-MR. Thus, our
121
hypothesis here was that BMI differences were the primary (causal) driver behind the improvement
122
behind some biomarkers. MR uses genetic variants as modifiable exposure (risk factor) proxies
123
to evaluate causal relationships in observational data while reducing the effects of confounders
124
and reverse causation (S1 Fig). These SNPs are used as instrumental variables and must meet 3
125
basic assumptions: (1) they must be robustly associated with the exposure; (2) they must exert
126
their effect on outcome via the exposure, and (3) there must be no unmeasured confounders of the
127
7
associations between the genetic variants and outcome (e.g. horizontal pleiotropy) [17].
128
Importantly, SNPs are proper randomization instruments because they are determined at birth and
129
thus serve as proxies of long-term exposures and cannot, in general, be modified by the
130
environment. If the 3 above mentioned assumptions hold, MR-estimate effects of exposure on
131
outcomes are not likely to be significantly affected by reverse causation or confounding. In the
132
2S-MR performed here, where GWAS summary statistics are used for both exposure and outcome
133
from independent cohorts, reverse causation and horizontal pleiotropy can readily be assessed, and
134
weak instrument bias and the likelihood of false positive findings are minimized as a result of the
135
much larger samples sizes [17]. Indeed, the bias in the 2S-MR using non-overlapping datasets as
136
performed here is towards the null [17]. Furthermore, to maintain the SNP-exposure associations
137
and linkage disequilibrium (LD) patterns in the non-overlapping populations we used GWAS
138
datasets from the MR-Base platform that were derived from ancestrally similar populations
139
(“ukb”: analysis of UK Biobank phenotypes, and “ieu”: GWAS summary datasets generated by
140
many different European consortia). To perform the analysis we used the R package
141
“TwoSampleMR” that combines the effects sizes of instruments on exposures with those on
142
outcomes via a meta-analysis. We used “TwoSampleMR” package functions for allele
143
harmonization between exposure and outcome datasets, proxy variant substitution when SNPs
144
from exposure were not genotyped in the outcome data (Rsq>0.8 using the 1000G EUR reference
145
data integrated into MR-Base), and clumping to prune instrument SNPs for LD (the R script used
146
for MR analyses is available upon request). We used 5 different MR methods that were included
147
as part of the “TwoSampleMR” package to control for bias inherent to any one technique [18].
148
For example, the multiplicative random effects inverse variance-weighted (IVW) method is a
149
weighted regression of instrument-outcome effects on instrument-exposure effects with the
150
intercept is set to zero. This method generates a causal estimate of the exposure trait on outcome
151
traits by regressing the, for example, SNP-BMI trait association on the SNP-biomarker measure
152
association, weighted by the inverse of the SNP-biomarker measure association, and constraining
153
the intercept of this regression to zero. This constraint can result in unbalanced horizontal
154
pleiotropy whereby the instruments influence the outcome through causal pathways distinct from
155
that through the exposure (thus violating the second above-mentioned assumption). Such
156
unbalanced horizontal pleiotropy distorts the association between the exposure and the outcome,
157
and the effect estimate from the IVW method can be exaggerated or attenuated. However,
158
unbalanced horizontal pleiotropy can be readily assessed by the MR Egger method (via the MR
159
Egger intercept), which provides a valid MR causal estimate that is adjusted for the presence of
160
such directional pleiotropy, albeit at the cost of statistical efficiency. Finally, to ascertain the
161
directionality of the various causal relationships examined, we also performed each MR analysis
162
in reverse where possible.
163
Results
164
Study population characteristics
165
Table 1 shows the demographic characteristics of the study population. We observed a
166
significant trend toward younger individuals reporting higher running volume, with more than
167
75% of the professional (PRO) group falling between the ages of 18 and 35 (S1 Table). Significant
168
differences were also observed in the distribution of males and females within study groups (Table
169
1). Moreover, higher running volume associated with significantly lower body mass index (BMI).
170
9
Thus, moving forward, combined comparisons of blood biomarkers as they relate to running
171
volume were adjusted for age, gender, and BMI.
172
173
Table 1. Study population demographics
174
Group
N
Female, %
Age, yrs
Body mass index, kg/m2
PRO
82
53.7%
33.68
20.15 ± 6.02
HVAM
1103
52.9%
39.48
22.57 ± 9.97
MVAM
6747
54.2%
41.49
23.35 ± 9.76
LVAM
10877
34.2%
41.16
24.72 ± 9.70
SED
4428
48.9%
44.25
27.83 ± 10.70
PRO = Professional, HVAM = high volume amateur (>10 h/week ), MVAM = medium volume
175
amateur (3-10 h/week), LVAM = low volume amateur (<3 h/week), SED = sedentary
176
177
Endurance exercise exhibits a modest association with clusters of
178
blood biomarker features
179
In order to begin to understand the most important variables that may associate with endurance
180
exercise in the form of running, we performed a principal component analysis (PCA), sub-dividing
181
the male and female cohorts into two most divergent groups in terms of exercise volume:
182
PRO/high volume amateur (HVAM) and sedentary (SED) groups. Using propensity matching,
183
PRO and amateur athletes who reported running >10h per week were combined into the PRO-
184
HVAM group to balance out the sample size between the exercising and non-exercising groups.
185
This approach yielded a modest degree of separation, with hematological, inflammation, and lipid
186
features, as well as BMI explaining some of the variance (Fig 1 A through D). We hypothesized
187
that there may more subtle relationships between running volume and the blood biomarker features
188
that contributed to distinguishing the endurance exercise and sedentary groups, thus we next
189
performed ANOVA analyses stratified by running volume as categorized in Table 1.
190
Fig 1. Principal component analysis and variables plots of PRO-HVAM runners and sedentary
191
user blood biomarkers. Females, (A) and (B); males (C) and (D). PRO-HVAM = combined
192
professional and high-volume amateur. Alb = albumin, ALT = alanine transaminase, AST =
193
aspartate aminotransferase, B12 = vitamin B12, Ca = calcium, Chol = total cholesterol, CK =
194
creatine kinase, Cor = cortisol, FE = iron, EOS_PCT = eosinophil percentage, Fer = ferritin, Fol =
195
folate, FT = free testosterone, GGT = gamma-glutamyl transferase, Glu = glucose, Hb =
196
hemoglobin, HCT = hematocrit, HDL = high density lipoprotein, HbA1c = glycated hemoglobin,
197
hsCRP = high-sensitivity C-reactive protein, LDL = low density lipoprotein, LYMPS_PCT =
198
lymphocyte percentage, MCH = mean cell hemoglobin, Mg = magnesium, MONOS_PCT =
199
monocytes percentage, MPV = mean platelet volume, Na = sodium, RBC = red blood cells,
200
RBC_Mg = red blood cell magnesium, RDW = red blood cell distribution width, SHBG = sex
201
hormone binding globulin, Tg = triglycerides, TIBC = total iron binding capacity, WBC = white
202
blood cells
203
Significant trends in glycemic, hematological, blood lipid, and
204
inflammatory serum traits with increasing running volumes
205
Weighted ANOVA analyses adjusted for age, gender, and BMI showed significant differences
206
among groups for multiple blood biomarkers (Table 2 and S2 , Figs 2 and 3). We observed a trend
207
11
toward lower HbA1c, hsCRP, RDW, WBC, ferritin, gamma-glutamyl transferase (GGT), and
208
LDL. HDL, hemoglobin (Hb), transferrin saturation (TS), alanine aminotransferase (ALT),
209
aspartate aminotransferase (AST), vitamin B12, folate, 25-hydroxy vitamin D, and creatine kinase
210
(CK) tended to be higher with increasing reported training volume, particularly in PRO runners
211
(Tables 2 and S2 , Figs 2 and S2 , Fig 3). Hct and Hb were higher only in PRO males, whereas
212
increased running volume associated with upward trend in these biomarkers in females (Fig 3 A
213
and B). Increased running volume was associated with markedly lower Fer in males, whereas
214
female runners did not exhibit varying levels, and SED females showed increased levels (Fig 3 C).
215
The low ferritin observed in male and female runners was not clinically significant. ALT
216
positively associated with running volume in females only (S2 Fig). Serum and RBC magnesium
217
(Mg) were both significantly lower in PRO runners relative to all other groups (Table 2 and Fig 3
218
D and E). Increasing levels of endurance exercise also appeared to be associated with higher sex-
219
hormone binding globulin (SHBG), particularly in PRO male runners (Fig 3 F).
220
221
Table 2. Blood biomarkers significantly different among sedentary individuals and those
222
who partake in running for exercise to various degrees
223
BIOMARKER
ANOVA P-VALUE TREND P-VALUE LOWEST MEAN
HIGHEST MEAN
ALB
<1e-16
<0.001
MVAM
PRO
ALT
<1e-16
<1e-16
SED
PRO
AST
<1e-16
<0.001
SED
PRO
B12
<0.001
<0.001
SED
PRO
CHOL
<0.001
0.005
PRO
SED
CK
<1e-16
<1e-16
SED
PRO
COR
<0.001
0.675
SED
PRO
FE
<0.001
0.119
SED
PRO
FER
<1e-16
<1e-16
MVAM
SED
FOL
<1e-16
<0.001
SED
PRO
FT
<0.001
0.013
SED
PRO
GGT
<1e-16
<0.001
PRO
SED
GLU
0.087
0.184
PRO
SED
HB
0.002
<0.001
MVAM
PRO
HCT
0.053
0.055
MVAM
PRO
HDL
<1e-16
<0.001
SED
PRO
HBA1C
<0.001
0.010
PRO
SED
HSCRP
<0.001
0.176
PRO
SED
LDL
<0.001
0.006
PRO
SED
MG
<0.001
0.276
PRO
SED
MPV
0.058
0.089
SED
HVAM
NA
<1e-16
0.622
HVAM
SED
RBC_MG
<0.001
0.773
PRO
SED
RDW
<1e-16
0.002
PRO
SED
SHBG
<1e-16
0.004
SED
PRO
TG
<1e-16
<1e-16
PRO
SED
13
WBC
<1e-16
<1e-16
PRO
SED
224
Fig 2. Blood biomarkers associated with running: Inflammation proxies, (A) hsCRP = high-
sensitivity C-reactive protein and (B) WBC = white blood cells; blood lipids, (C) HDL = high
density lipoprotein (D) LDL = low density lipoprotein, and (E) Tg = triglycerides; glycemia
proxies, (F) Glu = glucose and (G) HgbA1c = glycated hemoglobin, and (H) Cor =cortisol
Fig 3. Blood biomarkers associated with running: (A and B) Hb (hemoglobin) and Hct
(hematocrit) increase with increasing running volume, (C) Fer (ferritin) is reduced with
increasing running volume, (D and E) Serum and RBC Mg (red blood cell magnesium) are
reduced in professional runners, and (F) SHBG (sex hormone binding globulin) levels increase
with increasing running volume in males
225
Endurance exercise correlates with lower BMI across categories of
226
genetic risk
227
Using publicly available GWAS summary statistics, we constructed blood biomarker polygenic
228
risk scores (PGSs) to explore potential genetic risk-mitigating effects of endurance exercise. Since
229
only a subset of the individuals in our cohort were genotyped, we aggregated the groups into 2
230
categories—PRO-HVAM and sedentary—to increase statistical power. This across-group sample
231
size increase generally did not sufficiently power the ANOVA analysis to detect statistically
232
significant trends, though the BMI polygenic risk was suggestively mitigated for both males and
233
female PRO-HVAM runners across categories of genetic risk (Fig 4 B).
234
Fig 4. BMI significantly varied among running groups (A) with some suggestive effects on BMI
PGS modification (total number for observations (N) for T1, T2, and T3 were 87, 84, and 100,
respectively) (B) T1, T2, and T3 = 1st, and 2nd and 3rd tertials of the polygenic score distribution
235
Increased running volume is associated with lower BMI which may
236
drive biomarker changes
237
We observed a significant downward trend in the BMI with increased running volume for both
238
males and females, and, although some of the biomarker differences between sedentary and
239
exercising individuals remained significant after adjustment for BMI, their significance was
240
attenuated (Fig 4 A). Thus, we hypothesized that BMI may be driving a significant portion of the
241
observed variance in some of the biomarkers across the groups. Thus, to explore causal
242
relationships between weight and biomarker changes, we performed 2S-MR with BMI-associated
243
single-nucleotide polymorphisms (SNPs) as the instrumental variables (IVs) for a subset of the
244
healthspan-related biomarkers where BMI explained a relatively large portion of the variance in
245
our analysis. In general, these blood biomarkers associated with inflammation (hsCRP and
246
RDW), lipid metabolism (Tg and HDL), glycemic control (HbA1c and Glu), as well as Alb and
247
SHBG. We used GWAS summary statistics and found that most of these BMI-blood biomarker
248
relationships examined directionally aligned with our study (except for LDL), and some were
249
indicative of causal relationships in the BMI-biomarker direction even after considering directional
250
15
pleiotropy (S3 Table). We entertained the possibility of reverse causality and thus repeated the
251
2S-MR using each of the biomarker levels as the exposure and BMI as the outcome, and the results
252
were generally not significant (except for WBC – see S4 Table). Of note, to estimate the direct
253
causal effects of running on blood parameters, we attempted to find an instrumental variable for
254
to approximate running as the exposure from publicly available GWAS summary statistics.
255
Toward this end, we found that increasing levels of vigorous physical activity did associate with
256
lower hsCRP, HbA1C, higher HDL, and possibly higher SHBG (although the explained variance
257
(R2) in this exposure was just 0.001009, the F statistic was 37.7, thus meeting the criteria of F >
258
10 for minimizing weak instrument bias) (Figs 5 and S3; S5 Table).
259
Fig 5. Two-sample Mendelian randomization shows that increasing levels of vigorous physical
activity such as running is associated with improvement of (A) hsCRP = high-sensitivity C-
reactive protein, (B) HDL = high density lipoprotein, and (C) HbA1c = glycated hemoglobin
levels
260
Vigorous physical activity associates with healthier behaviors
261
We hypothesized that those who exercise regularly may also partake in other healthful lifestyle
262
habits that may be contributing to more optimal blood biomarker signatures of wellness. However,
263
our dataset did not allow for systematic accounting of other lifestyle habits across all running
264
groups. Thus, we again leveraged the potential of the 2S-MR approach to inform potential
265
confounding associations between modifiable exposures and found that vigorous physical activity
266
such as running is at least suggestively associated with several behaviors associated with improved
267
health (S4 Fig). Our analysis showed that those who participate in increasing levels of vigorous
268
physical activity may be less likely to eat processed meat (IVW p = 0.0000013), sweets (IVW p =
269
0.32), and nap during the day (IVW p = 0.13), while increasing their intake of oily fish (IVW p =
270
0.029), salad/raw vegetable intake (IVW p = 0.00016), and fresh fruit (IVW p = 0.0027) (S6
271
Table). Furthermore, following our assessment of reverse causality, we found evidence for the
272
bidirectionality in the causal relationship between vigorous activity and napping during the day
273
and salad/raw vegetable intake, perhaps suggesting some degree of confounding due to population
274
stratification (S7 Table). The suggestive positive effect of fresh fruit and processed meat intake on
275
vigorous physical activity appeared to violate MR assumption (3) (S1 Fig) (horizontal pleiotropy
276
p-values 0.051 and 0.17, respectively – S5 Fig).
277
Discussion
278
In this report, we describe the variance in wellness-related blood biomarkers among self-reported
279
recreational runners, PRO runners, and individuals who do not report any exercise. Overall, we
280
find that 1) recreational running as an exercise appears to be an effective intervention toward
281
modifying several biomarkers indicative of improved metabolic health, 2) an apparent dose-
282
response relationship between running volume and BMI may itself be responsible for a proportion
283
of the apparent metabolic benefits, and 3) both PRO-level status and gender appear to associate
284
with heterogeneous physiological responses, particularly in iron and magnesium metabolism, as
285
well as some hormonal traits.
286
Self-reported running improves glycemia and lipidemia
287
17
We did not observe distinct clusters corresponding to self-reported high-volume/PRO runners and
288
the sedentary upon dimension reduction. This is, perhaps, not unexpected due, in part, to the self-
289
selected healthspan-oriented nature of our cohort, where even the sedentary subset of individuals
290
tends to exhibit blood biomarker levels in the normal clinical reference ranges. Furthermore, the
291
measurement of running volume via self-report may be vulnerable to overestimation, which may
292
have contributed to the blending of sedentary and exercise groups with respect to the serum
293
markers measured, resulting in only marginal separation between the groups [19, 20]. However,
294
we did observe significant individual blood biomarker variance with respect to reported running
295
volumes when the dataset was subjected to ANOVA, even after adjustment for age, sex, and BMI.
296
From among glycemic control blood biomarkers, we were able to detect a relatively small exercise
297
effect in both fasting glucose and HbA1c in this generally healthy cohort, where the average
298
measures of glycemia were below the prediabetic thresholds in even the sedentary subset of the
299
cohort. Larger exercise intervention effects on metabolic biomarkers may be expected in cohorts
300
that include individuals with more clinically significant baseline values [21].
301
Similarly, blood lipids improved with higher self-reported running volume, and this result has been
302
reported before in multiple controlled endurance exercise trials [22]. The literature indicates that
303
HDL and Tg are two exercise-modifiable blood lipid biomarkers, with HDL being the most widely
304
reported to be modified by aerobic exercise [23, 24]. Although the mechanism behind this is not
305
entirely clear, it likely involves the modification of lecithin acyltransferase and lipoprotein lipase
306
activities following exercise training [25]. We observed a similar trend in our blood biomarker
307
analysis, with HDL exhibiting an upward trend with increasing reported running volume. While
308
we also found Tg and LDL to decrease with increasing exercise volume, these trends were less
309
pronounced. Reports generally suggest that, in order to reduce LDL more consistently, the
310
intensity of aerobic exercise must be high enough [23]. In the case of Tg, baseline levels may have
311
a significant impact on the exercise intervention effect, with individuals exhibiting higher baselines
312
showing greater improvements [13].
313
Importantly, these results suggests that exercise has a significant effect on glycemic control and
314
blood lipids even in the self-selected, already healthy individuals who are proactive about
315
preventing cardiometabolic disease.
316
Self-reported running and serum proxies of systemic inflammation
317
Chronic low-grade inflammation is one of the major risk factors for compromised cardiovascular
318
health and metabolic syndrome (MetS). While there is no shortage of inflammation-reducing
319
intervention studies on CVD patients with clinically high levels of metabolic inflammation, there
320
is less emphasis on modifiable lifestyle factors that can help stave off CVD and extend healthspan
321
in the generally healthy individual. Indeed, considering the pathological cardiovascular processes
322
begin shortly after birth, prevention in asymptomatic individuals may be a more appropriate
323
strategy toward decreasing the burden of CVD on the healthcare system [26].
324
Toward this end, increasing self-reported running volume appeared to associate with improved
325
markers of inflammation, as shown by the lower levels of hsCRP, WBC, as well as ferritin. Of
326
note, while the acute-phase protein, ferritin, is often used in the differential diagnosis of iron
327
deficiency anemia, the biomarker’s specificity appears to depend on the inflammatory state of the
328
individual, as it associates with hsCRP and inflammation more than iron stores, particularly in
329
those with higher BMI [27]. Although serum ferritin and iron is reported to be lower in male and
330
female elite athletes [28], the observed overall negative association of ferritin with increased
331
19
running volume in our cohort may be an indication of lower levels of inflammation rather than
332
compromised iron stores, particularly since the average ferritin level across all groups was above
333
the clinical iron deficiency thresholds. Moreover, increased levels of ferritin have been associated
334
with insulin resistance and lower levels of adiponectin in the general population, both indicators
335
of increased systemic inflammation [29]. Here, exercising groups with lower levels of ferritin also
336
exhibited glycemic and blood lipid traits indicative of improved metabolic states, further
337
supporting ferritin’s role as an inflammation proxy. Finally, Hb, TS and iron tended to be higher
338
in those who run for exercise compared to the SED group (with the TIBC lower), again suggesting
339
that runners, including the PRO group, were iron-sufficient in this cohort.
340
PRO endurance runners exhibit distinct biomarker signatures
341
PRO athletes exhibited lower serum and RBC Mg, which may be indication of the often-reported
342
endurance athlete hypomagnesaemia [30]. While the serum Mg was still within normal clinical
343
reference range for both PRO female and male athletes, RBC Mg, a more sensitive biomarker of
344
Mg status [31], was borderline low in female PRO athletes and might suggest suboptimal dietary
345
intakes and/or much higher volume of running training compared to the other running groups (i.e.
346
>>10h /week). Indeed, this group also had elevated baseline CK and AST, which suggests a much
347
higher training intensity and/or volume. Moreover, PRO level athletes had adequate iron status
348
and serum B12 and folate in the upper quartile of the normal reference range, suggesting that these
349
athletes’ general nutrition status may have been adequate. These observations suggest that elite
350
endurance runners may need to pay particular attention to their magnesium status.
351
Further, we observed higher levels of SHBG in PRO male runners, a biomarker whose levels
352
positively correlate with various indexes of insulin sensitivity [32]. However, since the average
353
SHBG levels in the SED group were not clinically low in both sexes, the observed increase in
354
SHBG levels induced by running in males may be a catabolic response, as cortisol levels in this
355
group were also higher. Indeed, Popovic et al have shown that endurance exercise may increase
356
SHBG, cortisol, and total testosterone levels at the expense of free testosterone levels [33]. This
357
could perhaps in part be explained by higher exercise-induced adiponectin levels, which have been
358
shown to increase SHBG via cAMP kinase (AMPK) activation [34]. However, since our data is
359
observational, we cannot rule out overall energy balance as a significant contributor to SHBG
360
levels. For example, caloric restriction (CR) has been shown to result in higher SHBG and cortisol
361
levels [32].
362
Finally, regarding the abovementioned PRO group elevated AST and CK biomarkers, evidence
363
suggests that normal reference ranges in both CK and AST in well-recovered athletes should be
364
adjusted up, as training and competition have a profound, non-pathological, impact on the activity
365
of these enzymes [35, 36]. Indeed, the recommendation appears to be not to use reference intervals
366
derived from the general population with hard-training (particularly competitive) athletes [36].
367
Effect of BMI on blood biomarkers
368
Since the current study is a cross-sectional analysis of self-reported running, we could not rule out
369
the possibility that factors other than exercise were the driving force behind the observed
370
biomarker variance among the groups examined. These factors, such as diet, sleep, and/or
371
medications were not readily ascertained in this free-living cohort at the time of this study, but
372
BMI was readily available to evaluate this biomarker’s potential relative contribution to the
373
observed mean biomarker differences among self-reported runner groups.
374
21
Multiple studies have attempted to uncouple the effects of exercise and BMI reduction on blood
375
biomarker outcomes, with mixed results [37]. For example, it is relatively well-known that acute
376
bouts of exercise improve glucose metabolism, but long-term effects are less well described [38].
377
Indeed, whether exercise without significant weight-loss is effective toward preventing metabolic
378
disease (and the associated blood biomarker changes) is inconclusive [39-41]. From the literature,
379
it appears that, for endurance exercise to have significant effect on most blood biomarkers, the
380
volume of exercise needs to be very high, and this typically results in significant reduction in
381
weight. Thus, in practice, it is difficult to demonstrably uncouple the effects of significant exercise
382
and the associated weight-loss, and the results may depend on the blood biomarker in question.
383
Indeed, there is evidence that exercise without weight-loss does improve markers of insulin
384
sensitivity but not chronic inflammation, with the latter apparently requiring a reduction in
385
adiposity in the general population [39-41].
386
In our study of apparently healthy individuals, we observed a downward trend in BMI with
387
increasing self-reported running volume, and, although this study was not longitudinal and we are
388
thus unable to claim weight-loss, our 2S-MR analysis using BMI as the exposure nonetheless
389
suggests this biomarker to be responsible for a significant proportion of the modification of some
390
blood biomarkers.
391
Serum markers of systemic inflammation
392
Through our 2S-MR analyses, we show that BMI is causally associated with markers of systemic
393
inflammation, including RDW, folate, and hsCRP [27, 42, 43]. Similar analyses have reported
394
that genetic variants that associate with higher BMI were associated with higher CRP levels, but
395
not the other way around [44]. The prevailing mechanism proposed to explain this relationship
396
appears to be the pathological nature of overweight/obesity-driven adipose tissue that results in
397
secretion of proinflammatory cytokines such as IL-6 and TNFa, which then stimulate an acute
398
hepatic response, resulting in increased hsCRP levels (among other effects) [45]. Thus, our 2S-
399
MR analyses and those of others [44] would indicate that the primary factor behind the lower
400
systemic inflammation in our cohort may be the exercise-associated lower BMI and not running
401
exercise per se, though the lower hsCRP in runners remained significant after adjustment for BMI
402
in our analysis.
403
Indeed, although a major driver behind reduced systemic inflammation may be a reduction in BMI
404
in the general population, additive effects of other lifestyle factors such as exercise cannot be
405
excluded. For example, a large body of cross-sectional investigations does indicate that physically
406
active individuals exhibit CRP levels that are 19-35% lower than less active individuals, even
407
when adjusted for BMI as was the case in the current analysis [41]. Further, it’s been reported that
408
physical activity at a frequency of as little as 1 day per week is associated with lower CRP in
409
individuals who are otherwise sedentary, while more frequent exercise further reduces
410
inflammation [41].
411
Significantly, our entire cohort of self-selected apparently healthy individuals did not exhibit
412
clinically high hsCRP, with average BMI also below the overweight thresholds. Because all
413
subjects were voluntarily participating in a personalized wellness platform intended to optimize
414
blood biomarkers that included hsCRP, it is possible that some individuals from across the study
415
groups (both running and sedentary) in our cohort partook in some form of inflammation-reducing
416
dietary and/or lifestyle-based intervention. Thus, that we detected a significant difference in
417
hsCRP between exercising and non-exercising individuals in this self-selected already generally
418
23
healthy cohort may be suggestive of the potential for additional preventative effect of scheduled
419
physical activity on low-grade systemic inflammation in the generally healthy individual.
420
Blood lipids
421
Controlled studies that tightly track exercise and the associated adiposity reduction have reported
422
that body fat reduction (and not improvement in fitness as measured via VO2max) is a predictor of
423
HDL, LDL, and Tg [46]. Similarly, though BMI is an imperfect measure of adiposity, our 2S-MR
424
analysis suggests that this biomarker is causally associated with improved levels of HDL and Tg,
425
though not LDL. This latter finding replicates a report by Hu et al. who, using the Global Lipids
426
Genetics Consortium GWAS summary statistics, applied a network MR approach that revealed
427
causal associations between BMI and blood lipids, where Tg and HDL, but not LDL, were found
428
to trend toward unhealthy levels with increasing adiposity [47]. On the other hand, others
429
implemented a robust BMI genetic risk score and demonstrated a causal association of adiposity
430
with peripheral artery disease and a multiple linear regression showed a strong association with
431
HDL, TC, and LDL, among other metabolic parameters [48]. In our cohort, given the lack of
432
evidence for a causal BMI-LDL association and the overall healthy levels of BMI, the observed a
433
significant improvement in LDL may be a result of marked running intensity and/or volume,
434
possibly combined with the aforementioned additional wellness program intervention variables.
435
Hormonal traits
436
As described above, we observed a trend toward increased plasma cortisol and SHBG in runners,
437
particularly PRO level athletes. The effects on cortisol are consistent with a report by Houmanrd
438
et al, who found male distance runners to exhibit higher levels of baseline cortisol [49]. With
439
respect to the effects of BMI on baseline cortisol levels, this observation is generally supported by
440
our 2S-MR analyses with evidence for a consistent effect of increased cortisol with decreasing
441
BMI. However, this association was suggestive at best, indicating that the higher levels of cortisol
442
exhibited in the PRO runners with significant lower adiposity are not likely to be solely explained
443
by their lower BMI. Indeed, the relationship between BMI and cortisol appears to be complex,
444
with some reports suggesting a U-shaped relationship, where the glucocorticoid’s levels associate
445
negatively up to about a BMI of 30 kg/m2, then exhibiting a positive correlation into obesity
446
phenotypes [50]. MR statistical models generally do not account for such non-linearity and would
447
require a more granular demographical treatment, which is not possible using only GWAS
448
summary statistics data in the context of 2S-MR [17, 51].
449
Behavioral traits associated with increase physical activity
450
The combination of the body of the literature that describes the effects of endurance training on
451
blood biomarkers, and our own analysis that included markers such as CK and AST, makes us
452
cautiously assured that most of the abovementioned blood biomarker signatures are indeed a result
453
of the interplay between self-reported running and the associated lower BMI. However, as this is
454
a self-report-based analysis and we were unable to track other subject behaviors in this free-living
455
cohort, we acknowledge that multiple behaviors that associate with exercise may be influencing
456
our results.
457
Toward this end, our exploratory 2S-MR analyses revealed potentially causal relationships
458
between vigorous exercise and multiple dietary habits that have been shown to improve the
459
biomarkers we examined. Indeed, diets that avoid processed meat and sweets while providing
460
ample amounts of fresh fruits, as well as oily fish have been shown to be anti-inflammatory, and
461
25
improve glycemic control and dyslipidemia [52, 53]. That physically active individuals are also
462
more likely to make healthier dietary choices adds insight to the potential confounders in ours and
463
others’ observational analyses, and this similar associations have previously been reported [54-
464
56]. For example, using a calculated healthy eating motivation score, Naughton et al. showed that
465
those who partake in more than 2 hours of vigorous physical activity are almost twice as likely to
466
be motivated to eat healthy [56]. Indeed, upon closer examination, the genetic instruments used
467
to approximate vigorous physical activity as the exposure in this work included variants in the
468
genes DPY19L1, CADM2, CTBP2, EXOC4, and FOXO3 [57]. Of these, CADM2 encodes proteins
469
that are involved in neurotransmission in brain regions well known for their involvement in
470
executive function, including motivation, impulse regulation and self-control [58]. Moreover,
471
variants within this locus have been associated with obesity-related traits [59]. Thus, it is likely
472
that the improved metabolic outcomes seen here with our self-reported runners are a composite
473
result of both these individuals exercise and dietary habits. Importantly, the above suggests that a
474
holistic wellness lifestyle approach is in practice the most likely to be most effective toward
475
preventing cardiometabolic disease. Nonetheless, the focus of this work – exercise in the form of
476
running – is known to significantly improve cardiorespiratory fitness (CRF), which has been
477
shown to be an independent predictor of CVD risk and total mortality, outcomes that indeed
478
correlate with dysregulated levels in many of the blood biomarkers examined in this work [7].
479
Study limitations
480
This study is based on self-reported running and thus has several limitations. First, it is generally
481
known that subjects tend to overestimate their commitment to exercise when self-reporting,
482
although in our cohort is a self-selected health-oriented population that is possibly less likely to
483
over-report their running volume. Furthermore, although the robust increasing trend in baselines
484
for muscle damage biomarkers (CK, AST) that have been shown to be associated with participation
485
in sports and exercise provides indirect evidence that the running groups were indeed participating
486
in increasing volumes of strenuous physical activity, we cannot confirm whether the reported
487
running was performed overground or on a treadmill, which may result in some heterogeneity in
488
physiological responses , nor can we ascertain the actual training volume of PRO-level runners.
489
We also cannot exclude the possibility that the running groups also participated in other forms of
490
exercise (such as strength training) or partook in other wellness program interventions that may
491
have influenced their blood biomarkers and/or BMI via lean muscle accretion. Toward this end,
492
we have attempted to shed light on potential behavioral covariates related to vigorous physical
493
activity via 2S-MR. Finally, while this cohort is generally healthy, we cannot exclude the potential
494
for unmeasured confounders such as medications, nutritional supplements, and unreported health
495
conditions.
496
2S- MR enables the assessment of causal relationships between modifiable traits and is less prone
497
to the so-called “winner’s curse” that more readily affects one-sample MR analyses [17, 51].
498
Because 2S-MR uses GWAS summary statistics for both exposure and outcome, it is possible to
499
increase statistical power because of the increased sample sizes. However, horizontal pleiotropy
500
is still a concern that can skew the results. Currently, there is no gold standard MR analysis
501
method, thus we used different techniques (IVW, MR-Egger, and median-based estimations – all
502
of which are based on different assumptions and thus biases) to evaluate the consistency among
503
these estimators and only reported associations as ‘causal’ if there was cross-model consistency.
504
Nonetheless, an exposure such as BMI is a complex trait that is composed of multiple sub-
505
phenotypes (such as years of education) that could be driving the causal associations.
506
27
Conclusions
507
Running is one of the most common forms of vigorous exercise practiced globally, thus making it
508
a compelling target of research studies toward understanding its applicability in chronic disease
509
prevention. Our cross-sectional study offers insight into the biomarker signatures of self-reported
510
running in generally healthy individuals that suggest improved insulin sensitivity, blood lipid
511
metabolism, and systemic inflammation. Furthermore, using 2S-MR in independent datasets we
512
provide additional evidence that some biomarkers are readily modified BMI alone, while others
513
appear
to
respond
to
the
combination
of
varying
exercise
and
BM
514
I. Our additional bi-directional 2S-MR analyses toward understanding the causal relationships
515
between partaking in vigorous physical activity and other healthy behaviors highlight the inherent
516
challenge in disambiguating exercise intervention effects in cross sectional studies of free-living
517
populations, where healthy behaviors such as exercising and healthy dietary habits co-occur.
518
Overall, our analysis shows that the differences between those who run and the sedentary in our
519
cohort are likely a combination of the specific physiological effects of exercise, the associated
520
changes in BMI, and lifestyle habits associated with those who exercise, such as food choices and
521
baseline activity level. Looking ahead, the InsideTracker database is continuously augmented
522
with blood chemistry, genotyping, and activity tracker data, facilitating further investigation of the
523
effects of various exercise modalities on phenotypes related to healthspan, including longitudinal
524
analyses and more granular dose-response dynamics.
525
Acknowledgments
526
We thank Michelle Cawley and Renee Deehan for their assistance with background subject matter
527
research and insightful conversations.
528
References
529
1.
Lavie CJ, Ozemek C, Carbone S, Katzmarzyk PT, Blair SN. Sedentary Behavior,
530
Exercise, and Cardiovascular Health. Circ Res. 2019;124(5):799-815. doi:
531
10.1161/CIRCRESAHA.118.312669. PubMed PMID: 30817262.
532
2.
Carlson SA, Adams EK, Yang Z, Fulton JE. Percentage of Deaths Associated With
533
Inadequate Physical Activity in the United States. Prev Chronic Dis. 2018;15:E38. Epub
534
20180329. doi: 10.5888/pcd18.170354. PubMed PMID: 29602315; PubMed Central PMCID:
535
PMCPMC5894301.
536
3.
Antonicelli R, Spazzafumo L, Scalvini S, Olivieri F, Matassini MV, Parati G, et al.
537
Exercise: a "new drug" for elderly patients with chronic heart failure. Aging (Albany NY).
538
2016;8(5):860-72. doi: 10.18632/aging.100901. PubMed PMID: 26953895; PubMed Central
539
PMCID: PMCPMC4931840.
540
4.
Sgro P, Emerenziani GP, Antinozzi C, Sacchetti M, Di Luigi L. Exercise as a drug for
541
glucose management and prevention in type 2 diabetes mellitus. Curr Opin Pharmacol.
542
2021;59:95-102. Epub 20210626. doi: 10.1016/j.coph.2021.05.006. PubMed PMID: 34182427.
543
5.
Vina J, Sanchis-Gomar F, Martinez-Bello V, Gomez-Cabrera MC. Exercise acts as a
544
drug; the pharmacological benefits of exercise. Br J Pharmacol. 2012;167(1):1-12. doi:
545
10.1111/j.1476-5381.2012.01970.x. PubMed PMID: 22486393; PubMed Central PMCID:
546
PMCPMC3448908.
547
6.
Lee EC, Fragala MS, Kavouras SA, Queen RM, Pryor JL, Casa DJ. Biomarkers in Sports
548
and Exercise: Tracking Health, Performance, and Recovery in Athletes. J Strength Cond Res.
549
2017;31(10):2920-37. doi: 10.1519/JSC.0000000000002122. PubMed PMID: 28737585;
550
PubMed Central PMCID: PMCPMC5640004.
551
7.
Lin X, Zhang X, Guo J, Roberts CK, McKenzie S, Wu WC, et al. Effects of Exercise
552
Training on Cardiorespiratory Fitness and Biomarkers of Cardiometabolic Health: A Systematic
553
Review and Meta-Analysis of Randomized Controlled Trials. J Am Heart Assoc. 2015;4(7).
554
Epub 20150626. doi: 10.1161/JAHA.115.002014. PubMed PMID: 26116691; PubMed Central
555
PMCID: PMCPMC4608087.
556
8.
Li X, Ploner A, Wang Y, Zhan Y, Pedersen NL, Magnusson PK, et al. Clinical
557
biomarkers and associations with healthspan and lifespan: Evidence from observational and
558
genetic data. EBioMedicine. 2021;66:103318. Epub 2021/04/05. doi:
559
10.1016/j.ebiom.2021.103318. PubMed PMID: 33813140; PubMed Central PMCID:
560
PMCPMC8047464.
561
9.
Mailliez A, Guilbaud A, Puisieux F, Dauchet L, Boulanger E. Circulating biomarkers
562
characterizing physical frailty: CRP, hemoglobin, albumin, 25OHD and free testosterone as best
563
biomarkers. Results of a meta-analysis. Exp Gerontol. 2020;139:111014. Epub 20200626. doi:
564
10.1016/j.exger.2020.111014. PubMed PMID: 32599147.
565
10.
Hirata T, Arai Y, Yuasa S, Abe Y, Takayama M, Sasaki T, et al. Associations of
566
cardiovascular biomarkers and plasma albumin with exceptional survival to the highest ages. Nat
567
29
Commun. 2020;11(1):3820. Epub 20200730. doi: 10.1038/s41467-020-17636-0. PubMed PMID:
568
32732919; PubMed Central PMCID: PMCPMC7393489.
569
11.
Erema VV, Yakovchik AY, Kashtanova DA, Bochkaeva ZV, Ivanov MV, Sosin DV, et
570
al. Biological Age Predictors: The Status Quo and Future Trends. Int J Mol Sci. 2022;23(23).
571
Epub 20221201. doi: 10.3390/ijms232315103. PubMed PMID: 36499430; PubMed Central
572
PMCID: PMCPMC9739540.
573
12.
Hartmann A, Hartmann C, Secci R, Hermann A, Fuellen G, Walter M. Ranking
574
Biomarkers of Aging by Citation Profiling and Effort Scoring. Front Genet. 2021;12:686320.
575
Epub 20210521. doi: 10.3389/fgene.2021.686320. PubMed PMID: 34093670; PubMed Central
576
PMCID: PMCPMC8176216.
577
13.
Trejo-Gutierrez JF, Fletcher G. Impact of exercise on blood lipids and lipoproteins. J Clin
578
Lipidol. 2007;1(3):175-81. Epub 20070607. doi: 10.1016/j.jacl.2007.05.006. PubMed PMID:
579
21291678.
580
14.
Westerman K, Reaver A, Roy C, Ploch M, Sharoni E, Nogal B, et al. Longitudinal
581
analysis of biomarker data from a personalized nutrition platform in healthy subjects. Sci Rep.
582
2018;8(1):14685. Epub 2018/10/04. doi: 10.1038/s41598-018-33008-7. PubMed PMID:
583
30279436; PubMed Central PMCID: PMCPMC6168584.
584
15.
Fox J WS. An R Companion to Applied Regression. Third ed: Sage, Thousand Oaks CA;
585
2019.
586
16.
Ho D, Imai K, King G, Stuart EA. MatchIt: Nonparametric Preprocessing for Parametric
587
Causal Inference. Journal of Statistical Software. 2011;42(8):1 - 28. doi: 10.18637/jss.v042.i08.
588
17.
Burgess S, Davey Smith G, Davies NM, Dudbridge F, Gill D, Glymour MM, et al.
589
Guidelines for performing Mendelian randomization investigations. Wellcome Open Res.
590
2019;4:186. Epub 20200428. doi: 10.12688/wellcomeopenres.15555.2. PubMed PMID:
591
32760811; PubMed Central PMCID: PMCPMC7384151.
592
18.
Hemani G, Zheng J, Elsworth B, Wade KH, Haberland V, Baird D, et al. The MR-Base
593
platform supports systematic causal inference across the human phenome. Elife. 2018;7. Epub
594
20180530. doi: 10.7554/eLife.34408. PubMed PMID: 29846171; PubMed Central PMCID:
595
PMCPMC5976434.
596
19.
Bogl LH, Pietilainen KH, Rissanen A, Kaprio J. Improving the accuracy of self-reports
597
on diet and physical exercise: the co-twin control method. Twin Res Hum Genet.
598
2009;12(6):531-40. doi: 10.1375/twin.12.6.531. PubMed PMID: 19943715.
599
20.
Yuen HK, Wang E, Holthaus K, Vogtle LK, Sword D, Breland HL, et al. Self-reported
600
versus objectively assessed exercise adherence. Am J Occup Ther. 2013;67(4):484-9. doi:
601
10.5014/ajot.2013.007575. PubMed PMID: 23791324; PubMed Central PMCID:
602
PMCPMC3722661.
603
21.
Ostman C, Smart NA, Morcos D, Duller A, Ridley W, Jewiss D. The effect of exercise
604
training on clinical outcomes in patients with the metabolic syndrome: a systematic review and
605
meta-analysis. Cardiovasc Diabetol. 2017;16(1):110. Epub 20170830. doi: 10.1186/s12933-017-
606
0590-y. PubMed PMID: 28854979; PubMed Central PMCID: PMCPMC5577843.
607
22.
Fikenzer K, Fikenzer S, Laufs U, Werner C. Effects of endurance training on serum
608
lipids. Vascul Pharmacol. 2018;101:9-20. Epub 20171201. doi: 10.1016/j.vph.2017.11.005.
609
PubMed PMID: 29203287.
610
23.
Mann S, Beedie C, Jimenez A. Differential effects of aerobic exercise, resistance training
611
and combined exercise modalities on cholesterol and the lipid profile: review, synthesis and
612
recommendations. Sports Med. 2014;44(2):211-21. doi: 10.1007/s40279-013-0110-5. PubMed
613
PMID: 24174305; PubMed Central PMCID: PMCPMC3906547.
614
24.
Tambalis K, Panagiotakos DB, Kavouras SA, Sidossis LS. Responses of blood lipids to
615
aerobic, resistance, and combined aerobic with resistance exercise training: a systematic review
616
of current evidence. Angiology. 2009;60(5):614-32. Epub 20081030. doi:
617
10.1177/0003319708324927. PubMed PMID: 18974201.
618
25.
Calabresi L, Franceschini G. Lecithin:cholesterol acyltransferase, high-density
619
lipoproteins, and atheroprotection in humans. Trends Cardiovasc Med. 2010;20(2):50-3. doi:
620
10.1016/j.tcm.2010.03.007. PubMed PMID: 20656215.
621
26.
Kemper HC, Snel J, Verschuur R, Storm-van Essen L. Tracking of health and risk
622
indicators of cardiovascular diseases from teenager to adult: Amsterdam Growth and Health
623
Study. Prev Med. 1990;19(6):642-55. doi: 10.1016/0091-7435(90)90061-n. PubMed PMID:
624
2263575.
625
27.
Khan A, Khan WM, Ayub M, Humayun M, Haroon M. Ferritin Is a Marker of
626
Inflammation rather than Iron Deficiency in Overweight and Obese People. J Obes.
627
2016;2016:1937320. Epub 20161227. doi: 10.1155/2016/1937320. PubMed PMID: 28116148;
628
PubMed Central PMCID: PMCPMC5223018.
629
28.
Nabhan D, Bielko S, Sinex JA, Surhoff K, Moreau WJ, Schumacher YO, et al. Serum
630
ferritin distribution in elite athletes. J Sci Med Sport. 2020;23(6):554-8. Epub 20191227. doi:
631
10.1016/j.jsams.2019.12.027. PubMed PMID: 31901316.
632
29.
Ku BJ, Kim SY, Lee TY, Park KS. Serum ferritin is inversely correlated with serum
633
adiponectin level: population-based cross-sectional study. Dis Markers. 2009;27(6):303-10. doi:
634
10.3233/DMA-2009-0676. PubMed PMID: 20075513; PubMed Central PMCID:
635
PMCPMC3835072.
636
30.
Pollock N, Chakraverty R, Taylor I, Killer SC. An 8-year Analysis of Magnesium Status
637
in Elite International Track & Field Athletes. J Am Coll Nutr. 2020;39(5):443-9. Epub
638
20191212. doi: 10.1080/07315724.2019.1691953. PubMed PMID: 31829845.
639
31.
Arnaud MJ. Update on the assessment of magnesium status. Br J Nutr. 2008;99 Suppl
640
3:S24-36. doi: 10.1017/S000711450800682X. PubMed PMID: 18598586.
641
32.
Simo R, Saez-Lopez C, Barbosa-Desongles A, Hernandez C, Selva DM. Novel insights
642
in SHBG regulation and clinical implications. Trends Endocrinol Metab. 2015;26(7):376-83.
643
Epub 20150601. doi: 10.1016/j.tem.2015.05.001. PubMed PMID: 26044465.
644
33.
Popovic B, Popovic D, Macut D, Antic IB, Isailovic T, Ognjanovic S, et al. Acute
645
Response to Endurance Exercise Stress: Focus on Catabolic/anabolic Interplay Between Cortisol,
646
Testosterone, and Sex Hormone Binding Globulin in Professional Athletes. J Med Biochem.
647
2019;38(1):6-12. Epub 20190301. doi: 10.2478/jomb-2018-0016. PubMed PMID: 30820178;
648
PubMed Central PMCID: PMCPMC6298450.
649
34.
Simo R, Saez-Lopez C, Lecube A, Hernandez C, Fort JM, Selva DM. Adiponectin
650
upregulates SHBG production: molecular mechanisms and potential implications.
651
Endocrinology. 2014;155(8):2820-30. Epub 20140514. doi: 10.1210/en.2014-1072. PubMed
652
PMID: 24828613.
653
35.
Mougios V. Reference intervals for serum creatine kinase in athletes. Br J Sports Med.
654
2007;41(10):674-8. Epub 20070525. doi: 10.1136/bjsm.2006.034041. PubMed PMID:
655
17526622; PubMed Central PMCID: PMCPMC2465154.
656
31
36.
Banfi G, Morelli P. Relation between body mass index and serum aminotransferases
657
concentrations in professional athletes. J Sports Med Phys Fitness. 2008;48(2):197-200. PubMed
658
PMID: 18427415.
659
37.
Ross R, Janiszewski PM. Is weight loss the optimal target for obesity-related
660
cardiovascular disease risk reduction? Can J Cardiol. 2008;24 Suppl D:25D-31D. doi:
661
10.1016/s0828-282x(08)71046-8. PubMed PMID: 18787733; PubMed Central PMCID:
662
PMCPMC2794451.
663
38.
Ross R. Does exercise without weight loss improve insulin sensitivity? Diabetes Care.
664
2003;26(3):944-5. doi: 10.2337/diacare.26.3.944. PubMed PMID: 12610063.
665
39.
Cerqueira E, Marinho DA, Neiva HP, Lourenco O. Inflammatory Effects of High and
666
Moderate Intensity Exercise-A Systematic Review. Front Physiol. 2019;10:1550. Epub
667
20200109. doi: 10.3389/fphys.2019.01550. PubMed PMID: 31992987; PubMed Central
668
PMCID: PMCPMC6962351.
669
40.
Church TS, Earnest CP, Thompson AM, Priest EL, Rodarte RQ, Saunders T, et al.
670
Exercise without weight loss does not reduce C-reactive protein: the INFLAME study. Med Sci
671
Sports Exerc. 2010;42(4):708-16. doi: 10.1249/MSS.0b013e3181c03a43. PubMed PMID:
672
19952828; PubMed Central PMCID: PMCPMC2919641.
673
41.
Plaisance EP, Grandjean PW. Physical activity and high-sensitivity C-reactive protein.
674
Sports Med. 2006;36(5):443-58. doi: 10.2165/00007256-200636050-00006. PubMed PMID:
675
16646631.
676
42.
Salvagno GL, Sanchis-Gomar F, Picanza A, Lippi G. Red blood cell distribution width:
677
A simple parameter with multiple clinical applications. Crit Rev Clin Lab Sci. 2015;52(2):86-
678
105. Epub 20141223. doi: 10.3109/10408363.2014.992064. PubMed PMID: 25535770.
679
43.
Carmel R, Green R, Rosenblatt DS, Watkins D. Update on cobalamin, folate, and
680
homocysteine. Hematology Am Soc Hematol Educ Program. 2003:62-81. doi:
681
10.1182/asheducation-2003.1.62. PubMed PMID: 14633777.
682
44.
Welsh P, Polisecki E, Robertson M, Jahn S, Buckley BM, de Craen AJ, et al. Unraveling
683
the directional link between adiposity and inflammation: a bidirectional Mendelian
684
randomization approach. J Clin Endocrinol Metab. 2010;95(1):93-9. Epub 20091111. doi:
685
10.1210/jc.2009-1064. PubMed PMID: 19906786; PubMed Central PMCID:
686
PMCPMC2805500.
687
45.
Maachi M, Pieroni L, Bruckert E, Jardel C, Fellahi S, Hainque B, et al. Systemic low-
688
grade inflammation is related to both circulating and adipose tissue TNFalpha, leptin and IL-6
689
levels in obese women. Int J Obes Relat Metab Disord. 2004;28(8):993-7. doi:
690
10.1038/sj.ijo.0802718. PubMed PMID: 15211360.
691
46.
Katzmarzyk PT, Leon AS, Rankinen T, Gagnon J, Skinner JS, Wilmore JH, et al.
692
Changes in blood lipids consequent to aerobic exercise training related to changes in body
693
fatness and aerobic fitness. Metabolism. 2001;50(7):841-8. doi: 10.1053/meta.2001.24190.
694
PubMed PMID: 11436192.
695
47.
Hu X, Zhuang XD, Mei WY, Liu G, Du ZM, Liao XX, et al. Exploring the causal
696
pathway from body mass index to coronary heart disease: a network Mendelian randomization
697
study. Ther Adv Chronic Dis. 2020;11:2040622320909040. Epub 20200527. doi:
698
10.1177/2040622320909040. PubMed PMID: 32523662; PubMed Central PMCID:
699
PMCPMC7257848.
700
48.
Huang Y, Xu M, Xie L, Wang T, Huang X, Lv X, et al. Obesity and peripheral arterial
701
disease: A Mendelian Randomization analysis. Atherosclerosis. 2016;247:218-24. Epub
702
20151229. doi: 10.1016/j.atherosclerosis.2015.12.034. PubMed PMID: 26945778.
703
49.
Houmard JA, Costill DL, Mitchell JB, Park SH, Fink WJ, Burns JM. Testosterone,
704
cortisol, and creatine kinase levels in male distance runners during reduced training. Int J Sports
705
Med. 1990;11(1):41-5. doi: 10.1055/s-2007-1024760. PubMed PMID: 2180832.
706
50.
Schorr M, Lawson EA, Dichtel LE, Klibanski A, Miller KK. Cortisol Measures Across
707
the Weight Spectrum. J Clin Endocrinol Metab. 2015;100(9):3313-21. Epub 20150714. doi:
708
10.1210/JC.2015-2078. PubMed PMID: 26171799; PubMed Central PMCID:
709
PMCPMC4570173.
710
51.
O'Donnell CJ, Sabatine MS. Opportunities and Challenges in Mendelian Randomization
711
Studies to Guide Trial Design. JAMA Cardiol. 2018;3(10):967. doi:
712
10.1001/jamacardio.2018.2863. PubMed PMID: 30326490.
713
52.
Hosseini B, Berthon BS, Saedisomeolia A, Starkey MR, Collison A, Wark PAB, et al.
714
Effects of fruit and vegetable consumption on inflammatory biomarkers and immune cell
715
populations: a systematic literature review and meta-analysis. Am J Clin Nutr. 2018;108(1):136-
716
55. doi: 10.1093/ajcn/nqy082. PubMed PMID: 29931038.
717
53.
Djousse L, Arnett DK, Coon H, Province MA, Moore LL, Ellison RC. Fruit and
718
vegetable consumption and LDL cholesterol: the National Heart, Lung, and Blood Institute
719
Family Heart Study. Am J Clin Nutr. 2004;79(2):213-7. doi: 10.1093/ajcn/79.2.213. PubMed
720
PMID: 14749225.
721
54.
L D. Physical Activity and Dietary Habits of College Students. The Journal of Nurse
722
Practitioners. 2015;11(2):192-8.e2.
723
55.
Shi X, Tubb L, Fingers ST, Chen S, Caffrey JL. Associations of physical activity and
724
dietary behaviors with children's health and academic problems. J Sch Health. 2013;83(1):1-7.
725
doi: 10.1111/j.1746-1561.2012.00740.x. PubMed PMID: 23253284.
726
56.
Naughton P, McCarthy SN, McCarthy MB. The creation of a healthy eating motivation
727
score and its association with food choice and physical activity in a cross sectional sample of
728
Irish adults. Int J Behav Nutr Phys Act. 2015;12:74. Epub 20150606. doi: 10.1186/s12966-015-
729
0234-0. PubMed PMID: 26048166; PubMed Central PMCID: PMCPMC4475298.
730
57.
Klimentidis YC, Raichlen DA, Bea J, Garcia DO, Wineinger NE, Mandarino LJ, et al.
731
Genome-wide association study of habitual physical activity in over 377,000 UK Biobank
732
participants identifies multiple variants including CADM2 and APOE. Int J Obes (Lond).
733
2018;42(6):1161-76. Epub 20180613. doi: 10.1038/s41366-018-0120-3. PubMed PMID:
734
29899525; PubMed Central PMCID: PMCPMC6195860.
735
58.
Arends RM, Pasman JA, Verweij KJH, Derks EM, Gordon SD, Hickie I, et al.
736
Associations between the CADM2 gene, substance use, risky sexual behavior, and self-control:
737
A phenome-wide association study. Addict Biol. 2021;26(6):e13015. Epub 20210218. doi:
738
10.1111/adb.13015. PubMed PMID: 33604983; PubMed Central PMCID: PMCPMC8596397.
739
59.
Morris J, Bailey MES, Baldassarre D, Cullen B, de Faire U, Ferguson A, et al. Genetic
740
variation in CADM2 as a link between psychological traits and obesity. Scientific Reports.
741
2019;9(1):7339. doi: 10.1038/s41598-019-43861-9.
742
743
Supporting information
744
33
S1 Table. Number of people in each category by age group. Significant trend toward
745
younger individuals reporting higher running volume, with more than 75% of the elite
746
group falling between the ages of 18 and 35.
747
S2 Table. Full running volume vs. blood biomarker results
748
S3 Table. 2S-MR results with BMI as the exposure and select biomarkers as outcomes.
749
S4 Table. 2S-MR results with BMI with biomarkers as exposures and BMI as outcome to
750
assess reverse causality
751
S5 Table. 2S-MR results with vigorous physical activity as exposure and blood biomarkers
752
as outcomes
753
S6 Table. 2S-MR results with vigorous physical activity as exposure and lifestyle habits as
754
outcomes
755
S7 Table. 2S-MR with healthy/unhealthy dietary habits as exposures and vigorous physical
756
activity as outcome to assess reverse causality
757
S1 Fig. Assumptions of Mendelian randomization
758
S2 Fig. Blood biomarker levels with respect to self-reported running volume and
759
professional athletes
760
S3 Fig. 2S-MR scatter plot showing effects of vigorous physical activity as the exposure on
761
blood biomarkers.
762
S4 Fig. 2S-MR scatter plot showing effects of vigorous physical activity as the exposure
763
dietary habits.
764
S5 Fig. 2S-MR scatter plot showing effects of dietary behaviors as the exposures on vigorous
765
physical activity
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
Fig. 1
Click here to access/download;Figure;Fig1.tif
Fig. 2
Click here to access/download;Figure;Fig2.tif
Fig. 3
Click here to access/download;Figure;Fig3.tif
Fig. 4
Click here to access/download;Figure;Fig4.tif
Fig. 5
Click here to access/download;Figure;Fig5.tif
Supporting Information
Click here to access/download
Supporting Information
Supplementary_Materials_PONE_rev.pdf
Minimum dataset
Click here to access/download
Supporting Information
Dataset.txt
1
Dose response of running on blood biomarkers of wellness in the
generally healthy individuals
Bartek Nogal PhD1¶ ナ, Svetlana Vinogradova PhD1ナ¶, Milena Jorge MD,PhD1, Ali
Torkamani PhD 2,3, Paul Fabian BSc1, and Gil Blander PhD1*
1InsideTracker, Cambridge, Massachusetts, United States of America.
2The Scripps Translational Science Institute, The Scripps Research Institute, La Jolla, CA,
United States of AmericaUSA.
3Department of Integrative Structural and Computational Biology, The Scripps Research
Institute, La Jolla, CA, United States of AmericaUSA.
* Corresponding author
E-mail: [email protected] (GB)
ナ¶Equal Contribution: These authors contributed equally to this work
* Correspondence and reprint requests: Gil Blander, [email protected]
Bartek Nogal
Formatted: Normal, Line spacing: single
Formatted: Indent: Left: 0", Line spacing: single
Formatted: Font:
Formatted: Indent: Left: 0", Line spacing: single
Formatted: Font: Times New Roman
Field Code Changed
Formatted: Indent: Left: 0"
Revised Manuscript with Track Changes
1
[email protected]
1
Svetlana Vinogradova
2
[email protected]
3
Ali Torkamani
4
[email protected]
5
Paul Fabian
6
[email protected]
7
Running title: Biomarker signature of runners
8
Abstract
9
Exercise is effective toward delaying or preventing chronic disease, with a large body of evidence
10
supporting its effectiveness. However, less is known about the specific healthspan-promoting
11
effects of exercise on blood biomarkers in the disease-free population. In this work, we examine
12
23,237 generally healthy individuals who self-report varying weekly running volumes and
13
compare them to 4,428 generally healthy sedentary individuals, as well as 82 professional
14
endurance athletesrunners. We estimate the significance of differences among blood biomarkers
15
for groups of increasing running levels using analysis of variance (ANOVA), adjusting for age,
16
gender, and BMI. We attempt and add insight to our observational dataset analysis via two-sample
17
Mendelian randomization (2S-MR) using large independent datasets. We find that self-reported
18
Formatted: Font: 18 pt
Formatted: Header distance from edge: 0.2", Footer
distance from edge: 0.35", Numbering: Continuous
running volume associates with biomarker signatures of improved wellness, with some serum
19
markers apparently being principally modified by BMI, whereas others show a dose-effect with
20
respect to running volume. We further detect hints of sexually dimorphic serum responses in
21
oxygen transport and hormonal traits, and we also observe a tendency toward pronounced
22
modifications in magnesium status in professional endurance athletes. Thus, our results further
23
characterize blood biomarkers of exercise and metabolic health, particularly regarding dose-effect
24
relationships, and better inform personalized advice for training and performance.
25
26
27
28
29
Keywords: physical activity, exercise, blood biomarkers, running, generally healthy, mendelian
30
randomization
31
32
33
34
35
36
37
3
38
39
40
41
42
43
44
45
46
1
Introduction
47
Physical inactivity is one of the leading modifiable behavioral causes of death in the US [1].
48
Worldwide, physical inactivity is estimated to account for about 8.3% of premature mortality, an
49
effect size that is on the same order as smoking and obesity [2]. At the same time, the potent health
50
benefits of exercise have been proven time and time again, with results so consistent across a wide
51
variety of chronic diseases that some posit it can be considered a medical intervention [3-5].
52
However, since most investigators report the effects of exercise in either diseased populations or
53
athletes [6, 7], there exists a significant gap in knowledge as to the measurable effects of exercise
54
Formatted: Font: 18 pt
Formatted: No bullets or numbering
in the generally healthy population who exercise for the purpose of improving their healthspan,
55
which can be projected via established measures such as blood biomarkers [8-11].
56
It is well established that routine laboratory biomarkers are validated proxies of the state of an
57
individual’s overall metabolic health and other healthpan-related parameters [12]. A large body
58
of evidence supports the effectiveness of exercise in modifying blood biomarkers toward disease
59
mitigation in clinical cohorts as well as athletes, where the effect sizes may be larger [6, 13].
60
Indeed, it’s been shown that more favorable changes in response to exercise training occur usually
61
in those with more pronounced dyslipidemia [13]. In professional athletes, the sheer volume
62
and/or intensity of physical activity may drive large effects in various hematological, lipid,
63
immune, and endocrine variables [6]. Our aim is to help fill the gap in understanding of the effects
64
of exercise on blood biomarkers in the generally healthy, free-living population. Toward this end,
65
we endeavored to explore the effects of vigorous exercise such as running in apparently healthy,
66
mostly non-athletic cohort to better understand the landscape of blood biomarker modifications
67
expected in the individual who partakes in recreational physical activity for the purpose of
68
maintaining good health.
69
For this purpose, we leveraged the InsideTracker dataset that includes information on self-reported
70
exercise habits combined with blood biomarker and genomics data. We have previously reported
71
on the results of a longitudinal analysis on blood biomarker data from 1032 generally healthy
72
individuals who used our automated, web-based personalized nutrition and lifestyle platform [14].
73
For the purpose of this investigation, we focused on running as the exercise of choice as it is one
74
of the most common (purposeful) physical activity modalities practiced globally by generally
75
healthy individuals and would thus be relevant. Moreover, since this was a cross-sectional study
76
based on self-reported exercise habits, we attempted to increase our capacity to infer intervention
77
5
effects, as well as tease out potential confounders, by performing 2S-MR in large independent
78
cohorts.
79
2
MethodsMaterials and methods
80
2.1 Dataset
81
82
We conducted an observational analysis of data from InsideTracker users. InsideTracker is a
83
direct-to-consumer (DTC) company established in 2009 that markets and sells InsideTracker
84
(insidetracker.com), a personalized lifestyle recommendation platform. The platform provides
85
serum biomarker and genomics testing, and performs integrative analysis of these datasets,
86
combined with activity/sleep tracker data toward biomarker and healthspan optimization (of note,
87
at the time of this analysis, we did not have sufficient users with activity/sleep tracker data to
88
include this data stream in the current study). New users were continuously added to the
89
InsideTracker database from January 2011 to March 2022.
90
91
2.2 Recruitment of participants
92
Recruitment of participants aged between 18 and 65 and residing in North America was conducted
93
through company marketing and outreach. Participants were subscribing members to the
94
InsideTracker platform and provided informed consent to have their blood test data and self-
95
reported information used in an anonymized fashion for research purposes. Research was
96
Formatted: Font: 18 pt
Formatted: Font: 18 pt
Formatted: Font: 16 pt
Formatted: Font: 16 pt
Formatted: Font: 16 pt
conducted according to guidelines for observational research in tissue samples from human
97
subjects. Eligible participants completed a questionnaire that included age, ethnicity, sex, dietary
98
preferences, physical activity, and exposure to sunlightother variables. This study employed data
99
from 23,237 participants that met our analysis inclusion requirements, namely absence of any
100
chronic disease as determined by questionnaire and metabolic blood biomarkers within normal
101
clinical reference ranges. The platform is not a medical service and does not diagnose or treat
102
medical conditions, so medical history and medication use were not collected. The Institutional
103
Review Board (IRB) determine this work was not subject to a review based on category 4
104
exemption (“secondary research” with de-identified subjects).
105
2.3 Biomarker collection and analysis
106
Blood samples were collected and analyzed by Clinical Laboratory Improvement Amendments
107
(CLIA)–approved, third-party clinical labs (primarily Quest Diagnostics and LabCorp).
108
Participants were instructed to fast for 12 hours prior to the phlebotomy, with the exception of
109
water consumption. Results from the blood analysis were then uploaded to the platform via
110
electronic integration with the CLIA-approved lab. Participants chose a specific blood panel from
111
7 possible offerings, each comprising some subset of the biomarkers available. Due to the variation
112
in blood panels offered, the participant sample size per biomarker is not uniform.
113
2.3 Biomarker dataset preparation
114
115
Formatted: Font: 16 pt
Formatted: Font: 16 pt
Formatted: Font: 16 pt
7
In our raw dataset, occasional outlier values were observed that were deemed implausible (e.g.
116
fasting glucose < 65 mg/dL). To remove anomalous outliers in a systematic way, we used the
117
Interquartile Range (IQR) method of identifying outliers, removing data points which fell below
118
Q1 – 1.5 IQR or above Q3 + 1.5 IQR. The cohort was divided into five groups: professional
119
endurance runners (PRO), high volume amateur (>10 h/week, HVAM ), medium volume amateur
120
(3-10 h/week, MVAM), low volume amateur (<3 h/week, LVAM), and sedentary (SED).
121
2.4 Calculation of polygenic scores
122
The variants (SNPs) comprising the polygenic risk scores were derived from publicly available
123
GWAS summary statistics (https://www.ebi.ac.uk/gwas/). Scores were calculated across users by
124
summing the product of effect allele doses weighted by the beta coefficient for each SNP, as
125
reported in the GWAS summary statistics. Variant p-value thresholds were generally chosen based
126
on optimization of respective PGS-blood biomarker correlation in the entire InsideTracker cohort
127
with both blood and genomics datasets (~1000-1500 depending on the blood biomarker at the time
128
of analysis). Genotyping data was derived from a combination of a custom InsideTracker array
129
and third party arrays such as 23andMe and Ancestry. Not all variants for any particular PGS were
130
genotyped on every array; proxies for missing SNPs were extracted via the “LDlinkR” package
131
using the Utah Residents (CEPH) with Northern and Western European ancestry (CEU) population
132
(R2 > 0.8 cut-off). Only results PGSs for which there was sufficient biomarker-genotyping dataset
133
overlap were reported (note that none of the blood biomarker PGSs met this requirement).
134
2.5 Blood biomarker analysis with respect to running volume and
135
polygenic scores
136
Formatted: Font: Font color: Text/Background 1
Formatted: Font: 16 pt
Formatted: Font: 16 pt
To estimate significance of differences for blood biomarkers levels among exercise groups, we
137
performed 3-way analysis of variance (ANOVA) analysis adjusting for age, gender, and BMI
138
(type-II analysis-of-variance tables function ANOVA from ‘car’ R package, version 3.0-12).
139
When estimating the effort of reported training volume on biomarkers, we assigned numerical
140
values corresponding to 4 levels of running and performed ANOVA analysis with those levels
141
treating it as an independent variable. P-values were adjusted using the Benjamini & Hochberg
142
method [15]. P-values for interaction plots were calculated with ANOVA including interaction
143
between exercise group and polygenic scores category. When comparing runners (PRO and
144
HVAM combined) versus sedentary individuals, we used propensity score matching method to
145
account for existing covariates (age and gender): we identified 745 sedentary individuals with
146
similar to runners’ age distributions among both males and females. We used ‘MatchIt’ R package
147
(version 4.3.3) implementing nearest neighbor method for matching [16].
148
2.6 Mendelian randomization
149
We attempted to add insight around the causality of exercise vs. BMI differences with respect to
150
serum marker improvement by performing MR analyses on a subset of biomarker observations
151
where BMI featured as a strong covariate and was thus used as the IV in the 2S-MR. Thus, our
152
hypothesis here was that BMI differences were the primary (causal) driver behind the improvement
153
behind some biomarkers. MR uses genetic variants as modifiable exposure (risk factor) proxies
154
to evaluate causal relationships in observational data while reducing the effects of confounders
155
and reverse causation (Figure 1SS1 Fig). These SNPs are used as instrumental variables and must
156
meet 3 basic assumptions: (1) they must be robustly associated with the exposure; (2) they must
157
exert their effect on outcome via the exposure, and (3) there must be no unmeasured confounders
158
Formatted: Font: 16 pt
9
of the associations between the genetic variants and outcome (e.g. horizontal pleiotropy) [17].
159
Importantly, SNPs are proper randomization instruments because they are determined at birth and
160
thus serve as proxies of long-term exposures and cannot, in general, be modified by the
161
environment. If the 3 above mentioned assumptions hold, MR-estimate effects of exposure on
162
outcomes are not likely to be significantly affected by reverse causation or confounding. In the
163
2S-MR performed here, where GWAS summary statistics are used for both exposure and outcome
164
from independent cohorts, reverse causation and horizontal pleiotropy can readily be assessed, and
165
weak instrument bias and the likelihood of false positive findings are minimized as a result of the
166
much larger samples sizes [17]. Indeed, the bias in the 2S-MR using non-overlapping datasets as
167
performed here is towards the null [17]. Furthermore, to maintain the SNP-exposure associations
168
and linkage disequilibrium (LD) patterns in the non-overlapping populations we used GWAS
169
datasets from the MR-Base platform that were derived from ancestrally similar populations
170
(“ukb”: analysis of UK Biobank phenotypes, and “ieu”: GWAS summary datasets generated by
171
many different European consortia). To perform the analysis we used the R package
172
“TwoSampleMR” that combines the effects sizes of instruments on exposures with those on
173
outcomes via a meta-analysis. We used “TwoSampleMR” package functions for allele
174
harmonization between exposure and outcome datasets, proxy variant substitution when SNPs
175
from exposure were not genotyped in the outcome data (Rsq>0.8 using the 1000G EUR reference
176
data integrated into MR-Base), and clumping to prune instrument SNPs for LD (the R script used
177
for MR analyses is available upon request). We used 5 different MR methods that were included
178
as part of the “TwoSampleMR” package to control for bias inherent to any one technique [18].
179
For example, the multiplicative random effects inverse variance-weighted (IVW) method is a
180
weighted regression of instrument-outcome effects on instrument-exposure effects with the
181
intercept is set to zero. This method generates a causal estimate of the exposure trait on outcome
182
traits by regressing the, for example, SNP-BMI trait association on the SNP-biomarker measure
183
association, weighted by the inverse of the SNP-biomarker measure association, and constraining
184
the intercept of this regression to zero. This constraint can result in unbalanced horizontal
185
pleiotropy whereby the instruments influence the outcome through causal pathways distinct from
186
that through the exposure (thus violating the second above-mentioned assumption). Such
187
unbalanced horizontal pleiotropy distorts the association between the exposure and the outcome,
188
and the effect estimate from the IVW method can be exaggerated or attenuated. However,
189
unbalanced horizontal pleiotropy can be readily assessed by the MR Egger method (via the MR
190
Egger intercept), which provides a valid MR causal estimate that is adjusted for the presence of
191
such directional pleiotropy, albeit at the cost of statistical efficiency. Finally, to ascertain the
192
directionality of the various causal relationships examined, we also performed each MR analysis
193
in reverse where possible.
194
3
Results
195
Study population characteristics
196
Table 1 shows the demographic characteristics of the study population. We observed a
197
significant trend toward younger individuals reporting higher running volume, with more than
198
75% of the professional (PRO) group falling between the ages of 18 and 35 (Table 1SS1 Table).
199
Significant differences were also observed in the distribution of males and females within study
200
groups (Table 1). Moreover, higher running volume associated with significantly lower body mass
201
Formatted: Font: 18 pt
Formatted: Font: 18 pt
Formatted: Font: 16 pt, Bold, Not Italic
11
index (BMI). Thus, moving forward, combined comparisons of blood biomarkers as they relate
202
to running volume were adjusted for age, gender, and BMI.
203
204
Table 1. Study population demographics
205
Group
N
Female, %
Age, yrs
Body mass index, kg/m2
PRO
82
53.7%
33.68
20.15 ± 6.02
HVAM
1103
52.9%
39.48
22.57 ± 9.97
MVAM
6747
54.2%
41.49
23.35 ± 9.76
LVAM
10877
34.2%
41.16
24.72 ± 9.70
SED
4428
48.9%
44.25
27.83 ± 10.70
PRO = Professional, HVAM = high volume amateur (>10 h/week ), MVAM = medium volume
206
amateur (3-10 h/week), LVAM = low volume amateur (<3 h/week), SED = sedentary
207
208
Endurance exercise exhibits a modest association with clusters of
209
blood biomarker features
210
In order to begin to understand the most important variables that may associate with endurance
211
exercise in the form of running, we performed a principal component analysis (PCA), sub-dividing
212
the male and female cohorts into two most divergent groups in terms of exercise volume:
213
PRO/high volume amateur (HVAM) and sedentary (SED) groups. Using propensity matching,
214
PRO and amateur athletes who reported running >10h per week were combined into the PRO-
215
HVAM group to balance out the sample size between the exercising and non-exercising groups.
216
Formatted: Font: (Default) Times New Roman, 12 pt
Formatted: Font: (Default) Times New Roman, 12 pt
Formatted: Font: (Default) Times New Roman, 12 pt
Formatted: Font: (Default) Times New Roman
Formatted: Font: (Default) Times New Roman
Formatted: Font: (Default) Times New Roman
Formatted: Font: (Default) Times New Roman
Formatted: Font: (Default) Times New Roman
Formatted: Font: (Default) Times New Roman
Formatted: Font: (Default) Times New Roman
Formatted: Font: (Default) Times New Roman
Formatted: Font: (Default) Times New Roman
Formatted: Font: (Default) Times New Roman
Formatted: Font: (Default) Times New Roman
Formatted: Font: (Default) Times New Roman, 12 pt
Formatted: Font: 16 pt, Bold, Not Italic
Using this approach, we did not observe a significant separation between these groups (data not
217
shown). However, dividing this dataset further into males and femalesThis approach yielded a
218
modest degree of separation, with hematological, inflammation, and lipid features, as well as BMI
219
explaining some of the variance (FigureFig 1 A through D). We hypothesized that there may more
220
subtle relationships between running volume and the blood biomarker features that contributed to
221
distinguishing the endurance exercise and sedentary groups, thus we next performed ANOVA
222
analyses stratified by running volume as categorized in Table 1.
223
Fig 1. Principal component analysis and variables plots of PRO-HVAM runners and sedentary
224
user blood biomarkers. Females, (A) and (B); males (C) and (D). PRO-HVAM = combined
225
professional and high-volume amateur. Alb = albumin, ALT = alanine transaminase, AST =
226
aspartate aminotransferase, B12 = vitamin B12, Ca = calcium, Chol = total cholesterol, CK =
227
creatine kinase, Cor = cortisol, FE = iron, EOS_PCT = eosinophil percentage, Fer = ferritin, Fol =
228
folate, FT = free testosterone, GGT = gamma-glutamyl transferase, Glu = glucose, Hb =
229
hemoglobin, HCT = hematocrit, HDL = high density lipoprotein, HbA1c = glycated hemoglobin,
230
hsCRP = high-sensitivity C-reactive protein, LDL = low density lipoprotein, LYMPS_PCT =
231
lymphocyte percentage, MCH = mean cell hemoglobin, Mg = magnesium, MONOS_PCT =
232
monocytes percentage, MPV = mean platelet volume, Na = sodium, RBC = red blood cells,
233
RBC_Mg = red blood cell magnesium, RDW = red blood cell distribution width, SHBG = sex
234
hormone binding globulin, Tg = triglycerides, TIBC = total iron binding capacity, WBC = white
235
blood cells
236
Significant trends in glycemic, hematological, blood lipid, and
237
inflammatory serum traits with increasing running volumes
238
Formatted: Font: 12 pt
Formatted: Font: 16 pt, Bold, Not Italic
Formatted: Font: 16 pt, Bold
Formatted: Font: 16 pt, Bold, Not Italic
13
Weighted ANOVA analyses adjusted for age, gender, and BMI showed significant differences
239
among groups for multiple blood biomarkers (Table 2 and 2SS2 , FigureFigs 2 and 3). We
240
observed a trend toward lower HbA1c, hsCRP, RDW, WBC, ferritin, gamma-glutamyl transferase
241
(GGT), and LDL. HDL, hemoglobin (Hb), transferrin saturation (TS), alanine aminotransferase
242
(ALT), aspartate aminotransferase (AST), vitamin B12, folate, 25-hydroxy vitamin D, and creatine
243
kinase (CK) tended to be higher with increasing reported training volume, particularly in PRO
244
runners (Tables 2 and 2SS2 , FigureFigs 2 and 2SS2 , FigureFig 3). Hct and Hb were higher only
245
in PRO males, whereas increased running volume associated with upward trend in these
246
biomarkers in females (FigureFig 3 A and B). Increased running volume was associated with
247
markedly lower Fer in males, whereas female runners did not exhibit varying levels, and SED
248
females showed increased levels (FigureFig 3 C). The low ferritin observed in male and female
249
runners was not clinically significant. ALT positively associated with running volume in females
250
only (Figure 2SS2 Fig). Serum and RBC magnesium (Mg) were both significantly lower in PRO
251
runners relative to all other groups (Table 2 and FigureFig 3 D and E). Increasing levels of
252
endurance exercise also appeared to be associated with higher sex-hormone binding globulin
253
(SHBG), particularly in PRO male runners (FigureFig 3 F).
254
255
Table 2. Blood biomarkers significantly different among sedentary individuals and those
256
who partake in running for exercise to various degrees
257
BIOMARKER ANOVA P-VALUE TREND P-VALUE LOWEST MEAN
HIGHEST MEAN
ALB
<1e-16
<0.001
MVAM
PRO
ALT
<1e-16
<1e-16
SED
PRO
Formatted: Font: (Default) Times New Roman, 12 pt
Formatted: Line spacing: single, Don't suppress line
numbers
Formatted: Font: (Default) Times New Roman, 12 pt, Bold
Formatted: Font: (Default) Times New Roman, 12 pt, Bold
Formatted: Font: (Default) Times New Roman, 12 pt, Bold
Formatted: Font: (Default) Times New Roman, 12 pt, Bold
Formatted: Font: (Default) Times New Roman, 12 pt, Bold
Formatted: Font: (Default) Times New Roman, 12 pt, Bold
Formatted: Font: (Default) Times New Roman, 12 pt, Bold
Formatted: Font: (Default) Times New Roman, 12 pt, Bold
Formatted: Font: (Default) Times New Roman, 12 pt, Bold
Formatted: Font: (Default) Times New Roman, 12 pt, Bold
Formatted: Font: (Default) Times New Roman, 12 pt, Bold
Formatted: Font: (Default) Times New Roman, 12 pt, Bold
Formatted: Font: (Default) Times New Roman, 12 pt, Bold
Formatted: Font: (Default) Times New Roman, 12 pt, Bold
Formatted: Font: Bold, Font color: Black
AST
<1e-16
<0.001
SED
PRO
B12
<0.001
<0.001
SED
PRO
CHOL
<0.001
0.005
PRO
SED
CK
<1e-16
<1e-16
SED
PRO
COR
<0.001
0.675
SED
PRO
FE
<0.001
0.119
SED
PRO
FER
<1e-16
<1e-16
MVAM
SED
FOL
<1e-16
<0.001
SED
PRO
FT
<0.001
0.013
SED
PRO
GGT
<1e-16
<0.001
PRO
SED
GLU
0.087
0.184
PRO
SED
HB
0.002
<0.001
MVAM
PRO
HCT
0.053
0.055
MVAM
PRO
HDL
<1e-16
<0.001
SED
PRO
HBA1C
<0.001
0.010
PRO
SED
HSCRP
<0.001
0.176
PRO
SED
LDL
<0.001
0.006
PRO
SED
MG
<0.001
0.276
PRO
SED
MPV
0.058
0.089
SED
HVAM
NA
<1e-16
0.622
HVAM
SED
RBC_MG
<0.001
0.773
PRO
SED
15
RDW
<1e-16
0.002
PRO
SED
SHBG
<1e-16
0.004
SED
PRO
TG
<1e-16
<1e-16
PRO
SED
WBC
<1e-16
<1e-16
PRO
SED
258
Fig 2. Blood biomarkers associated with running: Inflammation proxies, (A) hsCRP = high-
sensitivity C-reactive protein and (B) WBC = white blood cells; blood lipids, (C) HDL = high
density lipoprotein (D) LDL = low density lipoprotein, and (E) Tg = triglycerides; glycemia
proxies, (F) Glu = glucose and (G) HgbA1c = glycated hemoglobin, and (H) Cor =cortisol
Fig 3. Blood biomarkers associated with running: (A and B) Hb (hemoglobin) and Hct
(hematocrit) increase with increasing running volume, (C) Fer (ferritin) is reduced with
increasing running volume, (D and E) Serum and RBC Mg (red blood cell magnesium) are
reduced in professional runners, and (F) SHBG (sex hormone binding globulin) levels increase
with increasing running volume in males
259
Endurance exercise correlates with lower BMI across categories of
260
genetic risk
261
Using publicly available GWAS summary statistics, we constructed blood biomarker polygenic
262
risk scores (PGSs) to explore potential genetic risk-mitigating effects of endurance exercise. Since
263
Formatted: Line spacing: Double
Formatted: Font: 12 pt
Formatted: Font: 12 pt
Formatted: Font: 16 pt, Bold, Not Italic
only a subset of the individuals in our cohort were genotyped, we aggregated the groups into 2
264
categories—PRO-HVAM and sedentary—to increase statistical power. This across-group sample
265
size increase generally did not sufficiently power the ANOVA analysis to detect statistically
266
significant trends (data not shown), though the BMI polygenic risk was suggestively mitigated for
267
both males and female PRO-HVAM runners across categories of genetic risk (FigureFig 4 B).
268
Fig 4. BMI significantly varied among running groups (A) with some suggestive effects on BMI
PGS modification (total number for observations (N) for T1, T2, and T3 were 87, 84, and 100,
respectively) (B) T1, T2, and T3 = 1st, and 2nd and 3rd tertials of the polygenic score distribution
269
Increased running volume is associated with lower BMI which may
270
drive biomarker changes
271
We observed a significant downward trend in the BMI with increased running volume for both
272
males and females, and, although some of the biomarker differences between sedentary and
273
exercising individuals remained significant after adjustment for BMI, their significance was
274
attenuated (FigureFig 4 A, p-value attenuation data not shown). Thus, we hypothesized that BMI
275
may be driving a significant portion of the observed variance in some of the biomarkers across the
276
groups. Thus, to explore causal relationships between weight and biomarker changes, we
277
performed 2S-MR with BMI-associated single-nucleotide polymorphisms (SNPs) as the
278
instrumental variables (IVs) for a subset of the healthspan-related biomarkers where BMI
279
explained a relatively large portion of the variance in our analysis. In general, these blood
280
biomarkers associated with inflammation (hsCRP and RDW), lipid metabolism (Tg and HDL),
281
Formatted: Line spacing: Double
Formatted: Font: 12 pt
Formatted: Font: 12 pt
Formatted: Font: 16 pt, Bold, Not Italic
17
glycemic control (HbA1c and Glu), as well as Alb and SHBG. We used GWAS summary statistics
282
and found that most of these BMI-blood biomarker relationships examined directionally aligned
283
with our study (except for LDL), and some were indicative of causal relationships in the BMI-
284
biomarker direction even after considering directional pleiotropy (Table 3SS3 Table). We
285
entertained the possibility of reverse causality and thus repeated the 2S-MR using each of the
286
biomarker levels as the exposure and BMI as the outcome, and the results were generally not
287
significant (except for WBC – see Table 4SS4 Table). Of note, to estimate the direct causal effects
288
of running on blood parameters, we attempted to find an instrumental variable for to approximate
289
running as the exposure from publicly available GWAS summary statistics. Toward this end, we
290
found that increasing levels of vigorous physical activity did associate with lower hsCRP, HbA1C,
291
higher HDL, and possibly higher SHBG (although the explained variance (R2) in this exposure
292
was just 0.001009, the F statistic was 37.7, thus meeting the criteria of F > 10 for minimizing weak
293
instrument bias) (FigureFigs 5 and 3SS3; Table 5SS5 Table).
294
Fig 5. Two-sample Mendelian randomization shows that increasing levels of vigorous physical
activity such as running is associated with improvement of (A) hsCRP = high-sensitivity C-
reactive protein, (B) HDL = high density lipoprotein, and (C) HbA1c = glycated hemoglobin
levels
295
Vigorous physical activity associates with healthier behaviors
296
We hypothesized that those who exercise regularly may also partake in other healthful lifestyle
297
habits that may be contributing to more optimal blood biomarker signatures of wellness. However,
298
Formatted: Line spacing: Double
Formatted: Font: 12 pt
Formatted: Font: 12 pt
Formatted: Font: 16 pt, Bold, Not Italic
Formatted: Font: 16 pt, Bold
our dataset did not allow for systematic accounting of other lifestyle habits across all running
299
groups. Thus, we again leveraged the potential of the 2S-MR approach to inform potential
300
confounding associations between modifiable exposures and found that vigorous physical activity
301
such as running is at least suggestively associated with several behaviors associated with improved
302
health (Figure 4SS4 Fig). Our analysis showed that those who participate in increasing levels of
303
vigorous physical activity may be less likely to eat processed meat (IVW p = 0.0000013), sweets
304
(IVW p = 0.32), and nap during the day (IVW p = 0.13), while increasing their intake of oily fish
305
(IVW p = 0.029), salad/raw vegetable intake (IVW p = 0.00016), and fresh fruit (IVW p = 0.0027)
306
(Table 6SS6 Table). Furthermore, following our assessment of reverse causality, we found
307
evidence for the bidirectionality in the causal relationship between vigorous activity and napping
308
during the day and salad/raw vegetable intake, perhaps suggesting some degree of confounding
309
due to population stratification (Table 7SS7 Table). The suggestive positive effect of fresh fruit
310
and processed meat intake on vigorous physical activity appeared to violate MR assumption (3)
311
(Figure 1SS1 Fig) (horizontal pleiotropy p-values 0.051 and 0.17, respectively – Figure 5SS5 Fig).
312
4
Discussion
313
In this report, we describe the variance in wellness-related blood biomarkers among self-reported
314
recreational runners, PRO runners, and individuals who do not report any exercise. Overall, we
315
find that 1) recreational running as an exercise appears to be an effective intervention toward
316
modifying several biomarkers indicative of improved metabolic health, 2) an apparent dose-
317
response relationship between running volume and BMI may itself be responsible for a proportion
318
of the apparent metabolic benefits, and 3) both PRO-level status and gender appear to associate
319
Formatted: Font: 18 pt
19
with heterogeneous physiological responses, particularly in iron and magnesium metabolism, as
320
well as some hormonal traits.
321
4.1 Self-reported running improves glycemia and lipidemia
322
We did not observe distinct clusters corresponding to self-reported high-volume/PRO runners and
323
the sedentary upon dimension reduction. This is, perhaps, not unexpected due, in part, to the self-
324
selected healthspan-oriented nature of our cohort, where even the sedentary subset of individuals
325
tends to exhibit blood biomarker levels in the normal clinical reference ranges. Furthermore, the
326
measurement of running volume via self-report may be vulnerable to overestimation, which may
327
have contributed to the blending of sedentary and exercise groups with respect to the serum
328
markers measured, resulting in only marginal separation between the groups [19, 20]. However,
329
we did observe significant individual blood biomarker variance with respect to reported running
330
volumes when the dataset was subjected to ANOVA, even after adjustment for age, sex, and BMI.
331
From among glycemic control blood biomarkers, we were able to detect a relatively small exercise
332
effect in both fasting glucose and HbA1c in this generally healthy cohort, where the average
333
measures of glycemia were below the prediabetic thresholds in even the sedentary subset of the
334
cohort. Larger exercise intervention effects on metabolic biomarkers may be expected in cohorts
335
that include individuals with more clinically significant baseline values [21].
336
Similarly, blood lipids improved with higher self-reported running volume, and this result has been
337
reported before in multiple controlled endurance exercise trials [22]. The literature indicates that
338
HDL and Tg are two exercise-modifiable blood lipid biomarkers, with HDL being the most widely
339
reported to be modified by aerobic exercise [23, 24]. Although the mechanism behind this is not
340
Formatted: Font: 16 pt
entirely clear, it likely involves the modification of lecithin acyltransferase and lipoprotein lipase
341
activities following exercise training [25]. We observed a similar trend in our blood biomarker
342
analysis, with HDL exhibiting an upward trend with increasing reported running volume. While
343
we also found Tg and LDL to decrease with increasing exercise volume, these trends were less
344
pronounced. Reports generally suggest that, in order to reduce LDL more consistently, the
345
intensity of aerobic exercise must be high enough [23]. In the case of Tg, baseline levels may have
346
a significant impact on the exercise intervention effect, with individuals exhibiting higher baselines
347
showing greater improvements [13].
348
Importantly, these results suggests that exercise has a significant effect on glycemic control and
349
blood lipids even in the self-selected, already healthy individuals who are proactive about
350
preventing cardiometabolic disease.
351
4.2 Self-reported running and serum proxies of systemic
352
inflammation
353
Chronic low-grade inflammation is one of the major risk factors for compromised cardiovascular
354
health and metabolic syndrome (MetS). While there is no shortage of inflammation-reducing
355
intervention studies on CVD patients with clinically high levels of metabolic inflammation, there
356
is less emphasis on modifiable lifestyle factors that can help stave off CVD and extend healthspan
357
in the generally healthy individual. Indeed, considering the pathological cardiovascular processes
358
begin shortly after birth, prevention in asymptomatic individuals may be a more appropriate
359
strategy toward decreasing the burden of CVD on the healthcare system [26].
360
Formatted: Font: 16 pt
21
Toward this end, increasing self-reported running volume appeared to associate with improved
361
markers of inflammation, as shown by the lower levels of hsCRP, WBC, as well as ferritin. Of
362
note, while the acute-phase protein, ferritin, is often used in the differential diagnosis of iron
363
deficiency anemia, the biomarker’s specificity appears to depend on the inflammatory state of the
364
individual, as it associates with hsCRP and inflammation more than iron stores, particularly in
365
those with higher BMI [27]. Although serum ferritin and iron is reported to be lower in male and
366
female elite athletes [28], the observed overall negative association of ferritin with increased
367
running volume in our cohort may be an indication of lower levels of inflammation rather than
368
compromised iron stores, particularly since the average ferritin level across all groups was above
369
the clinical iron deficiency thresholds. Moreover, increased levels of ferritin have been associated
370
with insulin resistance and lower levels of adiponectin in the general population, both indicators
371
of increased systemic inflammation [29]. Here, exercising groups with lower levels of ferritin also
372
exhibited glycemic and blood lipid traits indicative of improved metabolic states, further
373
supporting ferritin’s role as an inflammation proxy. Finally, Hb, TS and iron tended to be higher
374
in those who run for exercise compared to the SED group (with the TIBC lower), again suggesting
375
that runners, including the PRO group, were iron-sufficient in this cohort.
376
4.3 PRO athletes endurance runners exhibit distinct biomarker
377
signatures
378
PRO athletes exhibited lower serum and RBC Mg, which may be indication of the often-reported
379
endurance athlete hypomagnesaemia [30]. While the serum Mg was still within normal clinical
380
reference range for both PRO female and male athletes, RBC Mg, a more sensitive biomarker of
381
Mg status [31], was borderline low in female PRO athletes and might suggest suboptimal dietary
382
Formatted: Font: 16 pt
intakes and/or much higher volume of running training compared to the other running groups (i.e.
383
>>10h /week). Indeed, this group also had elevated baseline CK and AST, which suggests a much
384
higher training intensity and/or volume. Moreover, PRO level athletes had adequate iron status
385
and serum B12 and folate in the upper quartile of the normal reference range, suggesting that these
386
athletes’ general nutrition status may have been adequate. These observations suggest that elite
387
endurance runners may need to pay particular attention to their magnesium status.
388
Further, we observed higher levels of SHBG in PRO male runners, a biomarker whose levels
389
positively correlate with various indexes of insulin sensitivity [32]. However, since the average
390
SHBG levels in the SED group were not clinically low in both sexes, the observed increase in
391
SHBG levels induced by running in males may be a catabolic response, as cortisol levels in this
392
group were also higher. Indeed, Popovic et al have shown that endurance exercise may increase
393
SHBG, cortisol, and total testosterone levels at the expense of free testosterone levels [33]. This
394
could perhaps in part be explained by higher exercise-induced adiponectin levels, which have been
395
shown to increase SHBG via cAMP kinase (AMPK) activation [34]. However, since our data is
396
observational, we cannot rule out overall energy balance as a significant contributor to SHBG
397
levels. For example, caloric restriction (CR) has been shown to result in higher SHBG and cortisol
398
levels [32].
399
Finally, regarding the abovementioned PRO group elevated AST and CK biomarkers, evidence
400
suggests that normal reference ranges in both CK and AST in well-recovered athletes should be
401
adjusted up, as training and competition have a profound, non-pathological, impact on the activity
402
of these enzymes [35, 36]. Indeed, the recommendation appears to be not to use reference intervals
403
derived from the general population with hard-training (particularly competitive) athletes [36].
404
23
4.4 Effect of BMI on blood biomarkers
405
Since the current study is a cross-sectional analysis of self-reported running, we could not rule out
406
the possibility that factors other than exercise were the driving force behind the observed
407
biomarker variance among the groups examined. While These factors, such as diet, sleep, and/or
408
medication medications use were not readily ascertained in this free-living cohort at the time of
409
this study, but BMI was readily available to evaluate this biomarker’s potential relative
410
contribution to the observed mean biomarker differences among self-reported runner groups.
411
Multiple studies have attempted to uncouple the effects of exercise and BMI reduction on blood
412
biomarker outcomes, with mixed results [37]. For example, it is relatively well-known that acute
413
bouts of exercise improve glucose metabolism, but long-term effects are less well described [38].
414
Indeed, whether exercise without significant weight-loss is effective toward preventing metabolic
415
disease (and the associated blood biomarker changes) is inconclusive [39-41]. From the literature,
416
it appears that, for endurance exercise to have significant effect on most blood biomarkers, the
417
volume of exercise needs to be very high, and this typically results in significant reduction in
418
weight. Thus, in practice, it is difficult to demonstrably uncouple the effects of significant exercise
419
and the associated weight-loss, and the results may depend on the blood biomarker in question.
420
Indeed, there is evidence that exercise without weight-loss does improve markers of insulin
421
sensitivity but not chronic inflammation, with the latter apparently requiring a reduction in
422
adiposity in the general population [39-41].
423
In our study of apparently healthy individuals, we observed a downward trend in BMI with
424
increasing self-reported running volume, and, although this study was not longitudinal and we are
425
thus unable to claim weight-loss, our 2S-MR analysis using BMI as the exposure nonetheless
426
Formatted: Font: 16 pt
suggests this biomarker to be responsible for a significant proportion of the modification of some
427
blood biomarkers.
428
4.5.1 Serum markers of systemic inflammation
429
Through our 2S-MR analyses, we show that BMI is causally associated with markers of systemic
430
inflammation, including RDW, folate, and hsCRP [27, 42, 43]. Similar analyses have reported
431
that genetic variants that associate with higher BMI were associated with higher CRP levels, but
432
not the other way around [44]. The prevailing mechanism proposed to explain this relationship
433
appears to be the pathological nature of overweight/obesity-driven adipose tissue that results in
434
secretion of proinflammatory cytokines such as IL-6 and TNFa, which then stimulate an acute
435
hepatic response, resulting in increased hsCRP levels (among other effects) [45]. Thus, our 2S-
436
MR analyses and those of others [44] would indicate that the primary factor behind the lower
437
systemic inflammation in our cohort may be the exercise-associated lower BMI and not running
438
exercise per se, though the lower hsCRP in runners remained significant after adjustment for BMI
439
in our analysis.
440
Indeed, although a major driver behind reduced systemic inflammation may be a reduction in BMI
441
in the general population, additive effects of other lifestyle factors such as exercise cannot be
442
excluded. For example, a large body of cross-sectional investigations does indicate that physically
443
active individuals exhibit CRP levels that are 19-35% lower than less active individuals, even
444
when adjusted for BMI as was the case in the current analysis [41]. Further, it’s been reported that
445
physical activity at a frequency of as little as 1 day per week is associated with lower CRP in
446
individuals who are otherwise sedentary, while more frequent exercise further reduces
447
inflammation [41].
448
Formatted: Font: 14 pt
25
Significantly, our entire cohort of self-selected apparently healthy individuals did not exhibit
449
clinically high hsCRP, with average BMI also below the overweight thresholds. Because all
450
subjects were voluntarily participating in a personalized wellness platform intended to optimize
451
blood biomarkers that included hsCRP, it is possible that some individuals from across the study
452
groups (both running and sedentary) in our cohort partook in some form of inflammation-reducing
453
dietary and/or lifestyle-based intervention. Thus, that we detected a significant difference in
454
hsCRP between exercising and non-exercising individuals in this self-selected already generally
455
healthy cohort may be suggestive of the potential for additional preventative effect of scheduled
456
physical activity on low-grade systemic inflammation in the generally healthy individual.
457
4.5.2 Blood lipids
458
Controlled studies that tightly track exercise and the associated adiposity reduction have reported
459
that body fat reduction (and not improvement in fitness as measured via VO2max) is a predictor of
460
HDL, LDL, and Tg [46]. Similarly, though BMI is an imperfect measure of adiposity, our 2S-MR
461
analysis suggests that this biomarker is causally associated with improved levels of HDL and Tg,
462
though not LDL. This latter finding replicates a report by Hu et al. who, using the Global Lipids
463
Genetics Consortium GWAS summary statistics, applied a network MR approach that revealed
464
causal associations between BMI and blood lipids, where Tg and HDL, but not LDL, were found
465
to trend toward unhealthy levels with increasing adiposity [47]. On the other hand, others
466
implemented a robust BMI genetic risk score and demonstrated a causal association of adiposity
467
with peripheral artery disease and a multiple linear regression showed a strong association with
468
HDL, TC, and LDL, among other metabolic parameters [48]. In our cohort, given the lack of
469
evidence for a causal BMI-LDL association and the overall healthy levels of BMI, the observed a
470
Formatted: Font: 14 pt
significant improvement in LDL may be a result of marked running intensity and/or volume,
471
possibly combined with the aforementioned additional wellness program intervention variables.
472
4.5.3 Hormonal traits
473
As described above, we observed a trend toward increased plasma cortisol and SHBG in runners,
474
particularly PRO level athletes. The effects on cortisol are consistent with a report by Houmanrd
475
et al, who found male distance runners to exhibit higher levels of baseline cortisol [49]. With
476
respect to the effects of BMI on baseline cortisol levels, this observation is generally supported by
477
our 2S-MR analyses with evidence for a consistent effect of increased cortisol with decreasing
478
BMI. However, this association was suggestive at best, indicating that the higher levels of cortisol
479
exhibited in the PRO runners with significant lower adiposity are not likely to be solely explained
480
by their lower BMI. Indeed, the relationship between BMI and cortisol appears to be complex,
481
with some reports suggesting a U-shaped relationship, where the glucocorticoid’s levels associate
482
negatively up to about a BMI of 30 kg/m2, then exhibiting a positive correlation into obesity
483
phenotypes [50]. MR statistical models generally do not account for such non-linearity and would
484
require a more granular demographical treatment, which is not possible using only GWAS
485
summary statistics data in the context of 2S-MR [17, 51].
486
4.6 Behavioral traits associated with increase physical activity
487
The combination of the body of the literature that describes the effects of endurance training on
488
blood biomarkers, and our own analysis that included markers such as CK and AST, makes us
489
cautiously assured that most of the abovementioned blood biomarker signatures are indeed a result
490
of the interplay between self-reported running and the associated lower BMI. However, as this is
491
Formatted: Font: 14 pt
Formatted: Font: 16 pt
27
a self-report-based analysis and we were unable to track other subject behaviors in this free-living
492
cohort, we acknowledge that multiple behaviors that associate with exercise may be influencing
493
our results.
494
Toward this end, our exploratory 2S-MR analyses revealed potentially causal relationships
495
between vigorous exercise and multiple dietary habits that have been shown to improve the
496
biomarkers we examined. Indeed, diets that avoid processed meat and sweets while providing
497
ample amounts of fresh fruits, as well as oily fish have been shown to be anti-inflammatory, and
498
improve glycemic control and dyslipidemia [52, 53]. That physically active individuals are also
499
more likely to make healthier dietary choices adds insight to the potential confounders in ours and
500
others’ observational analyses, and this similar associations have previously been reported [54-
501
56]. For example, using a calculated healthy eating motivation score, Naughton et al. showed that
502
those who partake in more than 2 hours of vigorous physical activity are almost twice as likely to
503
be motivated to eat healthy [56]. Indeed, upon closer examination, the genetic instruments used
504
to approximate vigorous physical activity as the exposure in this work included variants in the
505
genes DPY19L1, CADM2, CTBP2, EXOC4, and FOXO3 [57]. Of these, CADM2 encodes proteins
506
that are involved in neurotransmission in brain regions well known for their involvement in
507
executive function, including motivation, impulse regulation and self-control [58]. Moreover,
508
variants within this locus have been associated with obesity-related traits [59]. Thus, it is likely
509
that the improved metabolic outcomes seen here with our self-reported runners are a composite
510
result of both these individuals exercise and dietary habits. Importantly, the above suggests that a
511
holistic wellness lifestyle approach is in practice the most likely to be most effective toward
512
preventing cardiometabolic disease. Nonetheless, the focus of this work – exercise in the form of
513
running – is known to significantly improve cardiorespiratory fitness (CRF), which has been
514
shown to be an independent predictor of CVD risk and total mortality, outcomes that indeed
515
correlate with dysregulated levels in many of the blood biomarkers examined in this work [7].
516
4.7 Study limitations
517
This study is based on self-reported running and thus has several limitations. First, it is generally
518
known that subjects tend to overestimate their commitment to exercise when self-reporting,
519
although in our cohort is a self-selected health-oriented population that is possibly less likely to
520
over-report their running volume. First, it is generally known that subjects tend to overestimate
521
their commitment to exercise when self-reporting, although in our cohort is a self-selected health-
522
oriented population that is possibly less likely to over-report their running volume. Furthermore,
523
although the robust increasing trend in baselines for muscle damage biomarkers (CK, AST) that
524
have been shown to be associated with participation in sports and exercise provides indirect
525
evidence that the running groups were indeed participating in increasing volumes of strenuous
526
physical activity, we cannot confirm whether the reported running was performed overground or
527
on a treadmill, which may result in some heterogeneity in physiological responses , nor can we
528
ascertain the actual training volume of PRO-level runners. We also cannot exclude the possibility
529
that the running groups also participated in other forms of exercise (such as strength training) or
530
partook in other wellness program interventions that may have influenced their blood biomarkers
531
and/or BMI via lean muscle accretion. Toward this end, we have attempted to shed light on
532
potential behavioral covariates related to vigorous physical activity via 2S-MR. Finally, while
533
this cohort is generally healthy, we cannot exclude the potential for unmeasured confounders such
534
as medications, nutritional supplements, and unreported health conditions.
535
Formatted: Font: 16 pt
29
2S- MR enables the assessment of causal relationships between modifiable traits and is less prone
536
to the so-called “winner’s curse” that more readily affects one-sample MR analyses [17, 51].
537
Because 2S-MR uses GWAS summary statistics for both exposure and outcome, it is possible to
538
increase statistical power because of the increased sample sizes. However, horizontal pleiotropy
539
is still a concern that can skew the results. Currently, there is no gold standard MR analysis
540
method, thus we used different techniques (IVW, MR-Egger, and median-based estimations – all
541
of which are based on different assumptions and thus biases) to evaluate the consistency among
542
these estimators and only reported associations as ‘causal’ if there was cross-model consistency.
543
Nonetheless, an exposure such as BMI is a complex trait that is composed of multiple sub-
544
phenotypes (such as years of education) that could be driving the causal associations.
545
546
5
Conclusions
547
Running is one of the most common forms of vigorous exercise practiced globally, thus making it
548
a compelling target of research studies toward understanding its applicability in chronic disease
549
prevention. Our cross-sectional study offers insight into the biomarker signatures of self-reported
550
running in generally healthy individuals that suggest improved insulin sensitivity, blood lipid
551
metabolism, and systemic inflammation. Furthermore, using 2S-MR in independent datasets we
552
provide additional evidence that some biomarkers are readily modified BMI alone, while others
553
appear
to
respond
to
the
combination
of
varying
exercise
and
BM
554
I. Our additional bi-directional 2S-MR analyses toward understanding the causal relationships
555
between partaking in vigorous physical activity and other healthy behaviors highlight the inherent
556
challenge in disambiguating exercise intervention effects in cross sectional studies of free-living
557
Formatted: Justified
Formatted: Font: 18 pt
Formatted: Font: 18 pt
populations, where healthy behaviors such as exercising and healthy dietary habits co-occur.
558
Overall, our analysis shows that the differences between those who run and the sedentary in our
559
cohort are likely a combination of the specific physiological effects of exercise, the associated
560
changes in BMI, and lifestyle habits associated with those who exercise, such as food choices and
561
baseline activity level. Looking ahead, the InsideTracker database is continuously augmented
562
with blood chemistry, genotyping, and activity tracker data, facilitating further investigation of the
563
effects of various exercise modalities on phenotypes related to healthspan, including longitudinal
564
analyses and more granular dose-response dynamics.
565
Data Availability Statement
566
The full set of biomarker change correlations has been made available in the Supplementary
567
Information files. Specific components of the raw dataset are available upon reasonable request
568
from the corresponding author. 2S-MR analysis was performed using publicly available datasets
569
via the TwoSampleMR R package.
570
Ethics statement
571
This study was submitted to The Institutional Review Board (IRB), which determined this work
572
was not subject to a review based on category 4 exemption (“secondary research” with de-
573
identified subjects).
574
Author contributions
575
BN performed the 2S-MR analyses, calculated PGSs, and wrote the manuscript; SV performed
576
blood biomarker and blood biomarker X PGS interaction analysis; PF calculated PGSs; MJ, AT,
577
31
and GB provided guidance. All authors have read and agreed to the published version of the
578
manuscript.
579
Funding
580
InsideTracker was the sole funding source.
581
Conflict of interest
582
InsideTracker is a direct-to-consumer blood biomarker and genomics company providing its
583
users with nutritional and exercise recommendations toward improving wellness. B.N., S.V.,
584
P.F., and G.B. are employees of InsideTracker.
585
586
587
588
589
590
Acknowledgments
591
InsideTracker is the sole funding source. We thank Michelle Cawley and Renee Deehan for their
592
assistance with background subject matter research and insightful conversations.
593
594
Formatted: Font: 18 pt
595
596
597
598
599
600
601
602
603
604
605
606
607
References
608
1.
Lavie CJ, Ozemek C, Carbone S, Katzmarzyk PT, Blair SN. Sedentary Behavior,
609
Exercise, and Cardiovascular Health. Circ Res. 2019;124(5):799-815. doi:
610
10.1161/CIRCRESAHA.118.312669. PubMed PMID: 30817262.
611
2.
Carlson SA, Adams EK, Yang Z, Fulton JE. Percentage of Deaths Associated With
612
Inadequate Physical Activity in the United States. Prev Chronic Dis. 2018;15:E38. Epub
613
Formatted: Font: 18 pt
Formatted: Font: (Default) Times New Roman
33
20180329. doi: 10.5888/pcd18.170354. PubMed PMID: 29602315; PubMed Central PMCID:
614
PMCPMC5894301.
615
3.
Antonicelli R, Spazzafumo L, Scalvini S, Olivieri F, Matassini MV, Parati G, et al.
616
Exercise: a "new drug" for elderly patients with chronic heart failure. Aging (Albany NY).
617
2016;8(5):860-72. doi: 10.18632/aging.100901. PubMed PMID: 26953895; PubMed Central
618
PMCID: PMCPMC4931840.
619
4.
Sgro P, Emerenziani GP, Antinozzi C, Sacchetti M, Di Luigi L. Exercise as a drug for
620
glucose management and prevention in type 2 diabetes mellitus. Curr Opin Pharmacol.
621
2021;59:95-102. Epub 20210626. doi: 10.1016/j.coph.2021.05.006. PubMed PMID: 34182427.
622
5.
Vina J, Sanchis-Gomar F, Martinez-Bello V, Gomez-Cabrera MC. Exercise acts as a
623
drug; the pharmacological benefits of exercise. Br J Pharmacol. 2012;167(1):1-12. doi:
624
10.1111/j.1476-5381.2012.01970.x. PubMed PMID: 22486393; PubMed Central PMCID:
625
PMCPMC3448908.
626
6.
Lee EC, Fragala MS, Kavouras SA, Queen RM, Pryor JL, Casa DJ. Biomarkers in Sports
627
and Exercise: Tracking Health, Performance, and Recovery in Athletes. J Strength Cond Res.
628
2017;31(10):2920-37. doi: 10.1519/JSC.0000000000002122. PubMed PMID: 28737585;
629
PubMed Central PMCID: PMCPMC5640004.
630
7.
Lin X, Zhang X, Guo J, Roberts CK, McKenzie S, Wu WC, et al. Effects of Exercise
631
Training on Cardiorespiratory Fitness and Biomarkers of Cardiometabolic Health: A Systematic
632
Review and Meta-Analysis of Randomized Controlled Trials. J Am Heart Assoc. 2015;4(7).
633
Epub 20150626. doi: 10.1161/JAHA.115.002014. PubMed PMID: 26116691; PubMed Central
634
PMCID: PMCPMC4608087.
635
8.
Li X, Ploner A, Wang Y, Zhan Y, Pedersen NL, Magnusson PK, et al. Clinical
636
biomarkers and associations with healthspan and lifespan: Evidence from observational and
637
genetic data. EBioMedicine. 2021;66:103318. Epub 2021/04/05. doi:
638
10.1016/j.ebiom.2021.103318. PubMed PMID: 33813140; PubMed Central PMCID:
639
PMCPMC8047464.
640
9.
Mailliez A, Guilbaud A, Puisieux F, Dauchet L, Boulanger E. Circulating biomarkers
641
characterizing physical frailty: CRP, hemoglobin, albumin, 25OHD and free testosterone as best
642
biomarkers. Results of a meta-analysis. Exp Gerontol. 2020;139:111014. Epub 20200626. doi:
643
10.1016/j.exger.2020.111014. PubMed PMID: 32599147.
644
10.
Hirata T, Arai Y, Yuasa S, Abe Y, Takayama M, Sasaki T, et al. Associations of
645
cardiovascular biomarkers and plasma albumin with exceptional survival to the highest ages. Nat
646
Commun. 2020;11(1):3820. Epub 20200730. doi: 10.1038/s41467-020-17636-0. PubMed PMID:
647
32732919; PubMed Central PMCID: PMCPMC7393489.
648
11.
Erema VV, Yakovchik AY, Kashtanova DA, Bochkaeva ZV, Ivanov MV, Sosin DV, et
649
al. Biological Age Predictors: The Status Quo and Future Trends. Int J Mol Sci. 2022;23(23).
650
Epub 20221201. doi: 10.3390/ijms232315103. PubMed PMID: 36499430; PubMed Central
651
PMCID: PMCPMC9739540.
652
12.
Hartmann A, Hartmann C, Secci R, Hermann A, Fuellen G, Walter M. Ranking
653
Biomarkers of Aging by Citation Profiling and Effort Scoring. Front Genet. 2021;12:686320.
654
Epub 20210521. doi: 10.3389/fgene.2021.686320. PubMed PMID: 34093670; PubMed Central
655
PMCID: PMCPMC8176216.
656
13.
Trejo-Gutierrez JF, Fletcher G. Impact of exercise on blood lipids and lipoproteins. J Clin
657
Lipidol. 2007;1(3):175-81. Epub 20070607. doi: 10.1016/j.jacl.2007.05.006. PubMed PMID:
658
21291678.
659
14.
Westerman K, Reaver A, Roy C, Ploch M, Sharoni E, Nogal B, et al. Longitudinal
660
analysis of biomarker data from a personalized nutrition platform in healthy subjects. Sci Rep.
661
2018;8(1):14685. Epub 2018/10/04. doi: 10.1038/s41598-018-33008-7. PubMed PMID:
662
30279436; PubMed Central PMCID: PMCPMC6168584.
663
15.
Fox J WS. An R Companion to Applied Regression. Third ed: Sage, Thousand Oaks CA;
664
2019.
665
16.
Ho D, Imai K, King G, Stuart EA. MatchIt: Nonparametric Preprocessing for Parametric
666
Causal Inference. Journal of Statistical Software. 2011;42(8):1 - 28. doi: 10.18637/jss.v042.i08.
667
17.
Burgess S, Davey Smith G, Davies NM, Dudbridge F, Gill D, Glymour MM, et al.
668
Guidelines for performing Mendelian randomization investigations. Wellcome Open Res.
669
2019;4:186. Epub 20200428. doi: 10.12688/wellcomeopenres.15555.2. PubMed PMID:
670
32760811; PubMed Central PMCID: PMCPMC7384151.
671
18.
Hemani G, Zheng J, Elsworth B, Wade KH, Haberland V, Baird D, et al. The MR-Base
672
platform supports systematic causal inference across the human phenome. Elife. 2018;7. Epub
673
20180530. doi: 10.7554/eLife.34408. PubMed PMID: 29846171; PubMed Central PMCID:
674
PMCPMC5976434.
675
19.
Bogl LH, Pietilainen KH, Rissanen A, Kaprio J. Improving the accuracy of self-reports
676
on diet and physical exercise: the co-twin control method. Twin Res Hum Genet.
677
2009;12(6):531-40. doi: 10.1375/twin.12.6.531. PubMed PMID: 19943715.
678
20.
Yuen HK, Wang E, Holthaus K, Vogtle LK, Sword D, Breland HL, et al. Self-reported
679
versus objectively assessed exercise adherence. Am J Occup Ther. 2013;67(4):484-9. doi:
680
10.5014/ajot.2013.007575. PubMed PMID: 23791324; PubMed Central PMCID:
681
PMCPMC3722661.
682
21.
Ostman C, Smart NA, Morcos D, Duller A, Ridley W, Jewiss D. The effect of exercise
683
training on clinical outcomes in patients with the metabolic syndrome: a systematic review and
684
meta-analysis. Cardiovasc Diabetol. 2017;16(1):110. Epub 20170830. doi: 10.1186/s12933-017-
685
0590-y. PubMed PMID: 28854979; PubMed Central PMCID: PMCPMC5577843.
686
22.
Fikenzer K, Fikenzer S, Laufs U, Werner C. Effects of endurance training on serum
687
lipids. Vascul Pharmacol. 2018;101:9-20. Epub 20171201. doi: 10.1016/j.vph.2017.11.005.
688
PubMed PMID: 29203287.
689
23.
Mann S, Beedie C, Jimenez A. Differential effects of aerobic exercise, resistance training
690
and combined exercise modalities on cholesterol and the lipid profile: review, synthesis and
691
recommendations. Sports Med. 2014;44(2):211-21. doi: 10.1007/s40279-013-0110-5. PubMed
692
PMID: 24174305; PubMed Central PMCID: PMCPMC3906547.
693
24.
Tambalis K, Panagiotakos DB, Kavouras SA, Sidossis LS. Responses of blood lipids to
694
aerobic, resistance, and combined aerobic with resistance exercise training: a systematic review
695
of current evidence. Angiology. 2009;60(5):614-32. Epub 20081030. doi:
696
10.1177/0003319708324927. PubMed PMID: 18974201.
697
25.
Calabresi L, Franceschini G. Lecithin:cholesterol acyltransferase, high-density
698
lipoproteins, and atheroprotection in humans. Trends Cardiovasc Med. 2010;20(2):50-3. doi:
699
10.1016/j.tcm.2010.03.007. PubMed PMID: 20656215.
700
26.
Kemper HC, Snel J, Verschuur R, Storm-van Essen L. Tracking of health and risk
701
indicators of cardiovascular diseases from teenager to adult: Amsterdam Growth and Health
702
Study. Prev Med. 1990;19(6):642-55. doi: 10.1016/0091-7435(90)90061-n. PubMed PMID:
703
2263575.
704
35
27.
Khan A, Khan WM, Ayub M, Humayun M, Haroon M. Ferritin Is a Marker of
705
Inflammation rather than Iron Deficiency in Overweight and Obese People. J Obes.
706
2016;2016:1937320. Epub 20161227. doi: 10.1155/2016/1937320. PubMed PMID: 28116148;
707
PubMed Central PMCID: PMCPMC5223018.
708
28.
Nabhan D, Bielko S, Sinex JA, Surhoff K, Moreau WJ, Schumacher YO, et al. Serum
709
ferritin distribution in elite athletes. J Sci Med Sport. 2020;23(6):554-8. Epub 20191227. doi:
710
10.1016/j.jsams.2019.12.027. PubMed PMID: 31901316.
711
29.
Ku BJ, Kim SY, Lee TY, Park KS. Serum ferritin is inversely correlated with serum
712
adiponectin level: population-based cross-sectional study. Dis Markers. 2009;27(6):303-10. doi:
713
10.3233/DMA-2009-0676. PubMed PMID: 20075513; PubMed Central PMCID:
714
PMCPMC3835072.
715
30.
Pollock N, Chakraverty R, Taylor I, Killer SC. An 8-year Analysis of Magnesium Status
716
in Elite International Track & Field Athletes. J Am Coll Nutr. 2020;39(5):443-9. Epub
717
20191212. doi: 10.1080/07315724.2019.1691953. PubMed PMID: 31829845.
718
31.
Arnaud MJ. Update on the assessment of magnesium status. Br J Nutr. 2008;99 Suppl
719
3:S24-36. doi: 10.1017/S000711450800682X. PubMed PMID: 18598586.
720
32.
Simo R, Saez-Lopez C, Barbosa-Desongles A, Hernandez C, Selva DM. Novel insights
721
in SHBG regulation and clinical implications. Trends Endocrinol Metab. 2015;26(7):376-83.
722
Epub 20150601. doi: 10.1016/j.tem.2015.05.001. PubMed PMID: 26044465.
723
33.
Popovic B, Popovic D, Macut D, Antic IB, Isailovic T, Ognjanovic S, et al. Acute
724
Response to Endurance Exercise Stress: Focus on Catabolic/anabolic Interplay Between Cortisol,
725
Testosterone, and Sex Hormone Binding Globulin in Professional Athletes. J Med Biochem.
726
2019;38(1):6-12. Epub 20190301. doi: 10.2478/jomb-2018-0016. PubMed PMID: 30820178;
727
PubMed Central PMCID: PMCPMC6298450.
728
34.
Simo R, Saez-Lopez C, Lecube A, Hernandez C, Fort JM, Selva DM. Adiponectin
729
upregulates SHBG production: molecular mechanisms and potential implications.
730
Endocrinology. 2014;155(8):2820-30. Epub 20140514. doi: 10.1210/en.2014-1072. PubMed
731
PMID: 24828613.
732
35.
Mougios V. Reference intervals for serum creatine kinase in athletes. Br J Sports Med.
733
2007;41(10):674-8. Epub 20070525. doi: 10.1136/bjsm.2006.034041. PubMed PMID:
734
17526622; PubMed Central PMCID: PMCPMC2465154.
735
36.
Banfi G, Morelli P. Relation between body mass index and serum aminotransferases
736
concentrations in professional athletes. J Sports Med Phys Fitness. 2008;48(2):197-200. PubMed
737
PMID: 18427415.
738
37.
Ross R, Janiszewski PM. Is weight loss the optimal target for obesity-related
739
cardiovascular disease risk reduction? Can J Cardiol. 2008;24 Suppl D:25D-31D. doi:
740
10.1016/s0828-282x(08)71046-8. PubMed PMID: 18787733; PubMed Central PMCID:
741
PMCPMC2794451.
742
38.
Ross R. Does exercise without weight loss improve insulin sensitivity? Diabetes Care.
743
2003;26(3):944-5. doi: 10.2337/diacare.26.3.944. PubMed PMID: 12610063.
744
39.
Cerqueira E, Marinho DA, Neiva HP, Lourenco O. Inflammatory Effects of High and
745
Moderate Intensity Exercise-A Systematic Review. Front Physiol. 2019;10:1550. Epub
746
20200109. doi: 10.3389/fphys.2019.01550. PubMed PMID: 31992987; PubMed Central
747
PMCID: PMCPMC6962351.
748
40.
Church TS, Earnest CP, Thompson AM, Priest EL, Rodarte RQ, Saunders T, et al.
749
Exercise without weight loss does not reduce C-reactive protein: the INFLAME study. Med Sci
750
Sports Exerc. 2010;42(4):708-16. doi: 10.1249/MSS.0b013e3181c03a43. PubMed PMID:
751
19952828; PubMed Central PMCID: PMCPMC2919641.
752
41.
Plaisance EP, Grandjean PW. Physical activity and high-sensitivity C-reactive protein.
753
Sports Med. 2006;36(5):443-58. doi: 10.2165/00007256-200636050-00006. PubMed PMID:
754
16646631.
755
42.
Salvagno GL, Sanchis-Gomar F, Picanza A, Lippi G. Red blood cell distribution width:
756
A simple parameter with multiple clinical applications. Crit Rev Clin Lab Sci. 2015;52(2):86-
757
105. Epub 20141223. doi: 10.3109/10408363.2014.992064. PubMed PMID: 25535770.
758
43.
Carmel R, Green R, Rosenblatt DS, Watkins D. Update on cobalamin, folate, and
759
homocysteine. Hematology Am Soc Hematol Educ Program. 2003:62-81. doi:
760
10.1182/asheducation-2003.1.62. PubMed PMID: 14633777.
761
44.
Welsh P, Polisecki E, Robertson M, Jahn S, Buckley BM, de Craen AJ, et al. Unraveling
762
the directional link between adiposity and inflammation: a bidirectional Mendelian
763
randomization approach. J Clin Endocrinol Metab. 2010;95(1):93-9. Epub 20091111. doi:
764
10.1210/jc.2009-1064. PubMed PMID: 19906786; PubMed Central PMCID:
765
PMCPMC2805500.
766
45.
Maachi M, Pieroni L, Bruckert E, Jardel C, Fellahi S, Hainque B, et al. Systemic low-
767
grade inflammation is related to both circulating and adipose tissue TNFalpha, leptin and IL-6
768
levels in obese women. Int J Obes Relat Metab Disord. 2004;28(8):993-7. doi:
769
10.1038/sj.ijo.0802718. PubMed PMID: 15211360.
770
46.
Katzmarzyk PT, Leon AS, Rankinen T, Gagnon J, Skinner JS, Wilmore JH, et al.
771
Changes in blood lipids consequent to aerobic exercise training related to changes in body
772
fatness and aerobic fitness. Metabolism. 2001;50(7):841-8. doi: 10.1053/meta.2001.24190.
773
PubMed PMID: 11436192.
774
47.
Hu X, Zhuang XD, Mei WY, Liu G, Du ZM, Liao XX, et al. Exploring the causal
775
pathway from body mass index to coronary heart disease: a network Mendelian randomization
776
study. Ther Adv Chronic Dis. 2020;11:2040622320909040. Epub 20200527. doi:
777
10.1177/2040622320909040. PubMed PMID: 32523662; PubMed Central PMCID:
778
PMCPMC7257848.
779
48.
Huang Y, Xu M, Xie L, Wang T, Huang X, Lv X, et al. Obesity and peripheral arterial
780
disease: A Mendelian Randomization analysis. Atherosclerosis. 2016;247:218-24. Epub
781
20151229. doi: 10.1016/j.atherosclerosis.2015.12.034. PubMed PMID: 26945778.
782
49.
Houmard JA, Costill DL, Mitchell JB, Park SH, Fink WJ, Burns JM. Testosterone,
783
cortisol, and creatine kinase levels in male distance runners during reduced training. Int J Sports
784
Med. 1990;11(1):41-5. doi: 10.1055/s-2007-1024760. PubMed PMID: 2180832.
785
50.
Schorr M, Lawson EA, Dichtel LE, Klibanski A, Miller KK. Cortisol Measures Across
786
the Weight Spectrum. J Clin Endocrinol Metab. 2015;100(9):3313-21. Epub 20150714. doi:
787
10.1210/JC.2015-2078. PubMed PMID: 26171799; PubMed Central PMCID:
788
PMCPMC4570173.
789
51.
O'Donnell CJ, Sabatine MS. Opportunities and Challenges in Mendelian Randomization
790
Studies to Guide Trial Design. JAMA Cardiol. 2018;3(10):967. doi:
791
10.1001/jamacardio.2018.2863. PubMed PMID: 30326490.
792
52.
Hosseini B, Berthon BS, Saedisomeolia A, Starkey MR, Collison A, Wark PAB, et al.
793
Effects of fruit and vegetable consumption on inflammatory biomarkers and immune cell
794
populations: a systematic literature review and meta-analysis. Am J Clin Nutr. 2018;108(1):136-
795
55. doi: 10.1093/ajcn/nqy082. PubMed PMID: 29931038.
796
37
53.
Djousse L, Arnett DK, Coon H, Province MA, Moore LL, Ellison RC. Fruit and
797
vegetable consumption and LDL cholesterol: the National Heart, Lung, and Blood Institute
798
Family Heart Study. Am J Clin Nutr. 2004;79(2):213-7. doi: 10.1093/ajcn/79.2.213. PubMed
799
PMID: 14749225.
800
54.
L D. Physical Activity and Dietary Habits of College Students. The Journal of Nurse
801
Practitioners. 2015;11(2):192-8.e2.
802
55.
Shi X, Tubb L, Fingers ST, Chen S, Caffrey JL. Associations of physical activity and
803
dietary behaviors with children's health and academic problems. J Sch Health. 2013;83(1):1-7.
804
doi: 10.1111/j.1746-1561.2012.00740.x. PubMed PMID: 23253284.
805
56.
Naughton P, McCarthy SN, McCarthy MB. The creation of a healthy eating motivation
806
score and its association with food choice and physical activity in a cross sectional sample of
807
Irish adults. Int J Behav Nutr Phys Act. 2015;12:74. Epub 20150606. doi: 10.1186/s12966-015-
808
0234-0. PubMed PMID: 26048166; PubMed Central PMCID: PMCPMC4475298.
809
57.
Klimentidis YC, Raichlen DA, Bea J, Garcia DO, Wineinger NE, Mandarino LJ, et al.
810
Genome-wide association study of habitual physical activity in over 377,000 UK Biobank
811
participants identifies multiple variants including CADM2 and APOE. Int J Obes (Lond).
812
2018;42(6):1161-76. Epub 20180613. doi: 10.1038/s41366-018-0120-3. PubMed PMID:
813
29899525; PubMed Central PMCID: PMCPMC6195860.
814
58.
Arends RM, Pasman JA, Verweij KJH, Derks EM, Gordon SD, Hickie I, et al.
815
Associations between the CADM2 gene, substance use, risky sexual behavior, and self-control:
816
A phenome-wide association study. Addict Biol. 2021;26(6):e13015. Epub 20210218. doi:
817
10.1111/adb.13015. PubMed PMID: 33604983; PubMed Central PMCID: PMCPMC8596397.
818
59.
Morris J, Bailey MES, Baldassarre D, Cullen B, de Faire U, Ferguson A, et al. Genetic
819
variation in CADM2 as a link between psychological traits and obesity. Scientific Reports.
820
2019;9(1):7339. doi: 10.1038/s41598-019-43861-9.
821
822
Supporting information
823
S1 Table. Number of people in each category by age group. Significant trend toward
824
younger individuals reporting higher running volume, with more than 75% of the elite
825
group falling between the ages of 18 and 35.
826
S2 Table. Full running volume vs. blood biomarker results
827
S3 Table. 2S-MR results with BMI as the exposure and select biomarkers as outcomes.
828
S4 Table. 2S-MR results with BMI with biomarkers as exposures and BMI as outcome to
829
assess reverse causality
830
Formatted: Font: Times New Roman, 12 pt, Bold
Formatted: Font: Bold
Formatted: Font: Bold
S5 Table. 2S-MR results with vigorous physical activity as exposure and blood biomarkers
831
as outcomes
832
S6 Table. 2S-MR results with vigorous physical activity as exposure and lifestyle habits as
833
outcomes
834
S7 Table. 2S-MR with healthy/unhealthy dietary habits as exposures and vigorous physical
835
activity as outcome to assess reverse causality
836
S1 Fig. Assumptions of Mendelian randomization
837
S2 Fig. Blood biomarker levels with respect to self-reported running volume and
838
professional athletes
839
S3 Fig. 2S-MR scatter plot showing effects of vigorous physical activity as the exposure on
840
blood biomarkers.
841
S4 Fig. 2S-MR scatter plot showing effects of vigorous physical activity as the exposure
842
dietary habits.
843
S5 Fig. 2S-MR scatter plot showing effects of dietary behaviors as the exposures on vigorous
844
physical activity
845
846
847
Table 1 Study Population Demographics
848
849
Formatted: Line spacing: single, Don't suppress line
numbers
39
Group
N
Female, %
Age, yrs
Body mass index, kg/m2
PRO
82
53.7%
33.68
20.15
HVAM
1103
52.9%
39.48
22.57
MVAM
6747
54.2%
41.49
23.35
LVAM
10877
34.2%
41.16
24.72
SED
4428
48.9%
44.25
27.83
PRO = Professional, HVAM = high volume amateur (>10 hr), MVAM = medium
850
volume amateur (3-10hr), LVAM = low volume amateur (<3 hr), SED = sedentary
851
852
853
854
855
856
857
858
859
860
Formatted Table
Table 2 Blood Biomarkers Significantly Different Among Sedentary
861
Individuals and Those Who Partake in Running for Exercise to
862
Various Degrees
863
BIOMARKER ANOVA P-VALUE TREND P-VALUE LOWEST MEAN
HIGHEST MEAN
ALB
<1e-16
<0.001
MVAM
PRO
ALT
<1e-16
<1e-16
SED
PRO
AST
<1e-16
<0.001
SED
PRO
B12
<0.001
<0.001
SED
PRO
CHOL
<0.001
0.005
PRO
SED
CK
<1e-16
<1e-16
SED
PRO
COR
<0.001
0.675
SED
PRO
FE
<0.001
0.119
SED
PRO
FER
<1e-16
<1e-16
MVAM
SED
FOL
<1e-16
<0.001
SED
PRO
FT
<0.001
0.013
SED
PRO
GGT
<1e-16
<0.001
PRO
SED
GLU
0.087
0.184
PRO
SED
HB
0.002
<0.001
MVAM
PRO
HCT
0.053
0.055
MVAM
PRO
HDL
<1e-16
<0.001
SED
PRO
HBA1C
<0.001
0.010
PRO
SED
Formatted Table
41
HSCRP
<0.001
0.176
PRO
SED
LDL
<0.001
0.006
PRO
SED
MG
<0.001
0.276
PRO
SED
MPV
0.058
0.089
SED
HVAM
NA
<1e-16
0.622
HVAM
SED
RBC_MG
<0.001
0.773
PRO
SED
RDW
<1e-16
0.002
PRO
SED
SHBG
<1e-16
0.004
SED
PRO
TG
<1e-16
<1e-16
PRO
SED
WBC
<1e-16
<1e-16
PRO
SED
864
865
866
Reviewer #1: The majority parts of the articles are technically sound. Moreover, the purpose of
the study is very sound since it focused on healthy active population. Among the few drawbacks
of the study the way the study subjects categorized into groups based on the duration of the
activity (>10hours per week and less that 10hours per week is not appropriate. Moreover, the
reliability/validity of the information sources in relation to the biomarker tests and lifestyle
habits of the study subjects didn't consider the immediate effects of medical services and
medication conditions of the respondents at the time of reporting the volume of exercise and
biomarker test results. Medical services and lifestyle habits specially all the habits in addition to
exercises/running are very important to reach informative decision in this research. So, the
above two points need further explanation or modification.
Response to Reviewer #1: We appreciate the reviewer's feedback and are pleased that they
find the majority of our study technically sound and recognize the importance of our focus on a
healthy, active population. We also appreciate the reviewer pointing out an opportunity to
improve the clarity around our experimental design as it pertains subject groupings.
Regarding the categorization of study subjects, we want to clarify that we actually categorized
them into five groups. These groups include professional endurance runners, high volume
amateur runners (>10 hours per week), medium volume amateur runners (3-10 hours per
week), low volume amateur runners (<3 hours per week), and the sedentary. We now added a
sentence starting on line 125 the explicitly states this categorization (“The cohort was divided
into five groups:…”). These groupings were determined based on the respondents' self-
reported data.
We acknowledge the potential influence of medication use on our analysis, and we now address
it starting on line 461 (“These factors, such as diet, sleep, and/or medications were not readily
ascertained in this free-living cohort…”) and in the Study Limitations section (line 595). We
noted that unmeasured confounders such as medications, nutritional supplements, and
unreported health conditions may exist. However, given the nature of our cohort, which
primarily consists of self-selected, generally healthy individuals, the impact of significant
medication use is expected to be limited. We believe that the observed trends in healthier
biomarker levels with increased reported running volume support this assertion.
Furthermore, we recognize the importance of lifestyle habits beyond exercise in influencing our
results. To address this, we employed statistical genomics, specifically two-sample Mendelian
randomization with physical activity as the exposure. This analysis allowed us to explore other
potential habits and behaviors contributing to improved biomarker signatures in physically
active runners within our cohort. We kindly refer the reviewer to the "Vigorous physical activity
associates with healthier behaviors" section in the results for a detailed examination of this
aspect. Notably, our entire cohort is composed of health-conscious individuals within the same
health advisory platform, with the primary differentiator being self-reported running activity.
We also controlled for key variables such as age, sex, and BMI in our ANOVA analyses.
Response to Reviewers
We hope these explanations clarify our approach and address the reviewer's concerns
adequately.
Reviewer #2: How your data is reliable by using A cross-sectional study design? &
How again the Data is reliable by using self-reported running. I understand that Biomarkers are
objective measure, but do you think that Self-report is trustworthy? Thank you
Response to Reviewer #2: We appreciate the reviewer's questions and concerns regarding the
reliability of our runners data, which is largely derived from self-reported exercise habits. Cross-
sectional studies inherently have limitations when it comes to establishing causality, and we
acknowledge this challenge. To address potential confounding factors, we conducted additional
causal analyses, specifically investigating the effects of BMI on the biomarkers under
examination to begin to disentangle the relative contributions of known factors. Furthermore,
we performed secondary Mendelian randomization (MR) analyses to identify and account for
potential confounders in our findings. We kindly invite the reviewer to explore the "Vigorous
physical activity associates with healthier behaviors" section in the results for a comprehensive
exploration of these confounding aspects.
Regarding the reliability of self-reported running activity, we recognize that self-reports can be
subject to biases, and individuals may tend to overestimate their exercise commitment. To
address this drawback, we added language addressing these limitations in the “Study
limitations” section (Line 579: “First, it is generally known that subjects tend to overestimate
their commitment to exercise …”). We do note that our study cohort comprises self-selected
individuals who are health-conscious and possibly less prone to over-report their running
volume. Additionally, the robust increasing trend in baseline levels of muscle damage
biomarkers (CK, AST), which are known to be associated with participation in sports and
exercise, provides indirect evidence that the different running groups in our study were indeed
engaging in increasing volumes of strenuous physical activity.
While self-reporting has its limitations, it remains a valuable method for capturing individuals'
exercise behaviors in large-scale observational studies. We took measures to mitigate potential
biases, and our findings align with established trends in biomarker responses to physical
activity.
Reviewer #3: Upon a meticulous review of the article in question, I wish to commend the
authors for crafting a piece that not only carries immense scientific weight but is also articulated
with great clarity. Such insightful work surely merits publication in your distinguished journal.
It's admirable how the authors have navigated through a myriad of physiological and
biochemical variables (blood biomarkers) across five distinct participant categories and
presented their results with lucidity. The experimental framework is robust, the statistical
evaluations are apt, and the narrative progresses seamlessly. The references provided are both
relevant and adequate. Nevertheless, I'd like to offer a few observations and suggestions:
Response: We appreciate the reviewer's positive feedback and kind words about our
manuscript. We eagerly await their observations and suggestions should they see further
opportunities to improve our work based on our responses to the current suggestions.
Original Title: “Dose response of running on blood biomarkers of wellness in the generally
healthy.”
Proposed Title: “Dose-response relationship between running and blood biomarkers of wellness
in generally healthy individuals.”
Response: Thank you – title has been changed.
Page 2, Line 8: The mention of “exposure to sunlight” seems somewhat out of context. Could
the authors clarify its relevance or indicate if it has been discussed elsewhere in the article?
Response: Thank you for the suggestion, we removed this as we agree it was not relevant in
this manuscript.
Page 17, Lines 17-18: The text reads: "These observations suggest that elite endurance
runners………to their magnesium status."
Comments: It would be helpful to clarify whether the professional athletes (PRO) participating
in this study are specifically elite endurance runners. Kindly integrate this distinction into the
main text if accurate.
Response: Thank you for the clarifying suggestion. We included the pro/elite endurance
runners clarification within the abstract as well as a section heading (lines 7 and 425)
Page 19, Lines 1-2: The assertion: “Indeed whether exercise………..is inconclusive,” needs to be
substantiated with a relevant citation.
Response: Thank you – citations have been added.
Table 1: Please include standard deviation (SD) values. I also recommend expressing exercise
duration in terms of "h/week" instead of "hr".
Response: Thank you for the catch – units changed to “h/week” and SDs added to Table 1.
We are grateful for your valuable feedback, which has contributed to improving the clarity and
accuracy of our manuscript.
| Dose response of running on blood biomarkers of wellness in generally healthy individuals. | 11-15-2023 | Nogal, Bartek,Vinogradova, Svetlana,Jorge, Milena,Torkamani, Ali,Fabian, Paul,Blander, Gil | eng |
PMC4914003 | Original Article
Brazilian Cardiorespiratory Fitness Classification Based on Maximum
Oxygen Consumption
Artur Haddad Herdy1,2,3 and Ananda Caixeta1
Instituto de Cardiologia de Santa Catarina1; Clínica Cardiosport2; Universidade do Sul de Santa Catarina3, Florianópolis, SC – Brazil
Mailing Address: Artur Haddad Herdy •
Instituto de Cardiologia de Santa Catarina. Rua Newton Ramos 91- 601-A,
Centro. Postal Code 88015-395, Florianópolis, SC – Brazil
E-mail: [email protected]
Manuscript received October 13, 2014; revised manuscript April 30, 2015;
accepted June 26, 2015.
DOI: 10.5935/abc.20160070
Introduction
Cardiopulmonary exercise test (CPET) is considered
one of the most complete tools to assess functional aerobic
capacity, because it provides an integrated assessment of
response to exercise, involving the cardiovascular, pulmonary,
hematopoietic, neurophysiological and skeletal muscle
systems.1 In clinical practice, it has been widely used to
assess cardiac and pulmonary diseases, to stratify the risk of
patients with heart failure, and to optimize the prescription
Abstract
Background: Cardiopulmonary exercise test (CPET) is the most complete tool available to assess functional aerobic
capacity (FAC). Maximum oxygen consumption (VO2 max), an important biomarker, reflects the real FAC.
Objective: To develop a cardiorespiratory fitness (CRF) classification based on VO2 max in a Brazilian sample of healthy
and physically active individuals of both sexes.
Methods: We selected 2837 CEPT from 2837 individuals aged 15 to 74 years, distributed as follows: G1 (15 to 24); G2
(25 to 34); G3 (35 to 44); G4 (45 to 54); G5 (55 to 64) and G6 (65 to 74). Good CRF was the mean VO2 max obtained for
each group, generating the following subclassification: Very Low (VL): VO2 < 50% of the mean; Low (L): 50% - 80%; Fair
(F): 80% - 95%; Good (G): 95% -105%; Excellent (E) > 105%.
Results:
Men
VL < 50%
L 50-80%
F 80-95%
G 95-105%
E > 105%
G1
< 25.30
25.30-40.48
40.49-48.07
48.08-53.13
> 53.13
G2
< 23.70
23.70-37.92
37.93-45.03
45.04-49.77
> 49.77
G3
< 22.70
22.70-36.32
36.33-43.13
43.14-47.67
> 47.67
G4
< 20.25
20.25-32.40
32.41-38.47
38.48-42.52
> 42.52
G5
< 17.54
17.65-28.24
28.25-33.53
33.54-37.06
> 37.06
G6
< 15
15.00-24.00
24.01-28.50
28.51-31.50
> 31.50
Women
G1
< 19.45
19.45-31.12
31.13-36.95
36.96-40.84
> 40.85
G2
< 19.05
19.05-30.48
30.49-36.19
36.20-40.00
> 40.01
G3
< 17.45
17.45-27.92
27.93-33.15
33.16-34.08
> 34.09
G4
< 15.55
15.55-24.88
24.89-29.54
29.55-32.65
> 32.66
G5
< 14.30
14.30-22.88
22.89-27.17
27.18-30.03
> 30.04
G6
< 12.55
12.55-20.08
20.09-23.84
23.85-26.35
> 26.36
Conclusions: This chart stratifies VO2 max measured on a treadmill in a robust Brazilian sample and can be used as an
alternative for the real functional evaluation of physically and healthy individuals stratified by age and sex. (Arq Bras
Cardiol. 2016; 106(5):389-395)
Keywords: Respiratory Function Tests; Exercise; Exercise Test; Oxygen Consumption.
389
Original Article
Herdy & Caixeta
Cardiorespiratory fitness classification
Arq Bras Cardiol. 2016; 106(5):389-395
of physical exercise.2-5 In Brazil, CPET is preferably performed
on a treadmill, but, in many countries, a cycle ergometer is
preferred. Maximum oxygen consumption (VO2 max) reflects
the individual’s maximum capacity to absorb, transport and
consume oxygen.2 The major determinants of normal VO2 max
are: genetic factors, muscle mass amount, age, sex and body
weight.1,2 In practice, VO2 max is considered to be equivalent
to the highest VO2 value obtained in peak exertion, which
is usually used to classify cardiorespiratory fitness (CRF) in a
population. In this study, for practical purposes, we named
VO2 peak, which was actually measured, VO2 max.
Few studies have provided reference CRF charts for
populations, and it is yet to be clarified whether the existing
classifications can be extrapolated to other populations.
Most published studies have been based on small samples,
and the profiles of the populations studied have significantly
differed.6,7 The CRF classification charts most used in Brazil
are as follows: that of the American Heart Association (AHA),
published in 1972 (Table 1), and that by Cooper, of 1987.
Brazil does not have a solid and widely used CRF classification
for CPET; therefore, this study proposes a classification based
on Brazilian population data. Such data, resulting from a
recently published study, were used as reference for CPET
on a treadmill (ramp protocol) for sedentary and physically
active men and women.8
Methods
This study’s sample comprised 9,250 CPET performed at a
large cardiology referral center in southern Brazil.8 Based on
a questionnaire completed during the test, individuals with
the following characteristics were excluded from the study:
any symptom suggesting disease or pathology; amateur or
professional athletes; smokers; users of any medication; obese
individuals (body mass index - BMI > 30); and tests with the
ratio between the amount of carbon dioxide produced and
of oxygen used (respiratory exchange ratio - RER) < 1.1.
After applying the exclusion criteria, 3,922 CPET were
identified, of which, 2,837 CPET, corresponding to healthy
and active individuals, were selected. Those individuals, aged
between 15 and 74 years, were of both sexes and different
ethnicities, and practiced leisure-time aerobic physical activity
for at least 30 minutes a day, three times a week.8
All exercise tests were conducted by cardiologists
trained in ergometry and CPET by the Brazilian Society of
Cardiology Department of Ergometry and Cardiovascular
Rehabilitation. The tests were performed on a treadmill
(Inbrasport - ATL™, Brazil, 1999, Software ErgoPC Elite
Version 3.3.6.2, Micromed Brazil, 1999), using the ramp
protocol. A mixing chamber gas analyzer (MetaLyzer II,
CortexTM - Leipzig, Germany, 2004) was used to collect
the expired gases. For descriptive statistics, central trend
measures, such as means, were used, in addition to
dispersion measures (standard deviation). Excel software,
Microsoft 2008, was used for statistical analyses and charts.
Participants, classified according to sex (female and male),
were divided into six age groups between 15 and 74 years as
follows: G1 (15 to 24 years); G2 (25 to 34 years); G3 (35 to
44 years); G4 (45 to 54 years); G5 (55 to 64 years); and G6
(65 to 74 years).
The CRF classification proposed in this study was based
on 2,837 CPET performed in apparently healthy individuals.
We arbitrarily adopted as “Good” CRF the mean VO2 max
value expressed in mL.kg-1.min-1 obtained in each group, and,
taking that value as a reference, we classified CRF as follows:
“Very Low” (VO2 value < 50% of the mean); “low” (50-80%);
“fair” (80-95%); “good” (95-105%); and “excellent” (> 105%).
To internally validate our proposed CRF classification,
sedentary individuals of both sexes from the study population
sample were assessed, according to previous publication.8
This study was approved by the Ethics Committee in
Research of the Instituto de Cardiologia de Santa Catarina.
Table 1 – American Heart Association Cardiorespiratory Fitness Chart based on maximum oxygen consumption (VO2 max – mL/kg.min) – 1972
Men
Very Low
Low
Fair
Good
Excellent
Age group
20-29
< 25
25-33
34-42
43-52
≥ 53
30-39
< 23
23-30
31-38
39-48
≥ 49
40-49
< 20
20-26
27-35
36-44
≥ 45
50-59
< 18
18-24
25-33
34-42
≥ 43
60-69
< 16
16-22
23-30
31-40
≥ 41
Women
Very Low
Low
Fair
Good
Excellent
Age group
20-29
< 24
24-30
31-37
38-48
≥ 49
30-39
< 20
20-27
28-33
34-44
≥ 45
40-49
< 17
17-23
24-30
31-41
≥ 42
50-59
< 15
15-20
21-27
28-37
≥ 38
60-69
< 13
13-17
18-23
24-34
≥ 35
390
Original Article
Herdy & Caixeta
Cardiorespiratory fitness classification
Arq Bras Cardiol. 2016; 106(5):389-395
Table 2 – Distribution of the physically active and sedentary male population according to mean VO2 max (mL/kg.min) and age groups
Active men
Age (years)
15 – 24
25 – 34
35 – 44
45 – 54
55 – 64
65 – 74
n = 1818
343
597
427
285
134
32
Mean VO2 max (mL/kg.min)
50.6 ± 7.3
47.4 ± 7.4
45.4 ± 6.8
40.5 ± 6.5
35.3 ± 6.2
30 ± 6.1
Sedentary men
Age (years)
15 – 24
25 – 34
35 – 44
45 – 54
55 – 64
65 – 74
n = 570
85
188
157
100
30
10
Mean VO2 max (mL/kg.min)
47.4 ± 7.9
41.9 ± 7.2
39.9 ± 6.8
35.6 ± 7.7
30 ± 6.3
23.1 ± 6.3
Table 3 – Distribution of the physically active and sedentary female population according to mean VO2 max (mL/kg.min) and age groups
Active women
Age (years)
15 – 24
25 – 34
35 – 44
45 – 54
55 – 64
65 – 74
n = 1019
177
300
229
206
81
26
Mean VO2 max (mL/kg.min)
38.9 ± 5.7
38.1 ± 6.6
34.9 ± 5.9
31.1 ± 5.4
28.6 ± 6.1
25.1 ± 4.4
Sedentary women
Age (years)
15 – 24
25 – 34
35 – 44
45 – 54
55 – 64
65 – 74
n = 515
85
149
108
108
40
25
Mean VO2 max (mL/kg.min)
35.6 ± 5.7
34.0 ± 4.8
30.0 ± 5.4
27.2 ± 5.0
23.9 ± 4.2
21.3 ± 3.4
Results
Tables 2 and 3 show the mean VO2 max values of the
original population and the number of CPET performed,
stratified by sex and age groups, of physically active and
sedentary individuals. The VO2 max levels were higher in the
active groups as compared to the sedentary ones, and men
had greater VO2 max levels than women did. Tables 4 and 5
show our proposed CRF classification, with five different
categories, stratified by sex and age group, of apparently
healthy individuals. Table 6 shows the classification of the
sedentary population (men and women) from the original
sample, considering the new CRF chart proposed in this
study. It is worth noting that the CRF of sedentary individuals
is always classified as either fair or low.
As expected, VO2 max levels dropped throughout the age
groups for both sexes (Figures 1 and 2).
Discussion
We elaborated a CRF classification chart based on VO2
max levels measured during CPET (ramp protocol) performed
on an ergometric treadmill, to more accurately classify a solid
Brazilian sample of healthy and physically active individuals of
both sexes. We chose to base our analysis on data of physically
active individuals, who would provide CRF in the “good”
category, corresponding to mean CRF values. Not using data
of sedentary individuals allowed us to validate our proposed
CRF classification chart, observing in which category sedentary
individuals would fit.
According to our CRF classification chart, we confirmed
that the CRF of active men is higher than that of active
women of the same age group, and, for both sexes, active
individuals had a better CRF as compared to sedentary
ones. According to Nunes et al.,7 mean VO2 max values of
women are lower than those of men, the mean VO2 max
values of the former corresponding to only 70% of those of
the latter. The present study showed a mean VO2 max of
women corresponding to 76% to 83% of the mean VO2 max
of men of the same age group.
Sedentary individuals not only had a lower VO2 max as
compared to physically active ones, but also a twice higher
decrease in VO2 max as age advanced.9,10 Regular exercise
practice reduces the VO2 max rate of decrease as compared
to a sedentary lifestyle,11 and, the greater the VO2, the greater
the protection against cardiovascular events. An increase in
aerobic capacity is associated with an increase in survival,
as reported by Myers et al.,12 who have demonstrated a
significant increase in the relative risk of death from any
cause as functional capacity decreased, regardless of the risk
factors involved. In addition, those authors have reported
a 12%-increase in survival for each 1-MET increase in the
CRF level.12
Most CRF classification charts used in clinical practice
have been elaborated in other countries and have not been
validated for the Brazilian population. Extrapolating those
classifications to the Brazilian population can lead to relevant
discrepancies. Belli et al.13 have shown significant discrepancies
when comparing international charts with Brazilian data.
391
Original Article
Herdy & Caixeta
Cardiorespiratory fitness classification
Arq Bras Cardiol. 2016; 106(5):389-395
Table 4 – Classification of cardiorespiratory fitness based on maximum oxygen consumption (VO2 max – mL/kg.min) for the male sex
Age group (years)
Very Low
Low
Fair
Good
Excellent
15 – 24
< 25.30
25.30 – 40.48
40.49 – 48.07
48.08 – 53.13
> 53.13
25 – 34
< 23.70
23.70 – 37.92
37.93 – 45.03
45.04 – 49.77
> 49.77
35 – 44
< 22.70
22.70 – 36.32
36.33 – 43.13
43.14 – 47.67
> 47.67
45 – 54
< 20.25
20.25 – 32.40
32.41 – 38.47
38.48 – 42.52
> 42.52
55 – 64
< 17.54
17.65 – 28.24
28.25 – 33.53
33.54 – 37.06
> 37.06
65 – 74
< 15
15.00 – 24.00
24.01 – 28.50
28.51 – 31.50
> 31.50
Table 5 – Classification of cardiorespiratory fitness based on maximum oxygen consumption (VO2 max – mL/kg.min) for the female sex
Age group (years)
Very Low
Low
Fair
Good
Excellent
15 – 24
< 19.45
19.45 – 31.12
31.13 – 36.95
36.96 – 40.84
> 40.85
25 – 34
< 19.05
19.05 – 30.48
30.49 – 36.19
36.20 – 40.00
> 40.01
35 – 44
< 17.45
17.45 – 27.92
27.93 – 33.15
33.16 – 34.08
> 34.09
45 – 54
< 15.55
15.55 – 24.88
24.89 – 29.54
29.55 – 32.65
> 32.66
55 – 64
< 14.30
14.30 - 22.88
22.89 – 27.17
27.18 – 30.03
> 30.04
65 – 74
< 12.55
12.55 – 20.08
20.09 – 23.84
23.85 – 26.35
> 26.36
Table 6 – Classification of cardiorespiratory fitness based on maximum oxygen consumption (VO2 max – mL/kg.min) of the male and female
sedentary population from the original study and according to the new cardiorespiratory fitness chart proposed in this study
Men
Age group (years)
Very Low
Low
Fair
Good
Excellent
15 – 24
VO2 = 47.4
25 – 34
VO2 = 41.9
35 – 44
VO2 = 39.9
45 – 54
VO2 = 35.6
55 – 64
VO2 = 30.0
65 – 74
VO2 = 23.2
Women
15 – 24
VO2 = 35.6
25 – 34
VO2 = 34.0
35 – 44
VO2 = 30.0
45 – 54
VO2 = 27.2
55 – 64
VO2 = 23.9
65 – 74
VO2 = 21.2
Nunes et al.7 have classified CRF into percentiles, similarly to
Cooper et al., and have observed a difference in VO2 max
when comparing the two charts.
VO2 max depends on a frequent and constant physical
activity and can be enhanced with treinos.14 However, despite
the volume or intensity of the workout raise VO2 max by 10
to 30%, there is also an important genetic influence. Research
has shown that genetic inheritance is the main responsible
for max VO2 each individual and may be responsible for up
to 25% to 50% of the variation in the values of VO2 max, ie,
alone accounts for almost half of ACR.15
VO2 max can be measured directly by analyzing the gases
expired during CPET, or indirectly, by using calculations.
Although some prediction equations provide an acceptable
392
Original Article
Herdy & Caixeta
Cardiorespiratory fitness classification
Arq Bras Cardiol. 2016; 106(5):389-395
Figure 1 – Behavior of maximum oxygen consumption (VO2 max – mL/kg.min) throughout the years in men.
Figure 2 – Behavior of maximum oxygen consumption (VO2 max – mL/kg.min) throughout the years in women.
association with values obtained via direct measurements,
the difference varies, depending on the population studied.
The error for one certain individual can be extremely high,
ranging from 15% to 20% in some studies, and can even reach
or exceed 30%, a high margin of error, considering other
measurements in the biological area16.
According to data obtained in this study, VO2 max
drops with age. That drop in women varies less from
one age group to the other as compared to that in men.
We observed a higher drop in VO2 max among active women
from group 3 to group 4, with a mean of 0.38 mL.kg-1.min-1
per year. Among sedentary women, that drop was sharper
393
Original Article
Herdy & Caixeta
Cardiorespiratory fitness classification
Arq Bras Cardiol. 2016; 106(5):389-395
from group 2 to group 3, with a mean of 0.4 mL.kg-1.min-1
per year. Among both active and sedentary men, however,
the VO2 max drop was more marked from group 5 to
group 6, with a mean of 0.53 mL.kg-1.min-1 per year among
active men, and of 0.69 mL.kg-1.min-1 per year among
sedentary men. Nunes et al.7 have shown a VO2 max drop of
0.4 mL.kg-1.min-1 per year among men aged 20 to 60 years.
Belli et al.,13 using indirect VO2 max measurement, have
evidenced a drop of 20% to 25% per decade in mean VO2
max from the age of 50 years onward, that drop being sharper
after the age of 60 years. An approximate drop in VO2 max
of 0.4 mL.kg-1.min-1 per year is estimated to occur from the
age of 25 years onward, and that VO2 max decline is twice
greater in sedentary individuals as compared to physically
active ones.8,9
We used the new CRF classification chart to classify
sedentary individuals undergoing CPET under the same
conditions of the physically active ones from the original
population. This would allow us to validate our proposed
classification, considering how the VO2 max values of those
individuals would fit. Differently from the studies estimating
VO2 max indirectly, the direct measurement of VO2 max shows
that CRF in sedentary individuals is classified, at the most, as
fair, regardless of age and sex (Table 6). From the practical
viewpoint, sedentary individuals have decreased tolerance
to exertion, and, thus, physical exercise prescription to active
and sedentary individuals should differ.17
The CRF chart by Cooper18 and that of the AHA19
(Table 1) are the most commonly used tools to classify CRF
in CPET programs in Brazil. However, the literature lacks
data concerning sampling methods and sample types used
to elaborate the AHA chart. Therefore, the comparison of
data obtained in this study with the AHA chart is limited. Our
classification comprises a wider age range, from 15 to 74 years,
as compared to that of the AHA (20 to 69 years). The VO2 max
analysis in both charts evidences, in younger age groups, very
similar VO2 max values. However, in the other age groups, a
greater difference is observed between our data and the VO2
max values of the AHA chart.
Most CRF charts published so far have been elaborated
with CPET performed on a cycle ergometer. The VO2 max
obtained in tests performed on a treadmill, as opposed to those
performed on a bicycle, is approximately 5% to 17% higher
(mean of 8%).20,21 The difference is attributed to the amount
of active muscle mass involved in the test, which is greater
for the inclined treadmill. Another important factor relates to
the pedaling effect, which causes localized muscle fatigue by
using the large muscle groups of the thigh, and that fatigue can
occur before maximum exertion is imposed to the circulatory
and respiratory systems, generating a lower VO2 max.2
In our study sample, the age range was wide, including
adolescents older than 15 years. We believe that from that
age on, individuals already have muscle maturation and
performance close to those of young adults under the age
of 25 years.22,23 The classification chart proposed should be
assessed as an instrument to predict risk for morbidity and
mortality, according to each individual’s functional profile.
Further studies are required.
This study has limitations, such as the lack of standardization
of ramp protocols. Individuals classified as physically active
practiced different types of activities and sports, making the
comparison of the results in different populations difficult.
Further studies are required, using the same intensities
and inclinations in the protocol ramp and with individuals
practicing the same type of aerobic exercise, because that
would improve the analysis and comparison of the results.
Individuals with hypertension, diabetes or dyslipidemia, those
on any type of medication, and those with a BMI greater
than 30 (obese) were excluded, making the applicability of
that classification in those subgroups uncertain. The Brazilian
population is known to be diversified, and, in southern Brazil,
the European colonization predominates (smaller percentage
of Afrodescendant and Native individuals, as compared to
other Brazilian regions). New studies should be developed,
including different ethnicities and individuals from other
Brazilian regions, aiming at comparing with the classification
proposed to verify whether the values differ.
Conclusion
This is one of the few Brazilian studies to propose a CRF
chart with data extracted from a robust population sample,
and based on VO2 max measured via CPET on a treadmill.
These data can be used for functional capacity classification
according to sex and age group and considering different
risk profiles.
Author contributions
Conception and design of the research: Herdy AH e Caixeta
A. Acquisition of data: Herdy AH. Analysis and interpretation
of the data: Herdy AH e Caixeta A. Statistical analysis: Herdy
AH e Caixeta A. Obtaining financing: Herdy AH. Writing of
the manuscript: Herdy AH e Caixeta A. Critical revision of the
manuscript for intellectual content: Herdy AH e Caixeta A.
Potential Conflict of Interest
No potential conflict of interest relevant to this article
was reported.
Sources of Funding
There were no external funding sources for this study.
Study Association
This study is not associated with any thesis or dissertation work.
394
Original Article
Herdy & Caixeta
Cardiorespiratory fitness classification
Arq Bras Cardiol. 2016; 106(5):389-395
1.
Wasserman K, Whipp BJ. Exercise physiology in health and disease. Am Rev
Resp Dis. 1975;112(2):219-49.
2.
Wasserman K. Principles of exercise testing and interpretation: including
pathophysiology and clinical applications. 5th ed. Philadelphia: Wolters
Kluwer Health/Lippincott Williams & Wilkins; 2012.
3.
Meneghelo RS, Araújo CG, Stein R, Mastrocolla LE, Albuquerque PF, Serra SM, et
al; Sociedade Brasileira de Cardiologia. III Diretrizes da Sociedade Brasileira de
Cardiologia sobre teste ergométrico. Arq Bras Cardiol. 2010;95(5 sup.1):1-26.
4.
Herdy AH, López-Jimenez F, Terzic CP, Milani M, Stein R, Carvalho T, et
al. South American guidelines for cardiovascular disease prevention and
rehabilitation. Arq Bras Cardiol. 2014;103(2 Suppl.1):1-31.
5.
Arena R, Sietsema KE. Cardiopulmonary exercise testing in the clinical evaluation
of patients with heart an lung disease. Circulation. 2011;123(6):668-80.
6.
Koch B, Shaper C, Ittermannn T, Spielhagen T, Dorr M, Volzke H, et al.
Reference values for cardiopulmonary exercise testing in health volunteers:
the SHIP study. Eur Respir J. 2009;33(2):389-97
7.
Nunes RA, Pontes GF, Dantas PM, Fernandes Filho J. Tabela referencial
de condicionamento cardiorrespiratório. Fitness & Performance Journal.
2005;4(1):27-33.
8.
Herdy AH, Uhlendorf D. Reference values for cardiopulmonary exercise
testing for sedentary and active men and women. Arq Bras Cardiol.
2011;96(1):54-9.
9.
McArdle WD, Katch FI, Katch VL. Fisiologia do exercício: energia, nutrição
e desempenho humano. 3ª. ed. Rio de Janeiro: Guanabara Koogan; 1992.
10. Williams RA. O atleta e a doença cardíaca. Diagnóstico, avaliação e conduta.
Rio de Janeiro: Guanabara Koogan; 2002.
11. Rogers MA, Hagberg JM, Martin WH, Ehsani AA, Holloszy JO. Decline in
VO2 max with aging in master athletes and sedentary men. J Appl Physiol.
1990;68(5):2195-9.
12. Myers J, Prakash M, Froelicher V, Do D, Partington S, Atwood JE. Exercise
capacity and mortality among men referred for exercise testing. N Engl J
Med. 2002;346(11):793-801.
13. Belli KC, Calegaro C, Richter CM, Klafke JZ, Stein R, Viecili PR.
Cardiorespiratory fitness of a Brazilian regional sample distributed in
different tables. Arq Bras Cardiol. 2012;99(3):811-7. Erratum in: Arq Bras
Cardiol. 2012;99(4):965.
14. Duscha BD, Slentz CA, Johnson JL, Houmard JA, Bensimhon DR, Knetzger
KJ, et al. Effects of exercise training amount and intensity on peak oxygen
cosumption in middle-age men and women at risk for cardiovascular
disease. Chest. 2005;128(4):2787-93.
15. Bouchard C, Dionne FT, Simoneau AJ, Boulay MR. Genetics of aerobic and
anaerobic performances. Exerc Sport Sic Rev. 1992;20:27-58.
16. Araújo CG, Herdy AH, Stein R. Maximum oxygen consumption
measurement: valuable biological marker in health and in sickness. Arq
Bras Cardiol. 2013;100(4):e51-3.
17. Costa EC, Costa FC, Oliveira GW, e col. Capacidade cardiorrespiratória de
mulheres jovens com diferentes níveis de atividade física. Revista Brasileira
de Prescrição e Fisiologia do Exercício. 2009;3(14):139-45.
18. Cooper K. The new aerobics. New York: M Evans and Company; 1970.
19. Washington A. Ergometria, reabilitação e cardiologia desportiva. Rio de
Janeiro: Revinter; 2011.
20. Astrand PO. Experimental studies of physical working capacity in relation to
sex and age. Fiep Bulletin. 1952(2):19-21.
21. Neiderberger M, Bruce RA, Kusumi F, Whitkanack S. Disparities in ventilatory
and circulatory responses to bicycle and treadmill exercise. Br Heart J.
1974;36(4):377-82.
22. Rodrigues AN, Perez AJ, Carletti L, Bissoli NS, Abreu GR. Maximum oxygen
uptake in adolescents as measured by cardiopulmonary exercise testing: a
classification proposal. J Pediatr. 2006;82(6):426-30.
23. Ghorayeb N, Costa RV, Castro I, Daher DJ, Oliveira Filho JA, Oliveira
MA, et al; Sociedade Brasileira de Cardiologia. [Guidelines on exercise
and sports cardiology from the Brazilian Society of Cardiology and the
Brazilian Society of Sports Medicine]. Arq Bras Cardiol. 2013;100(1
Suppl. 2):1-41.
References
395
| Brazilian Cardiorespiratory Fitness Classification Based on Maximum Oxygen Consumption. | [] | Herdy, Artur Haddad,Caixeta, Ananda | eng |
PMC7230843 | medicina
Article
Comparison of Subjective Workout Intensities
between Aquatic and Land-based Running in Healthy
Young Males: A Pilot Study
Chang-Hyung Lee 1, Jun Hwan Choi 2
and Soo-Yeon Kim 1,*
1
Rehabilitation Medicine, Pusan National University School of Medicine and Research Institute for
Convergence of Biomedical Science and Technology, Pusan National University Yangsan Hospital,
Yangsan 50612, Korea; [email protected]
2
Department of Rehabilitation Medicine, Jeju National University Hospital, Jeju National University School of
Medicine, Jeju 63241, Korea; [email protected]
*
Correspondence: [email protected]
Received: 17 February 2020; Accepted: 26 March 2020; Published: 28 March 2020
Abstract: Background and objectives: Aquatic exercises have demonstrated several advantages over
land-based exercise, but only a few studies have compared the workout intensities and efficiencies
in a stage-specific manner. This study aimed to investigate workout intensity during aquatic and
land-based running, based on the rating of perceived exertion (RPE). Materials and Methods: Twenty
healthy young male subjects underwent a land-based running test (LRT) and an aquatic running test
(ART), in the form of a cardiopulmonary exercise treadmill test and a shallow-water running test. The
seven stages of the ART were composed of 3 minutes each of the Bruce protocol performed during
the LRT. In the ART, the participants were instructed to run in a swimming pool with matching RPE
to that obtained at each stage of the LRT. Results: Heart rate (HR) during both LRT and ART exhibited
a linear relationship (r = 0.997 and 0.996, respectively; p < 0.001). During the initial and middle
period, HR was higher in the ART than in the LRT. However, in the final period, HR was higher in
the LRT than in the ART. Conclusions: In aquatic exercises based on the RPE obtained from the LRT,
HR exhibited a linear relationship in both the ART and the LRT. The ART appears to increase cardiac
loading more efficiently in the initial period and does not increase cardiac loading abruptly at a later
period. Although there is no precise, objective, controlled parameter to compare the ART and the
LRT, the RPE may be used as a convenient measurement for workout intensity in aquatic running.
Keywords: aquatic exercise;
cardiorespiratory response;
shallow-water running;
rating of
perceived exertion
1. Introduction
Aquatic exercises are popular activities in the context of fitness, therapy, and rehabilitation [1].
Aquatic exercises have several advantages over land-based exercises; they lower risks of injury by
reducing musculoskeletal loading [2], provide a sense of comfort, safety and psychological benefit due
to freedom from concerns of falling down [3], enable aerobic and resistance exercises to be combined
by utilizing the resistance of water [4,5], and, finally, aid in weight reduction by increasing energy
expenditure [6]. Accordingly, aquatic exercises have recently been prescribed for cardiorespiratory
fitness, injured athletes, and the elderly [3,7,8].
Aquatic running methods include deep-water running (DWR) and shallow-water running
(SWR) [9]. During DWR, participants run while fitted with a buoyant belt or vest across a swimming
pool. DWR differs from land running in terms of kinematics and lower limb muscle recruitment [10],
due to the absence of a ground support phase and the involvement of water resistance. On the other
Medicina 2020, 56, 151; doi:10.3390/medicina56040151
www.mdpi.com/journal/medicina
Medicina 2020, 56, 151
2 of 9
hand, during SWR, participants run without any buoyant device and are typically immersed between
the waist and xiphoid process. SWR differs from DWR in terms of the presence of ground reaction
force, which is lower compared to land running and depends on the depth of immersion. Regardless
of water depth, aquatic exercise has been suggested as a good therapeutic method, as it decreases axial
loading on the musculoskeletal structure [1,11]. One can exercise safely in water without inflicting
unnecessary load on the musculoskeletal structure due to weight [7].
However, the efficiency of aquatic exercise has not been clearly demonstrated in previous studies.
A majority of the studies on this topic have demonstrated that aquatic running is characterized by lower
workout intensity than land-based treadmill running, as indicated by lower cardiorespiratory responses
such as peak oxygen consumption (VO2, mL/kg/min) and heart rate [3,12,13]. This observation can be
explained by several factors such as the absence of, or a low, ground reaction force [14], different water
temperatures [15,16], no objective measurement method such as self-selected exercise intensity [2],
and lack of familiarity with aquatic activities [17]. Unlike other exercises, methods for conventional
exercise prescription and the measurement of workout intensity for aquatic exercises are rather limited.
The most frequently used method of measuring results from aquatic exercises depends on ratings
of perceived exertion (RPE) and heart rate (HR, bpm) [1,5,18]. The former is generally rated on the
Borg CR10 Scale [19] in clinics and could be defined as a subjective rating of the intensity of a specific
exercise. Physicians usually prescribe water-based exercises using an appropriate intensity scale of 0 to
10, in accordance with the workout capacity of the patients. However, the measurement of the precise
workout intensity in a controlled manner could enable an objective comparison between aquatic and
land-based exercises.
Although there has been a growth in interest in the use of aquatic exercises for therapy or training,
only a few studies have examined means of measuring or controlling the workout intensities associated
with aquatic exercises. On the basis of their findings, and on its strong correlation with heart rate
during aquatic running, Wilder et al. suggested a cadence that provided subjects with a rhythm for
regular limb movements as a measure of the intensity of aquatic running [18]. In a study on the exercise
intensities of water-based activities, Raffaelli et al. demonstrated that various workout intensities can
be obtained by changing movement frequencies and exercise types, such as jumping, running, and
kicking [20]. However, previous studies on aquatic exercises with buoyancy devices or anaerobic
exercises showed that these conditions could influence workout intensities. In addition to the simple
comparison of total workout intensity, there are no previous studies demonstrating the differences
in terms of intensity increment. Apparently, the above-mentioned results did not provide a precise
comparison between aquatic and land-based exercise. To compare the precise workout intensity
between the two methods, the standard protocol for exercise method and stage evaluation should be
compared using the same standard measurement methods. In addition, although aquatic exercises
have demonstrated several advantages over land-based exercises, only a few studies have compared
the workout intensities and efficiencies in a stage-specific manner [3,7,13]. In previous studies of
aquatic exercises, aquatic treadmills were used to control walking speed and workout intensities (HR,
VO2, and RPE) for comparison with land-based running. However, measuring cardiorespiratory
responses, such as VO2,, during running in a swimming pool in clinical practice is challenging. The aim
of this study was to assess whether subjective perception reflects exercise workout intensity through a
comparison between aquatic and land-based running based on the RPE.
2. Materials and Methods
Twenty healthy young males with no history of musculoskeletal injury and not undergoing any
pharmacological therapy were recruited for this study. To rule out age and sex differences, we recruited
only young males regardless of prior experience in aquatic running. To determine inclusion and
exclusion criteria, the participants were asked to respond to a questionnaire. The inclusion criteria
were as follows: (1) age ranging from 20 to 39 years (mean, 26.0 ± 7.3 years); (2) stature over 160 cm
(mean, 172.1 ± 3.7 cm); (3) body weight ranging from 60 to 75 kg (mean, 66.2 ± 7.3 kg); and (4) body
Medicina 2020, 56, 151
3 of 9
mass index (BMI) ranging from 20 to 26 kg/m2 (mean, 22.3 ± 2.7 kg/m2). All subjects received written
and oral instructions before the test and provided informed consent. This study was approved on 14
Jan 2014 by the institutional review board (IRB 05-2012-018).
2.1. Experimental Setting for Aquatic Running Test (ART)
We suggested an ART protocol for subjective exercise intensity that can be conveniently used
in clinical practice. The water level was adjusted to be between the xiphoid process and the jugular
notch. No buoyancy device was utilized to prevent any possible influence on the workout intensities.
ART was performed in the form of SWR to gain optimal exercise efficiencies with relatively moderate
ground reaction force and much lower physical burden compared to the land-based running test (LRT).
The workout intensities of LRT were assessed by measuring cardiorespiratory responses such as HR,
VO2, and RPE, which provide objective and subjective measures of exercise intensity. Additionally, the
workout intensities of ART were assessed by measuring HR and RPE.
2.2. Aquatic Running and Land-Based Running Procedures
All 20 subjects participated in two running tests. One was a land-based running test in the form of
a cardiopulmonary exercise treadmill test (CPET), and the other was an aquatic running test (Figure 1).
Medicina 2020, 56, x FOR PEER REVIEW
3 of 9
stature over 160 cm (mean, 172.1 ± 3.7 cm); (3) body weight ranging from 60 to 75 kg (mean, 66.2 ± 7.3
kg); and (4) body mass index (BMI) ranging from 20 to 26 kg/m2 (mean, 22.3 ± 2.7 kg/m2). All subjects
received written and oral instructions before the test and provided informed consent. This study was
approved on 14 Jan 2014 by the institutional review board (IRB 05-2012-018).
2.1. Experimental Setting for Aquatic Running Test (ART)
We suggested an ART protocol for subjective exercise intensity that can be conveniently used
in clinical practice. The water level was adjusted to be between the xiphoid process and the jugular
notch. No buoyancy device was utilized to prevent any possible influence on the workout
intensities. ART was performed in the form of SWR to gain optimal exercise efficiencies with
relatively moderate ground reaction force and much lower physical burden compared to the
land-based running test (LRT). The workout intensities of LRT were assessed by measuring
cardiorespiratory responses such as HR, VO2, and RPE, which provide objective and subjective
measures of exercise intensity. Additionally, the workout intensities of ART were assessed by
measuring HR and RPE.
2.2. Aquatic Running and Land-Based Running Procedures
All 20 subjects participated in two running tests. One was a land-based running test in the form
of a cardiopulmonary exercise treadmill test (CPET), and the other was an aquatic running test
(Figure 1).
Figure 1. The land-based running test (LRT) (A) and the aquatic running test (ART) (B). LRT was
performed on an incline-adjustable treadmill with continuous vital sign and electrocardiographic
monitoring. ART was performed in a swimming pool with a water level between the xiphoid-process
and the jugular notch, with monitoring of heart rate (HR) using a water-resistant chest-strap
transmitter.
The LRT was performed before the ART and participants had a rest interval between the two
testing sessions of at least 72 hours to maximize performance in each protocol.
Each running test included a warm-up exercise for 5 minutes. The peak VO2, HR, and RPE
during LRT were measured during the test. LRT was performed on a calibrated, incline-adjustable
treadmill (STEX 8100T, TaeHa, Korea) with real-time recording 12-channel electrocardiographic
monitoring (Philips Health Care 3000 Minuteman Rd., Andover, MA, USA) and vital sign
monitoring based on the Bruce protocol. The Bruce protocol is a standard test in cardiology and
comprises multiple exercise stages that each last 3 minutes. At each stage, the gradient and speed of
the treadmill are elevated to increase work output, called METs (metabolic equivalent of task).
Figure 1. The land-based running test (LRT) (A) and the aquatic running test (ART) (B). LRT was
performed on an incline-adjustable treadmill with continuous vital sign and electrocardiographic
monitoring. ART was performed in a swimming pool with a water level between the xiphoid-process
and the jugular notch, with monitoring of heart rate (HR) using a water-resistant chest-strap transmitter.
The LRT was performed before the ART and participants had a rest interval between the two
testing sessions of at least 72 hours to maximize performance in each protocol.
Each running test included a warm-up exercise for 5 minutes. The peak VO2, HR, and RPE during
LRT were measured during the test. LRT was performed on a calibrated, incline-adjustable treadmill
(STEX 8100T, TaeHa, Korea) with real-time recording 12-channel electrocardiographic monitoring
(Philips Health Care 3000 Minuteman Rd., Andover, MA, USA) and vital sign monitoring based on the
Bruce protocol. The Bruce protocol is a standard test in cardiology and comprises multiple exercise
stages that each last 3 minutes. At each stage, the gradient and speed of the treadmill are elevated
to increase work output, called METs (metabolic equivalent of task). Stage 1 of the Bruce protocol is
performed at 1.7 miles per hour and at a 10% gradient. VO2 was determined by analyzing expired
air through a breath-by-breath method using a portable telemetric system (Ultima Series™ metabolic
stress-testing system, MGC Diagnostics, Saint Paul, Minnesota). For LRT, all the participants were
Medicina 2020, 56, 151
4 of 9
instructed to increase their workout intensity on the treadmill test until the achievement of submaximal
threshold (80% of maximal heart rate or an RPE of 8~9). The exercise test was terminated on the
participant’s request or according to the guidelines of the American College of Sports Medicine [20].
RPE was assessed using a numerical rating scale, the Borg CR10 Scale (0–10).
ART was performed in a swimming pool, while heart rate was monitored using a water-resistant
chest-strap transmitter (Polar T34, Polar Electro, Inc, Kempele, Oulu, Finland), with the water level
reaching between the xiphoid process and the jugular notch (water temperature, 31 ◦C). During ART, all
participants were instructed to run by moving their arm back and forth without swimming while their
legs continued to run in the pool for 3 minutes in each stage, similar to the Bruce protocol performed in
the LRT. To gradually and constantly increase the workout intensity in water, we defined the workout
intensity as HR of ART as the matching RPE that was obtained at each stage of LRT. After acquiring all
seven stages, the stages were classified as follows: initial (stages 1 and 2), middle (stages 3 to 5), and
final (stages 6 and 7). During ART, heart rate was measured at rest and at each stage.
2.3. Statistical Analysis
Kolmogorov–Smirnov verification was used to prove the normality of the data, which were found
to be not normally distributed. A Wilcoxon signed-rank test was used to assess the difference in
heart rate between the ART and the LRT at each stage. Spearman’s correlation was used to analyze
the stage-associated differences. Statistical analyses were performed using SPSS version 21.0 for
Windows (SPSS Inc., Chicago, IL, USA). Statistical significance was accepted for p values < 0.05 and
<0.001 respectively.
3. Results
Twenty male subjects completed LRT and ART on separate days. At the end of each exercise,
the peak VO2, HR, and RPE in LRT were 43.8 ± 3.9 mL/kg/min, 179.5 ± 9.7 bpm, and 8.70 ± 0.82,
respectively. The peak HR for ART at stage 7 were 172.2 ± 4.7 bpm. The final workload for the LRT
(which is equivalent to stage 7 of ART) was at a speed of 6.0 mph and at an inclination of 22% (Table 1).
Table 1. Heart rate (HR), rating of perceived exertion (RPE), and oxygen consumption (VO2) measured
at rest state and during the seven stages of the land-based running test (LRT).
Stage
HR (bpm)
RPE
VO2 (mL/kg/min)
Rest
74.9 ± 9.6
0
3.3 ± 0.8
Stage 1
91.8 ± 10.4
0.3 ± 0.2
8.9 ± 1.0
Stage 2
103.8 ± 8.5
1.2 ± 0.6
13.2 ± 1.3
Stage 3
117.3 ± 9.9
2.5 ± 0.7
18.0 ± 1.4
Stage 4
128.5 ± 9.2
3.7 ± 0.6
24.1 ± 2.4
Stage 5
144.2 ± 10.1
5.1 ± 0.8
30.3 ± 3.9
Stage 6
161.4 ± 11.2
6.8 ± 1.1
36.3 ± 4.3
Stage 7
179.5 ± 9.7
8.7 ± 0.8
43.8 ± 3.9
RPE scores were obtained using the Borg CR10 Scale (0 to 10); Values are presented as mean ± standard deviation
(SD); bpm, beats per minute.
HR during both LRT and ART exhibited a linear relationship (r = 0.997 and 0.996, respectively; p
< 0.001) (Figure 2). During the initial and middle period, HR was higher in ART than LRT; however, in
the final period, HR was higher in LRT than ART. Statistically significant differences were observed
between LRT and ART for HR during stages 2, 3 and 7 (p < 0.05) (Table 2).
Medicina 2020, 56, 151
5 of 9
Medicina 2020, 56, x FOR PEER REVIEW
5 of 9
in the final period, HR was higher in LRT than ART. Statistically significant differences were
observed between LRT and ART for HR during stages 2, 3 and 7 (p < 0.05) (Table 2).
Figure 2. Increase in heart rate (HR) observed during the seven stages of the land-based running test
(LRT) (A) and the aquatic running test (ART) (B) from the rest state. Linear relationship between HR
and the seven stages in LRT (A) and ART (B) (r = 0.997 and 0.996, respectively, p < 0.001, Spearman’s
correlation test); bpm, beats per minute.
Table 2. Comparison of heart rate (HR) between land-based running test (LRT) and aqua-based
running test (ART) according to stages.
Stage
HR (bpm)
Z-score
p Value
LRT
ART
(Mean ± SD)
(Median)
(Mean ± SD)
(Median)
Rest
74.9 ± 9.6
77.5
76.1 ±6.6
78.0
−1.029
0.304
Stage 1
91.8 ± 10.4
91.5
93.9 ± 7.7
93.5
−1.188
0.235
Stage 2
103.8 ± 8.5
106.0
112.7 ± 11.1
113.0
−2.298
0.022 *
Stage 3
117.3 ± 9.9
116.5
123.4 ± 7.5
124.0
−1.989
0.047 *
Stage 4
128.5 ± 9.2
129.0
136.4 ± 8.4
135.5
−1.888
0.059
Stage 5
144.2 ± 10.1
144.0
149.4 ± 7.6
145.5
−1.071
0.284
Stage 6
161.4 ± 11.2
161.0
160.4 ± 6.3
160.0
−0.358
0.721
Stage 7
179.5 ± 9.7
179.5
172.2 ± 4.7
172.5
−2.052
0.040 *
* p < 0.05, Wilcoxon signed-rank test; bpm, beats per minute; Z-score using the normal
approximation to the binomial distribution.
4. Discussion
To prescribe exercise intensity or predict the resulting outcome, precise and controlled exercise
protocols should be prepared. Although the beneficial effects of aquatic exercises have been widely
accepted, the workout intensity should be considered in exercise prescription for therapy or
rehabilitation to obtain optimal positive effects while avoiding possible injury [7,21].
Measurements of total calorie loss or VO2max are useful for acquiring an objective measure of
workout intensity. However, in the absence of appropriate respiratory aquatic analysis equipment,
obtaining such data is challenging, especially in a swimming-pool-based exercise. Theoretically,
subjective parameters used to measure workout intensity have an advantage over objective
parameters in reflecting the actual 'perceived intensity' of individuals. Although the same protocol
Figure 2. Increase in heart rate (HR) observed during the seven stages of the land-based running test
(LRT) (A) and the aquatic running test (ART) (B) from the rest state. Linear relationship between HR
and the seven stages in LRT (A) and ART (B) (r = 0.997 and 0.996, respectively, p < 0.001, Spearman’s
correlation test); bpm, beats per minute.
Table 2. Comparison of heart rate (HR) between land-based running test (LRT) and aqua-based running
test (ART) according to stages.
Stage
HR (bpm)
Z-score
p Value
LRT
ART
(Mean ± SD)
(Median)
(Mean ± SD)
(Median)
Rest
74.9 ± 9.6
77.5
76.1 ±6.6
78.0
−1.029
0.304
Stage 1
91.8 ± 10.4
91.5
93.9 ± 7.7
93.5
−1.188
0.235
Stage 2
103.8 ± 8.5
106.0
112.7 ± 11.1
113.0
−2.298
0.022 *
Stage 3
117.3 ± 9.9
116.5
123.4 ± 7.5
124.0
−1.989
0.047 *
Stage 4
128.5 ± 9.2
129.0
136.4 ± 8.4
135.5
−1.888
0.059
Stage 5
144.2 ± 10.1
144.0
149.4 ± 7.6
145.5
−1.071
0.284
Stage 6
161.4 ± 11.2
161.0
160.4 ± 6.3
160.0
−0.358
0.721
Stage 7
179.5 ± 9.7
179.5
172.2 ± 4.7
172.5
−2.052
0.040 *
* p < 0.05, Wilcoxon signed-rank test; bpm, beats per minute; Z-score using the normal approximation to the
binomial distribution.
4. Discussion
To prescribe exercise intensity or predict the resulting outcome, precise and controlled exercise
protocols should be prepared.
Although the beneficial effects of aquatic exercises have been
widely accepted, the workout intensity should be considered in exercise prescription for therapy or
rehabilitation to obtain optimal positive effects while avoiding possible injury [7,21].
Measurements of total calorie loss or VO2max are useful for acquiring an objective measure of
workout intensity. However, in the absence of appropriate respiratory aquatic analysis equipment,
obtaining such data is challenging, especially in a swimming-pool-based exercise. Theoretically,
subjective parameters used to measure workout intensity have an advantage over objective parameters
in reflecting the actual ’perceived intensity’ of individuals. Although the same protocol for workout
intensity is performed in water, the output (HR in this study) could be inconsistent because exercise
output could be strongly influenced by individual physical characteristics (e.g., running performance in
water) [10] and environmental factors (e.g., water temperature, relative depth of water) [16,22]. In this
study, we measured the subjective RPE value, and the gradual increase in its intensity in accordance
with each stage of the LRT. Due to the viscosity and consistency of water, cadence should not be used
Medicina 2020, 56, 151
6 of 9
as the standard measurement method as speed increases. Difficulty perceptions in the ART increased
with increasing intensity, which corresponded to the RPE in the LRT. HR in both the LRT and the ART
with matching RPE exhibited a linear relationship (r = 0.997 and 0.996, respectively; p < 0.001), and
there were no statistical differences, except in stages 2, 3, and 7. Although there is no precise objective
parameter compared between the ART and the LRT, the RPE may potentially provide a fundamental,
convenient, and meaningful parameter in aqua that reflects an “actual perceived workout intensity”
measure in individuals.
In the present study, HR as exercise workout of ART and LRT were assessed and compared
(Table 2). The plots shown in Figure 2 show an increase in HR and during the three defined test
periods for both ART and LRT. Initially, ART exhibited higher HR values than LRT during the initial
and middle periods. Conversely, HR was lower in ART than LRT during the final period (stage 5 to
6). Theoretically, the ground reaction force due to counter dragging force can be controlled in our
method, thus enabling exercise intensity to be increased gradually during the course of the test. The
maximum kinetic effort was observed during the early stage of the ART, presumably due to the higher
metabolic energy required to overcome the drag force on lower limbs to initiate aquatic running [14],
and at this stage HRs were higher for the ART than in the LRT. Secondly, the mechanical counterforce
corresponding to the water depth should be considered. If a person moves more vigorously in water,
more counterforce is imposed, thus leading to an increase in HR and RPE. However, the kinetic effort
could not increase exponentially, due to the enhanced counterforce at a later period. Despite the greater
effort at a later period, the actual increase in cardiac loading turned out to be small compared to LRT.
It was hypothesized that precise measurement of the workout intensity and effectiveness of aquatic
exercise in a stage-specific manner could allow objective comparisons, thus enabling prescription of
aquatic exercises in a more objective and safe manner in the clinical setting. In the present study,
comparison of the aquatic workout intensity with LRT using RPE as a parameter provided subjective
data which could be useful in clinical practice. A higher HR in ART compared to LRT was observed
until the middle exercise period, therefore suggesting a higher magnitude of enhancement of workout
intensities during the initial and middle period. Aquatic running, a higher-intensity exercise workout
that is less physically burdening than land-based running, is likely to be especially advantageous for
elderly, obese, and severely ill patients. Consequently, aquatic exercise can be recommended over land
exercise for patients with a deconditioned or weakened physical status (for example, patients suffering
from osteoarthritis or rheumatoid arthritis) [23]. Furthermore, the lower HR during the final period of
ART relative to LRT provides a safer exercise window. Considering these aspects, ART could be safely
recommended in clinical settings, as it is also not as demanding as LRT with regard to its physical
burden, especially in patients with a low cardiac ejection fraction below 55% (e.g. heart failure) [24].
There are several possible explanations for the observed workout differences. The physiological
adaptation to water should be considered. In a cross-sectional study, it was demonstrated that
subjects who participated in a session of aquatic exercise achieved acute adaptation. For example, as
pool water temperature decreased, HR increased rapidly in response to sympathetic nervous system
stimulation [16,22]. The increase in peripheral resistance with vasoconstriction is due to the blood being
redirected from the periphery to maintain core temperature [25]. It has been shown that immersion at
neutral temperature (32 ◦C) lowers HR by 15%, but immersion in cold water (14 ◦C) increases HR by
5% [16]. In the present study, the swimming pool temperature was maintained at 31 ◦C; therefore, we
believe that the effect of water temperature on HR was minimal during the early stage of exercise. On
the other hand, RPE has been reported to decrease with the depth of immersion [26], and, in another
study, high RPE appeared to be positively related to ground reaction force, the amount of drag force
on lower limbs [14], and changes in the neuromuscular patterns of active muscles [27]. As ground
reaction forces are always lower during shallow water running (SWR) than during land running,
musculoskeletal burden and the risk of injury could be reduced during SWR.
However, previous studies have demonstrated a decrease in HR with an increase in the water
depth [17,26], and that the decline in HR is associated with the influence of hydrostatic pressure
Medicina 2020, 56, 151
7 of 9
and buoyancy on the stroke volume of heart and consequent alteration in blood distribution in the
body [18]. In the present study, the water level was adjusted to a level between the xiphoid process
and the jugular notch. This depth was more than the specific level used for SWR but shallower than
that used for DWR. Consequently, it was hypothesized that the lower HR observed during the later
period of ART compared to LRT was due to the water depth employed in this study.
Generally, the water depth in swimming pools is usually at adult chest height, and a similar level
was employed in the present study. Apparently, it is hypothesized that our findings are meaningful,
based on the observation that the workout intensities of aquatic exercises reflect real situations, and are
applicable to aquatic exercises in clinical practice.
There are several results of clinical importance in our study. As there was previously no exact
guideline on intensity in aquatic therapy, these findings might suggest a possible guideline for aquatic
exercise. Initially, ART can increase the heart rate efficiently, which appears to be a good choice for
patients in a deconditioned state or unable to walk. Meanwhile HR during the ART rises slower than
during the LRT at later stages. This relatively safe ‘window period’ could be safely suggested in
patients with cardiac problems. In general, heart rate increases linearly in relation to RPE. However,
the heart rate did not increase linearly in later stages of ART. This might reflect the increasing viscosity
and resistance with the increase in running effort, increasing demand for endurance and muscle fatigue
in the water, and hence increasing thermal loss in ART.
There are several limitations of this study. First, all study subjects were young male adults; thus,
our results are applicable primarily to this population and not to female subjects, the elderly, or patients
with cardiac problems. Second, we only considered the partial influence of cardiorespiratory response
on the workout intensity of the ART by measuring HR with controlled RPE, but not VO2, because of
the absence of an appropriate respiratory gas analysis system at the swimming pool. In future studies,
it will be necessary to confirm whether RPE, a subjective parameter, is a meaningful indicator that can
reflect and objectively measure exercise intensity in aquatic exercises in various conditions.
5. Conclusions
This study demonstrates the workout intensities of aquatic and land-based running based on
RPE. The workout difficulty perceptions in aquatic running increased with increasing intensity, which
corresponded to RPE in the land-based running test. Although there is no precise comparison of
controlled cardiorespiratory measurement to compare aquatic and land-based running, RPE may be
used as a meaningful measurement for workout intensity in aquatic running.
Author Contributions: Writing original draft, C.-H.L.; writing-review & editing, S.-Y.K.; methodology, J.H.C. All
authors have read and agreed to the published version of the manuscript.
Funding: This research was supported by a grant of the Korea Health Technology R&D Project through the Korea
Health Industry Development Institute (KHIDI), funded by the Ministry of Health & Welfare, Republic of Korea
(grant number: HI17C2397).
Conflicts of Interest: The authors declare no conflict of interest.
References
1.
Barker, A.L.; Talevski, J.; Morello, R.T.; Brand, C.A.; Rahmann, A.E.; Urquhart, D.M. Effectiveness of aquatic
exercise for musculoskeletal conditions: A meta-analysis. Arch. Phys. Med. Rehabil. 2014, 95, 1776–1786.
[CrossRef]
2.
Colado, J.C.; Triplett, N.T. Monitoring the intensity of aquatic resistance exercises with devices that increase
the drag force: An update. Strength Cond. J. 2009, 31, 94–100. [CrossRef]
3.
Choi, J.H.; Kim, B.R.; Joo, S.J.; Han, E.Y.; Kim, S.Y.; Kim, S.M.; Lee, S.Y.; Yoon, H.M. Comparison of
cardiorespiratory responses during aquatic and land treadmill exercise in patients with coronary artery
disease. J. Cardiopulm. Rehabil. Prev. 2015, 35, 140–146. [CrossRef] [PubMed]
Medicina 2020, 56, 151
8 of 9
4.
Valtonen, A.; Pöyhönen, T.; Sipilä, S.; Heinonen, A. Effects of aquatic resistance training on mobility limitation
and lower-limb impairments after knee replacement. Arch. Phys. Med. Rehabil. 2010, 91, 833–839. [CrossRef]
[PubMed]
5.
Colado, J.C.; Tella, V.; Triplett, N.T. A method for monitoring intensity during aquatic resistance exercises.
J. Strength Cond. Res. 2008, 22, 2045–2049. [CrossRef] [PubMed]
6.
Barbosa, T.M.; Marinho, D.A.; Reis, V.M.; Silva, A.J.; Bragada, J.A. Physiological assessment of head-out
aquatic exercises in healthy subjects: A qualitative review. J. Sports Sci. Med. 2009, 8, 179–189.
7.
Kim, E.; Kim, T.; Kang, H.; Lee, J.; Childers, M.K. Aquatic versus land-based exercises as early functional
rehabilitation for elite athletes with acute lower extremity ligament injury: A pilot study. PM&R 2010, 2,
703–712.
8.
Irandoust, K.; Taheri, M. The effects of aquatic exercise on body composition and nonspecific low back pain
in elderly males. J. Phys. Ther. Sci. 2015, 27, 433–435. [CrossRef]
9.
Dowzer, C.N.; Reilly, T.; Cable, N.T.; Neville, A. Maximal physiological responses to deep and shallow water
running. Ergonomics 1999, 42, 275–281. [CrossRef]
10.
Killgore, G.L.; Wilcox, A.R.; Caster, B.L.; Wood, T.M. A lower-extremities kinematic comparison of deep-water
running styles and treadmill running. J. Strength Cond. Res. 2006, 20, 919–927. [CrossRef]
11.
Waller, B.; Lambeck, J.; Daly, D. Therapeutic aquatic exercise in the treatment of low back pain: A systematic
review. Clin. Rehabil. 2009, 23, 3–14. [CrossRef] [PubMed]
12.
Silvers, W.M.; Rutledge, E.R.; Dolny, D.G. Peak cardiorespiratory responses during aquatic and land treadmill
exercise. Med. Sci. Sports Exerc. 2007, 39, 969–975. [CrossRef] [PubMed]
13.
Phillips, V.K.; Legge, M.; Jones, L.M. Maximal physiological responses between aquatic and land exercise in
overweight women. Med. Sci. Sports Exerc. 2008, 40, 959–964. [CrossRef] [PubMed]
14.
Miyoshi, T.; Shirota, T.; Yamamoto, S.; Nakazawa, K.; Akai, M. Effect of the walking speed to the lower limb
joint angular displacements, joint moments and ground reaction forces during walking in water. Disabil.
Rehabil. 2004, 26, 724–732. [CrossRef] [PubMed]
15.
Bergamin, M.; Ermolao, A.; Matten, S.; Sieverdes, J.C.; Zaccaria, M. Metabolic and cardiovascular responses
during aquatic exercise in water at different temperatures in older adults. Res. Q. Exerc. Sport 2015, 86,
163–171. [CrossRef] [PubMed]
16.
Sramek, P.; Simeckova, M.; Jansky, L.; Savlikova, J.; Vybiral, S. Human physiological responses to immersion
into water of different temperatures. Eur. J. Appl. Physiol. 2000, 81, 436–442. [CrossRef] [PubMed]
17.
Town, G.P.; Bradley, S.S. Maximal metabolic responses of deep and shallow water running in trained runners.
Med. Sci. Sports Exerc. 1991, 23, 238–241. [CrossRef]
18.
Wilder, R.P.; Brennan, D.; Schotte, D.E. A standard measure for exercise prescription for aqua running. Am. J.
Sports Med. 1993, 21, 45–48. [CrossRef]
19.
Borg, E.; Kaijser, L. A comparison between three rating scales for perceived exertion and two different work
tests. Scand. J. Med. Sci. Sports 2006, 16, 57–69. [CrossRef]
20.
Thompson, W.R.; Gordon, N.F.; Pescatello, L.S. ACSM’s Guidelines for Exercise Testing and Prescription;
Lippincott Williams & Wilkins: Philadelphia, PA, USA, 2010.
21.
Wilcock, I.M.; Cronin, J.B.; Hing, W.A. Physiological response to water immersion. Sports Med. 2006, 36,
747–765. [CrossRef]
22.
Bartels, E.M.; Juhl, C.B.; Christensen, R.; Hagen, K.B.; Danneskiold-Samsøe, B.; Dagfinrud, H.; Lund, H.
Aquatic exercise for the treatment of knee and hip osteoarthritis. Cochrane Database Syst. Rev. 2016, 23.
[CrossRef] [PubMed]
23.
Adsett, J.A.; Mudge, A.M.; Morris, N.; Kuys, S.; Paratz, J.D. Aquatic exercise training and stable heart failure:
A systematic review and meta-analysis. Int. J. Cardiol. 2015, 86, 22–28. [CrossRef] [PubMed]
24.
Bonde-Petersen, F.; Schultz-Pedersen, L.; Dragsted, N. Peripheral and central blood flow in man during cold,
thermoneutral, and hot water immersion. Aviat. Space Environ. Med. 1992, 63, 346–350. [PubMed]
25.
Barbosa, T.M.; Garrido, M.F.; Bragada, J. Physiological adaptations to head-out aquatic exercises with
different levels of body immersion. J. Strength Cond. Res. 2007, 21, 1255–1259. [PubMed]
Medicina 2020, 56, 151
9 of 9
26.
Fujisawa, H.; Suenaga, N.; Minami, A. Electromyographic study during isometric exercise of the shoulder in
head-out water immersion. J. Shoulder Elb. Surg. 1998, 7, 491–494. [CrossRef]
27.
Reilly, T.; Dowzer, C.N.; Cable, N.T. The physiology of deep-water running. J. Sports Sci. 2003, 21, 959–972.
[CrossRef]
© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access
article distributed under the terms and conditions of the Creative Commons Attribution
(CC BY) license (http://creativecommons.org/licenses/by/4.0/).
| Comparison of Subjective Workout Intensities between Aquatic and Land-based Running in Healthy Young Males: A Pilot Study. | 03-28-2020 | Lee, Chang-Hyung,Choi, Jun Hwan,Kim, Soo-Yeon | eng |
PMC10681188 | Mouse
Mouse
Rat
Mouse
Rat
Mouse
Rat
Rat
Mouse
Rat
Mouse
Rat
Mouse
Rat
Rat
Mouse
Rat
Mouse
Rat
Mouse
Rat
| Rat and mouse cardiomyocytes show subtle differences in creatine kinase expression and compartmentalization. | 11-27-2023 | Branovets, Jelena,Soodla, Kärol,Vendelin, Marko,Birkedal, Rikke | eng |
PMC7892879 | 1
Vol.:(0123456789)
Scientific Reports | (2021) 11:4091
| https://doi.org/10.1038/s41598-021-83538-w
www.nature.com/scientificreports
Music‑based biofeedback to reduce
tibial shock in over‑ground running:
a proof‑of‑concept study
Pieter Van den Berghe
1*, Valerio Lorenzoni2, Rud Derie
1, Joren Six
2, Joeri Gerlo
1,
Marc Leman
2 & Dirk De Clercq
1
Methods to reduce impact in distance runners have been proposed based on real‑time auditory
feedback of tibial acceleration. These methods were developed using treadmill running. In this study,
we extend these methods to a more natural environment with a proof‑of‑concept. We selected ten
runners with high tibial shock. They used a music‑based biofeedback system with headphones in
a running session on an athletic track. The feedback consisted of music superimposed with noise
coupled to tibial shock. The music was automatically synchronized to the running cadence. The level
of noise could be reduced by reducing the momentary level of tibial shock, thereby providing a more
pleasant listening experience. The running speed was controlled between the condition without
biofeedback and the condition of biofeedback. The results show that tibial shock decreased by 27% or
2.96 g without guided instructions on gait modification in the biofeedback condition. The reduction
in tibial shock did not result in a clear increase in the running cadence. The results indicate that a
wearable biofeedback system aids in shock reduction during over‑ground running. This paves the way
to evaluate and retrain runners in over‑ground running programs that target running with less impact
through instantaneous auditory feedback on tibial shock.
Real‑time feedback on tibial acceleration during treadmill running.
Gait retraining intends to
alter a motor pattern that has become habituated over many years1. Gait retraining has been put forward as a
method to reduce or treat injuries in distance runners2,3. Various studies have focused on the reduction in tibial
shock (i.e., the axial peak tibial acceleration)1,4–7, presumably because the magnitude of the tibial shock has been
associated with tibial stress fracture susceptibility. Evidence for this association is provided in female distance
runners8. Other case–control studies failed to observe a clear difference in groups of runners with and without
a history of tibial stress injury9,10. Nevertheless, gait retraining on a treadmill with the intention of lowering the
impact loading has led to fewer running-related injuries (62% lower injury risk) in novice runners2. These run-
ners could reduce the maximum instantaneous vertical loading rate of the ground reaction force2, an impact
measure that has been correlated with tibial shock during level over-ground running11,12. In several studies4–7,
a reduction in tibial shock has been stimulated by providing biofeedback while participants were running on a
treadmill (Supplementary information file, supplement 1). For instance, Crowell and colleagues provided bio-
feedback that comprised a visual stream of the axial component of tibial acceleration in real-time5. The biofeed-
back was shown to the runners using a screen in front of a treadmill during a single session of gait retraining in
the laboratory. The concept of real-time biofeedback for lower impact running was further developed by Wood
and Kipp, who provided auditory biofeedback in the form of pitched “beeps” scaled relative to a runner’s base-
line of peak tibial acceleration13. This simple auditory feedback was found to be equally effective for tibial shock
reduction compared to the visual biofeedback during a short run on a treadmill in the laboratory14. Another
lab-based study used a combination of visual (traffic lights) and auditory (pitched beeps) feedback modalities in
runners screened for high tibial shock7. The authors reported a reduction in tibial shock of 3.28 g or 31% after
completing a multi-sessions program of gait retraining on a treadmill7. All these lab-studies demonstrate the
effectiveness of biofeedback at reducing tibial shock. Importantly, they pave the way to study lower impact run-
ning through such biofeedback in real-world running environments.
OPEN
1Biomechanics and Motor Control of Human Movement, Department of Movement and Sports Sciences, Ghent
University, 9000 Ghent, Belgium. 2Department of Arts, Music and Theatre Sciences, IPEM, Ghent University,
9000 Ghent, Belgium. *email: [email protected]
2
Vol:.(1234567890)
Scientific Reports | (2021) 11:4091 |
https://doi.org/10.1038/s41598-021-83538-w
www.nature.com/scientificreports/
From pitched beeps to music with superimposed noise.
Efforts have been made to apply music-
based biofeedback15,16. Music can act as a strong motivator for walking and running17–19, so music may be imple-
mented to achieve a pleasant or motivational stimulus. Music-based biofeedback has been explored in people
with brain damage after ischemic stroke or traumatic brain injury, to stimulate weight-shift training in patients
with impairment in balance function20. Music-based biofeedback has also been proven effective to steer posture
parameters while performing a weightlifting task16. There are indications that music can be used as stimulus in
a context of reinforcement learning16. The application of music to reduce tibial shock fits well in the context of
distance running as about half of the recreational runners regularly train with music18. In short, the development
and testing of a wearable music-based biofeedback system will advance the ecological validity of studying run-
ners who engage in gait retraining.
Based on the above considerations, we have developed a wearable music-based biofeedback system. It consists
of a measurement module and a feedback module. The measurement module detects tibial shock and cadence in
real-time using accelerometers12. The feedback module generates shock-dependent pink noise which is superim-
posed onto synchronized music to stimulate lower impact running12,21. The feedback subsystem has the capabil-
ity to synchronize the tempo of the music with the running cadence in real-time, which has been experienced
as motivating22. The use of synchronized audio in an exercise program consisting of locomotor activities has
improved adherence to physical activity23, emphasizing the idea that interaction with music is empowering24,25.
The beats per minute of the music continuously adapt to the steps per minute of the runner21, so the music-
based biofeedback system allows for cadence-induced changes if desired by the user. A high momentary level
of tibial shock results in a high level of noise. If the runner adopts a self-selected gait adaptation which reduces
tibial shock, then the noise level is reduced and the acoustical quality of the music improves. In terms of reinforce-
ment learning this creates a punishment/reward dynamic. The whole wearable music-based biofeedback system
opens the possibility to test whether runners can reduce the cyclic shock experienced in the lower extremities
with the aid of a runner-friendly form of auditory biofeedback.
A self‑discovery approach for lower impact running.
In previous studies, explicit instructions about
running technique have been given to participants with the intention of reducing impact26,27. In these studies27–33
groups of shod runners were asked to substantially increase their running cadence (i.e., steps per minute) or to
change to an anterior foot strike pattern6–28.. Besides explicitly imposing a particular change in running tech-
nique, a more personalized approach is to let the runner discover his or her own motor strategy of lower impact
running with the use of biofeedback as in7,13. In one such preliminary report, Morgan and colleagues observed a
systematic increase in running cadence when groups of about ten runners received visual or auditory real-time
biofeedback with the intention of reducing the magnitude of an unspecified component of peak tibial accelera-
tion on a treadmill14.
Aim and hypotheses.
We sought (1) to determine the potential of music-based feedback to induce lower
impact running in a group of runners who had high tibial shock and (2) to investigate if an eventual shock reduc-
tion would be achieved by a clear increase in the running cadence. Runners with high tibial shock experienced
shock magnitudes in the highest one-third of the population. A systematic review indicated that feedback on
tibial shock has been effective in reducing tibial shock while running on a treadmill34. Consequently, our first
hypothesis was that runners with high tibial shock would be able to decrease their level of tibial shock while
running over-ground with the use of continuous, real-time, auditory biofeedback on tibial shock at a stable
running pace. In a first step toward understanding how runners adapt to real-time auditory biofeedback outside
the traditional laboratory, the running cadence was included in the analysis. Therefore, our second hypothesis
was that the group of high impact runners would spontaneously increase the running cadence in an attempt to
reduce tibial shock at a stable running pace.
Methods
Participants.
For screening purposes, a total of 88 runners were recruited from the Flemish running popu-
lation. Analogous to Clansey and Crowell and colleagues4,7, the current study targeted runners experiencing
high tibial shock. In this case, the runners with a one-legged averaged value of tibial shock in the highest one-
third of the 88 screened runners were contacted to take part in the intervention. The first ten runners who
volunteered were selected. An a priori power analysis (GPower; α = 0.05, an effect size of 1.5, paired testing)
estimated a required sample size of at least seven (n = 7). The effect size in the power analysis was based on results
of a treadmill-based gait retraining program for runners experiencing high tibial shock4. Based on the email
response time, the first ten participants who agreed to participate were selected and engaged in the intervention.
The sample size is in line with previous studies that offered a single session of real-time feedback on tibial shock
to a group of runners6,13,35 (Supplementary information file, supplement 1). The ten selected participants were at
least 6 months injury-free and ran in non-minimalist footwear36. They ran at least 15 km/week distributed over
at least two sessions at the time of the study. Training habits were questioned (Table 1). All participants signed
an informed consent approved by the ethical committee of the Ghent University hospital (Bimetra Number
2015/0864). The methods were carried out following their guidelines and regulations. Informed consent was
obtained from the participants to publish the information/image(s) in an online open-access publication. This
consent was also obtained from test leaders who might be recognizable in some images.
Research design.
The quasi-experimental study was unblinded and used a pre-post design without con-
trols (Fig. 1). Two over-ground running sessions were completed in the runner’s regular sportswear at a speed
of 3.2 ± 0.2 m∙s−1, a common speed range to evaluate endurance running2,6,12,27,38, while instrumented with a
3
Vol.:(0123456789)
Scientific Reports | (2021) 11:4091 |
https://doi.org/10.1038/s41598-021-83538-w
www.nature.com/scientificreports/
wearable system that was developed for real-time identification of tibial shock and auditory biofeedback on tibial
shock. First, we identified the runners with high tibial shock following a screening session (October 2017—Feb-
ruary 2018) in a sports laboratory. Then, a supervised intervention session with auditory biofeedback on tibial
shock took place at an indoor track-and-field site (January—March 2018, supplementary video 1). This study
employed a within-subjects design to examine changes in tibial shock. The days of individual testing were sup-
plemented (digital supplementary file, datasheet). The time required to complete the two sessions was about 2 h
(2 · 1 h). The days between the sessions ranged from 59 to 138 (89 ± 28, mean ± SD).
Screening session.
Set‑up. A test leader instrumented the standing participant with a stand-alone back-
pack system. A 7″ tablet (Windows 10) was fixated to a stripped backpack and connected via USB port to a mi-
crocontroller (Teensy 3.2, PJRC). The microcontroller was connected to two lightweight, tri-axial accelerometers
(LIS331, Sparkfun, Colorado, USA;1000 Hz; ± 24 g) to measure tibial acceleration bilaterally12. The test leader
who instrumented the participant was part of a research team with varying experience and expertise level; from
a last-year student in sports sciences to a post-doctoral researcher. The tibial skin was pre-stretched by strappal
tape at ~ 8 cm above the left and right medial malleolus to minimize unwanted oscillations of the skin in the ver-
tical direction during impact7,12. Thereafter, an accelerometer fitted in a shrink socket with a total mass less than
3 grams was firmly attached to the anteromedial aspect of each lower leg by means of non-elastic zinc oxide tape
(Supplementary information file, supplement 2 a). The axial axis of the accelerometer was aligned visually with
the longitudinal axis of the lower leg before mounting. The tape was tightly fastened by one of the test leaders
to the limit of subject tolerance. The applied alignment has been common practice for research involving tibial
acceleration in running7,12,37.
Procedure. An initial warm-up and familiarization period of five minutes was given along an oval track (circa
32-m length ∙ 5-m width). Participants subsequently ran for circa 20 min. The running speed was monitored on
a trial-by-trial basis by timing gates spanning 6-m near the middle of a straight section. The first five satisfactory
trials of each foot were collected for processing. Trials were discarded if the running speed fell outside the set
boundary of 3.2 ± 0.2 m∙s−1.
Table 1. Participants’ characteristics: anthropometrics and self-reported training habits.
Variable
Mean
SD
Range
Minimum
Maximum
Body height (m)
1.70
0.07
1.59
1.79
Body mass (kg)
67.7
7.4
56.2
82.1
Age (year)
33
9
24
49
Training volume (km/week)
29
12
15
50
Training speed (m∙s−1)
2.88
0.31
2.36
3.33
Figure 1. Schematic overview of the experimental design involving two running sessions (screening and
intervention). A red icon represents a distance runner with high tibial shock. A filled circle indicates a system
check and self-selected rest. Tibial shocks were detected in both sessions. The music-based feedback module was
activated in the biofeedback condition.
4
Vol:.(1234567890)
Scientific Reports | (2021) 11:4091 |
https://doi.org/10.1038/s41598-021-83538-w
www.nature.com/scientificreports/
Data processing. The recorded tibial accelerations were imported for signal processing via custom-built MAT-
LAB scripts. Tibial shock magnitudes corresponding to the first five contacts on a force platform were averaged
for each foot side and per participant. Unfiltered magnitudes of tibial shock were preferred because the tibial
shocks detected by the biofeedback system were derived from the raw signal for the instantaneous auditory bio-
feedback. The leg with the highest value was retained. We evaluated the distribution of tibial shock in the group
of screened runners and invited the runners who experienced shock magnitudes in the highest one-third of that
population.
Intervention session.
Set‑up. The single-session intervention was supervised and took place at a track
and field facility (supplementary video 1). The accelerometers of the wearable system were re-applied to the
participant’s lower leg (supplementary information file, supplement 2 b). The manner of attachment of the ac-
celerometer in the intervention session was intended to be identical to that of the screening session. The simple
mounting technique has resulted in repeatable mean values of the tibial shock between running sessions12. The
participant wore an on-ear headphone (HD25-ii, Sennheiser, Wedemark, Germany).
A smart music player for real‑time music‑based feedback on tibial shock. A peak detection algorithm repetitively
detected the magnitude and timing of tibial shock in each leg12. A custom-built JAVA program operated on the
backpack system and detected a peak every time the axial acceleration exceeded 3 g with no higher axial accel-
eration value measured in the next 375 ms. This simple algorithm was based on a peak detection algorithm taken
from a previous gait retraining study7. The magnitudes and timings were transmitted in real-time through Open
Sound Control to a MAX/MSP patch that was built with the intention of providing music-based biofeedback in
real-time21. Real-time in this context means with negligible delay. For instance, when a new magnitude of tibial
shock was detected, the auditory manipulations were executed in the same stride cycle.
The real-time, continuous, auditory biofeedback consisted of commercially available music tracks with super-
imposed pink noise of variable loudness (Fig. 2). The loudness of the noise depended on the momentary level
of tibial shock of the leg that experienced the greatest mean shock in the baseline measurement. The five last
values of that leg’s tibial shock were averaged through a 5-point moving average to account for inherent step-to-
step variability in tibial shock7. That momentary level of tibial shock was mapped using an empirically validated
fitting to obtain a distinct level of noise loudness21. Six discrete loudness levels (0, 20, 40, 60, 80, 100% of noise)
were created for good discretization (supplementary audio, fragment 1)21, thereby accounting for inter-subject
differences in the decoding accuracies39. The loudness levels were calculated as a percentage of the root-mean-
square amplitude level. So, the upper limit of 100% corresponded to noise with the same amplitude as the
root-mean-square amplitude level of the music. Shock values below the target resulted in music only, meaning
Figure 2. Schematic representation of the biofeedback system’s main components for continuous biofeedback
on tibial shock (axial peak tibial acceleration). An interaction loop of the smart music player that provided
the auditory biofeedback in real-time and that continuously accounted for (in)voluntary alterations in the
running cadence by aligning the tempo (beats per minute) of the music to the cadence (steps per minute) of the
runner. The red horizontal line indicates the baseline tibial shock. The five most recent values of tibial shock are
averaged and mapped to a discretized level of noise loudness, which is added to the music playing.
5
Vol.:(0123456789)
Scientific Reports | (2021) 11:4091 |
https://doi.org/10.1038/s41598-021-83538-w
www.nature.com/scientificreports/
without pink noise (0% of noise). The target of minus ~ 50% of the baseline tibial shock was taken from previous
gait retraining studies4,5,7,14.
The running cadence was derived from the timings of the tibial shocks detected during running. We intended
to repetitively align the tempo of the music (i.e., the beats per minute) to the running cadence (i.e., the steps per
minute) in the biofeedback condition22,40. This real-time synchronization prevents the runner from adjusting his
or her cadence to the tempo of the music and is based on the idea that interaction with music is empowering24,25.
Music of a preferred genre (pop, rock, electronic dance, swing, world) was chosen by the participant. A music
database consisting of seventy-seven tracks with a clear beat in the tempo range of running at sub-maximal
speed was created (supplementary information file, supplement 4). Songs with the right tempo were selected
by a smart music player that instantaneously and continuously adjusted for a change in the running cadence.
Music tempi were manipulated up to ± 4% of the steps per minute without pitch shift21 (supplementary audio,
fragment 2). When a change in steps per minute exceeded this tempo shift for eight seconds, another song started
playing at a tempo that more closely resembled the altered running cadence. An illustrative audio fragment of
a change in a music track was supplemented (supplementary audio, fragment 3). The momentary ratio of the
music-to-motion alignment is described by the ratio of the running cadence (steps per minute) to the tempo of
the music (beats per minute). The ratio should be close to 1 when the beats per minute of the music are aligned
with the steps per minute of the runner.
Procedure. Once bilaterally instrumented with the accelerometers and the backpack (Supplementary informa-
tion file, supplement 2 b, c), participants ran an initial 4.5 min at ~ 3.2 ± 0.2 m/s. This warm-up period functioned
as the no feedback condition wherein no auditory feedback on tibial shock was provided. In the software patch,
the baseline tibial shock of the leg exhibiting the highest overall tibial shock was automatically determined for a
sequence of 90 s (≈ 1 lap of 289 m) in the middle of the no feedback condition. Before the biofeedback condition
started, the runners (i) were familiarized with the different levels of noise loudness by listening to the discrete
noise levels going from minimum to maximum and vice versa (supplementary audio, fragment 1); (ii) chose
their preferred sound volume; (iii) chose their preferred music genre; (iv) received verbal instructions in mother
tongue: “This may be very difficult, but I would like you to try your best to concentrate on the task throughout
the entire intervention. Listen carefully to the distorted music. Try to run with the music as clear as possible
without any distortion at all. If impossible, keep the music distortion as low as possible by modifying your run-
ning technique. The amount of distortion is linked to your tibial shock. The music stops playing when the trial
is over.” So each runner was instructed to find a way to run with a lower level of tibial shock. However, to elicit
self-discovery strategies, no instructions were given on how to reduce the shock magnitude7,13,14. An illustrative
fragment of auditory biofeedback with the different noise levels was supplemented (supplementary audio, frag-
ment 4).
Biofeedback was provided for 20 min in total with a pause after 10 min. The instructions were repeated during
the pause of self-selected duration. The software was configured in such a way that the music and the detection
of tibial shock automatically stopped after the set period of time. The runner finished the lap and met the test
leader at the checkpoint (Supplementary information file, supplement 2 d). Subsequently, the accelerometers
and the backpack were removed from the lower limb. Meanwhile, the runner reported if he or she perceived
any difference in the amount of superimposed noise in the biofeedback condition (yes/no). If so, we asked to
describe the perceived change in running technique. An estimation of exercise intensity was obtained by asking
the runner to give a score (from 1 to 10; from very easy to maximal effort) based on the session rating of perceived
exertion scale41. The subject’s score was collected ~ 5 min after the end of the running session. Three participants
did not report their level of exertion. Accelerometer data were continuously acquired during the no feedback
and biofeedback conditions. Lap times were hand clocked throughout the session to derive the running speed
of a lap. Verbal feedback about the running speed was given on a lap-by-lap basis to the runner.
Data processing. The proportion of the pink noise generated during the 20-min biofeedback run and the
detected tibial shocks were imported for processing using custom-built MATLAB scripts. The tibial shock values
of each individual were extracted for a period of 90 s in both the no feedback and biofeedback conditions. The
period of the no feedback condition corresponded to the period of the baseline measurement. The tibial shocks
belonging to the biofeedback condition were extracted for another period of 90 s at the end of the biofeedback
run. Post hoc inspection of all the registered peaks revealed that the peak detection algorithm worked sub-
optimally by occasionally detecting false-positive peaks. The values belonging to the falsely identified peaks were
post hoc excluded (supplementary information file, supplement 3). The time period at the end of the biofeedback
run was chosen for comparison, like that seen in previous research5,6,13,14. We wanted to obtain a representa-
tive level of overall tibial shock per participant compared to previous research on gait retraining (i.e., 5 to 20
footfalls) (supplementary information file, supplement 1). Therefore, the values of the tibial shock (g) and the
running cadence (steps per minute) of the detected footfalls that belong to the no feedback and the biofeedback
conditions were retained for the larger 90 s time period. The analyzed peaks were considered to be indicative of
foot–ground contact. The number of analyzed footfalls was respectively 125 ± 10 and 132 ± 7, mean ± SD. Hence,
it becomes possible to show the distribution in tibial shock, for example, in someone maximally responding to
the music-based biofeedback. The time between sequential tibial shocks was used to derive the steps per minute
in order to assess the running cadence. The average running speeds of the no feedback and biofeedback condi-
tions were calculated for each participant using the lap times clocked at the indoor track. The running speed was
also determined for those laps corresponding to the extracted tibial shocks.
For further statistical analysis, the tibial shock, the running cadence and the running speeds were averaged
per participant for each condition. Wilcoxon exact signed-rank tests were used for comparison due to the low
6
Vol:.(1234567890)
Scientific Reports | (2021) 11:4091 |
https://doi.org/10.1038/s41598-021-83538-w
www.nature.com/scientificreports/
number of participants. Tibial shock, running cadence and running speeds were compared between the no
feedback condition and the biofeedback condition. Tibial shock and running cadence were tested one-tailed
(p1) because of the directional hypothesis. The Pearson correlation coefficient was calculated post hoc between
the session rating of perceived exertion as reported by the runner and the difference in tibial shock. The alpha
level was set at 0.05 (SPSS). The effect size rES was calculated by dividing the absolute z-score by the square root
of the total number of observations, being rES =|z|/√20. Guidelines for rES are that a small effect is 0.1, a medium
effect is 0.3, and a large effect is 0.542. The individual metrics can be retrieved online (digital supplementary file,
datasheet). The reported values are mean ± SD.
Results
Tibial shock in the intervention session.
Tibial shock was 11.14 ± 1.83 g in the no feedback condition.
The individual averages of tibial shock ranged from 8.92 g to 13.71 g between the participants. Tibial shock
scores were reduced by 27% to 8.19 ± 1.79 g (p1 = 0.001, z = −2.803, rES = 0.627 (large), mean negative rank = 5.50,
absolute range: −0.94 to −7.14 g; relative range: −7 to –53%) in the biofeedback condition (Fig. 3 a), and this
without guided instruction on gait modification.
3
5
7
9
11
13
15
17
No feedback
Biofeedback
hoc ( k g)
s laibi
T
150
160
170
180
190
200
No feedback
Biofeedback
Running cadence (steps per minute)
a
b
*
Figure 3. (a) The axial peak tibial acceleration representing the tibial shock and the (b) running cadence for the
[left] no feedback and [right] biofeedback conditions. Every color is a different participant. The short horizontal
line indicates the mean level of the variable of interest in a condition. * indicates p < 0.05.
0
0.05
5
0.1
Least responder
Normalized count
0.15
10
0.2
Tibial Shock (g)
Average responder
15
20
Greatest responder
Figure 4. Histogram of the tibial shock magnitudes in the analysis period (90-s) for the no feedback (dark)
and biofeedback (light) conditions in the greatest, average and least responders. The footfalls of each runner in a
condition have been normalized to the number of total footfalls in that condition.
7
Vol.:(0123456789)
Scientific Reports | (2021) 11:4091 |
https://doi.org/10.1038/s41598-021-83538-w
www.nature.com/scientificreports/
Figure 4 shows the distribution in tibial shock for the average, most and least pronounced responder. While
there is an overall decrease in the magnitude of tibial shock in these three runners, few footfalls had a tibial shock
that would still be categorized as high. Figure 5 shows the group’s distribution in tibial shock for both conditions.
Music‑based biofeedback characteristics.
The average momentary ratio of the running cadence to the
music tempo was 1.01 ± 0.01 in the 20-min period of biofeedback and 1.02 ± 0.04 in the time period selected for
comparison. The noise loudness to the synchronized music varied from zero to maximum on the group level
(Fig. 6). This means that tibial shocks did occur both below the target (0% of noise) and above the baseline level
of tibial shock (100% of noise). All noise levels were experienced in this group of runners with high tibial shock
(Fig. 6). The individual proportions of the noise levels have been supplemented (Supplementary information
file, supplement 5). The questioned runners responded quasi-immediately after completing the running session
to have perceived a change in noise loudness or quality of the audio during the biofeedback run (digital sup-
plementary file, datasheet).
Temporospatial characteristics.
Figure 3b shows the individual evolution in the running cadence
between the conditions of no feedback and biofeedback. The increase of 4 steps per minute or 2.3% in the
running cadence was not statistically significant (p1 = 0.065, z = − 1.580, rES = 0.353 (moderate), positive mean
rank = + 6.14). The running speed in the 4.5-min no feedback and 20-min biofeedback runs were respectively
3.15 ± 0.12 m∙s−1 and 3.13 ± 0.15 m∙s−1, and did not differ significantly (p = 0.52, z = − 0.71, rES = 0.159 (small)).
The respective running speeds for the laps chosen for tibial shock comparison were 3.18 ± 0.15 m∙s−1 and
3.04 ± 0.10 m∙s−1, and did not differ statistically (p = 0.090, z = − 1.72, rES = 0.385 (moderate)). In addition, the
running speeds remained within the a priori permitted boundary of ± 0.20 m∙s−1.
Perceived exercise intensity.
The mean and median scores of the session rating of perceived exertion
were respectively 4 (somewhat hard) and 3 (moderate) with individual values ranging from 2 to 9 (digital sup-
plementary file, datasheet). In this cohort, the participant reporting the highest rating of perceived exertion also
reported the lowest combined training volume and training speed. The perceived exertion did not correlate to
the absolute (p = 0.460, r = 0.337) nor relative (p = 0.561, r = 0.268) decreases in tibial shock, suggesting that the
attained level of exertion did not influence the achieved reduction in tibial shock.
0
0.05
0.1
0.15
0.2
0.25
4
6
8
10
12
14
16
18
20
22
Tibial shock (g)
0
0.05
0.1
0.15
0.2
0.25
Normalised count
Normalised count
Figure 5. Histogram of the tibial shock magnitudes in the analysis period for the [upper panel] no feedback
condition and the [lower panel] biofeedback condition. Each color represents a participant (n = 10). The number
of footfalls within a single bin has been normalized to the total number of detected footfalls in that condition.
The solid vertical line indicates the tibial shock averaged for all footfalls in that condition.
8
Vol:.(1234567890)
Scientific Reports | (2021) 11:4091 |
https://doi.org/10.1038/s41598-021-83538-w
www.nature.com/scientificreports/
Discussion
The purpose of this proof-of-concept study was twofold: (1) to determine if real-time, continuous, music-based
feedback on tibial shock helps to reduce the shock magnitude during over-ground running at an instructed and
common running speed, and (2) to examine if runners with high tibial shock systematically increase the running
cadence in response to the real-time feedback. A single-session intervention was performed at an instructed run-
ning speed with pre and post measurements in a screened group of runners. The runners who participated in the
intervention session had an averaged value in one of the limbs of at least 9.7 g in tibial shock when screened in
the laboratory. A wearable system provided real-time auditory feedback on a modifiable mechanical parameter
to stimulate lower impact running in a controlled, indoor training environment.
Key implications and discussion regarding the reduction in tibial shock.
In support of our first
hypothesis, runners with high tibial shock decreased their tibial shock by − 27% or − 2.96 g while running
over-ground with the music-based biofeedback. This is the first study performed over-ground in which high
impact runners realized shock reduction with the use of unimodal biofeedback. Our findings build on previous
research about gait retraining in high impact runners4,7,29, and support the limited literature documenting that
self-discovery strategies to achieve shock reduction are effective7,13. For instance, Clansey and colleagues carried
out a randomized controlled trial and reported a decrease of 3.28 g in male runners with high tibial shock who
completed multiple sessions of continuous real-time feedback on tibial shock at the controlled running speed
3.7 m∙s−17. The decrease in tibial shock we found corresponds to the decrease reported by Clansey and col-
leagues in the experimental group, though the present study was performed at the slightly lower running speed
of ~ 3.2 m∙s−1 and in a single-session design. The mixed-sex runners in the present study could run a total of
25 min at 3.2 ± 0.2 m∙s−1 and all achieved shock reduction at the end of the biofeedback run. These runners’ shock
reduction did not correlate to the reported session rating of perceived exertion. Hence, a substantial reduction in
tibial shock is achievable in a heterogeneous group of recreational runners with the aid of a wearable biofeedback
system. The participants were informed about the aim of the intervention (i.e., shock reduction) and they were
aware of the fact that an auditory element was linked with the tibial shock. However, no explicit instructions
about gait modification were given.
Key implications and discussion regarding the expected increase in running cadence.
The
spontaneous self-adaptation in response to the music-based feedback permitted the runners to find their own
solution to cover ground with less tibial shock magnitudes, without reducing the running speed. Self-induced
changes in running cadence were possible because the music’s tempo was continuously and successfully synchro-
nized to the runner’s cadence. Contrary to our second hypothesis, a reduction in tibial shock was not accom-
panied by a systematic increase in the running cadence (or a decrease in step length because the running speed
remained stable). A preliminary and treadmill-based study has reported a systematic reduction in an unspeci-
0
20
40
60
80
100
Noise level
(rms)
Tibial shock
(%)
> 113%
96 - 113%
80 - 95%
65 - 79%
48 - 64%
< 48%
Distribution in the biofeedback run
Figure 6. The proportion of the pink noise generated during the 20-min biofeedback run for the group of high
impact runners. Level 0 represents the ‘music only’ category without superimposed noise. The level of noise
loudness added to the synchronized music has been subdivided into 5 categories. Each level of noise loudness
corresponds to a level of tibial shock relative to the baseline g-value of the runner, which was determined
during the no feedback condition. The value corresponding to 100% of tibial shock is identical to the value of
tibial shock determined in the no feedback condition. Tibial shock is here synonymous to the axial peak tibial
acceleration, rms: root mean square.
9
Vol.:(0123456789)
Scientific Reports | (2021) 11:4091 |
https://doi.org/10.1038/s41598-021-83538-w
www.nature.com/scientificreports/
fied component of peak tibial acceleration when providing real-time auditory feedback in response to that peak
tibial acceleration, that was accompanied by a systematic increase in running cadence of 2 steps per minute or
1.4%14. In the present study the cadence response between the participants was more variable (Figs. 3, 4). The
discrepancy in a systematic change in step frequency between study results highlights the fact that more work is
needed to fully understand the motor strategy or strategies for tibial shock reduction. For instance, another way
to reduce tibial shock may be a change in the discrete foot strike pattern.
An anterior change in foot strike pattern has been found in rearfoot runners with high tibial shock who com-
pleted a treadmill-based, multi-week, retraining program by means of visual and auditory biofeedback on tibial
shock7. In the current study performed in an over-ground running environment and at a slower running speed,
half of our participants claimed to have tried a non-rearfoot strike in the biofeedback condition. Only a single
runner declared to have maintained a forefoot strike until the end of the run. Almost all of the participants (9
out of 10) claimed to have performed a rearfoot strike near the end of the biofeedback condition. Based on our
observations and on the comments made by the participants, we speculate that the real-time feedback on tibial
shock elicits gait alterations with inter-individual differences in kinematic adaptations. Consequently, the gait
alterations may influence shock attenuation strategies. A shift from active shock attenuation to more passive
mechanisms has, for instance, been proposed as possible adaptation during prolonged running at a submaximal
intensity43. When providing biofeedback on the axial peak tibial acceleration, the shock attenuation may rely
more heavily on the active mechanisms (e.g., eccentric muscle contractions, changes to joint angles, and modu-
lating limb stiffness) than passive deformation of the body tissues. Future research may verify our speculations
because 3D kinematics, head nor sacral acceleration were measured.
Discussion regarding the targeted reduction in tibial shock using a music‑based approach.
We
attribute the large effect size obtained in our study to the use of reinforcement. Previous studies that used a
manipulation of music to modulate gait parameters have relied on a steering paradigm that is based on reinforce-
ment learning25,44 according to which people tend to modify their behavior in order to maximize reward and
recursively minimize error (i.e., distance from the target behavior). In this specific case, we sought to reward the
runner by providing a way of obtaining maximum acoustic quality of the synchronized music. The rewarding
effect of running with only music, thus without superposition of pink noise, occurs if the target is reached. Sur-
prisingly, a 50% reduction in tibial shock was reached only for 4.8% of the 20-min biofeedback run. The quote “I
heard several noise levels, but I never heard music without noise” of a participant illustrates this finding. Even the
greatest responder could not fully supress the level of superimposed noise (i.e., so that only synchronized music
would be heard) for the majority of the time (supplementary information file, supplement 5).
Many studies on gait retraining with biofeedback aimed to reduce the runner’s baseline value in tibial shock
with 50%4–7,14,45. But this relative threshold was difficult to achieve or to maintain in the present study. Accord-
ing to our data, a more realistic and relative target for the population of interest seems to be approximately
− 30% in tibial shock. Given that some gait adaptations felt unnatural when trying to achieve a 50% reduction
in tibial shock, a more feasible target of shock reduction may also counteract the slight discomfort reported by
several participants at the end of the run. Nevertheless, more retraining sessions are likely required before the
self-discovered gait pattern is perceived as natural. A cohort of runners with high tibial shock namely reported
that the new gait pattern felt natural by the end of the sixth retraining session, comprising the instruction to run
softer and the use of real-time feedback about tibial acceleration4.
Next to feasibility, it is debatable whether an extreme target of − 50% in tibial shock is required to be clinically
relevant. Chan and colleagues have executed a randomized controlled trial with one-year follow-up and reported
fewer running-related injuries in novice runners who completed a gait retraining program on treadmill2. Even
within the multifactorial nature of injury development, their findings are promising to consider gait retraining as
a preventive strategy for running-related injuries in distance runners who appear to be at risk for injury2. Their
multi-week gait retraining program was performed on an instrumented treadmill with an instruction intended
to reduce the vertical impact peak force. The group of runners who engaged in the retraining program could
reduce the instantaneous vertical loading rate of the ground reaction force by about 15 to 18%, estimated by
manual digitization of the results visualized in Fig. 4 of that publication, and depending on the running speed
tested2. Such a reduction in vertical loading rate might be linked with a reduction in tibial shock because of the
moderate correlation between the vertical loading rate and the tibial shock during over-ground level running11,12.
Multiple lab studies have provided real-time feedback on tibial shock and did report a substantial reduction in
tibial shock and in vertical loading rate post-retraining4,7,45. A reduction of about 30% in both tibial shock and
vertical loading rate has been achieved by runners with high tibial shock post-retraining in a laboratory setting45.
So, a more feasible target of approximately -30% in tibial shock relative to the baseline measurement may still
have potential to reduce or to treat running-related injuries in at-risk runners during level over-ground run-
ning. The evidence for an association between measures of impact over time and running-related injuries has
been conflicting9,10,46–51. Nevertheless, guided usage of a wearable biofeedback system that induces and retains
substantial impact-like reduction over time may have clinical implications for injury risk management.
Limitations.
The self-selected or fixed running speed has been held constant in gait retraining studies that
aim to reduce tibial shock4,7,45. The instructed and lap-by-lap monitored speed of 3.2 ± 0.2 m∙s−1 was slightly
above the group’s self-reported training pace for their typical distance runs (Table 1). It was still less than the
running speed of 3.7 m∙s−1 imposed by Clansey and colleagues7 in male runners during the 20-min retraining
sessions.
The instructed running speed of the present study may affect results since it influences the absolute magni-
tude of impact measures in the time domain, such as tibial shock and the instantaneous vertical loading rate12.
10
Vol:.(1234567890)
Scientific Reports | (2021) 11:4091 |
https://doi.org/10.1038/s41598-021-83538-w
www.nature.com/scientificreports/
Nonetheless, Chan and colleagues showed that the vertical loading rate was lowered at multiple running speeds
after gait retraining2.
Individualization of the instructed speed to the training speed of the participant’s typical distance run may
further increase the ecological value of gait retraining. Given that the session rating of perceived exertion indi-
cates the exercise intensity41, we estimate that the running session was generally performed near the first ven-
tilatory threshold. The average score of 4 on the session rating of perceived exertion scale resembles a physical
effort that was “somewhat hard” in this group of mixed-sex runners. Even the participant who reported the
highest score of 9 was able to reduce tibial shock. No linear relationship was found between shock reduction and
perceived exertion. These results suggest sufficient attention is required for lower impact running with the use
of the biofeedback at the instructed speed. This may not be the case at higher exercise intensities, for instance,
when the runner needs to cope with maintenance of the running pace during exhaustive runs.
The exploration of gait adaptations might affect running economy. Tibial shock reduction has led to more
oxygen being consumed whilst running on treadmill in a single session of gait retraining35. In contrast, a multi-
sessions program comprising real-time feedback on tibial shock resulted in a clear reduction in tibial shock
without affecting the running economy7. Future research may verify the hypothesis of a temporary decrease in
running economy in an over-ground setting because oxygen consumption was not measured in the present study.
The design of this study does not allow confirmation of whether the synchronised music influences the
tibial shock via the biofeedback system. The results can only be attributed to the auditory biofeedback, being
the combination of synchronized music and superimposed noise. Besides a positive effect of music to training
adherence, there might also be other effects because of the ability of music to distract from a task52. It could be
further investigated which kinds of music perform best in a retraining context.
In line with previous studies4,7,45, the study was conducted in healthy runners who demonstrated a charac-
teristic previously associated with a history of tibial stress fracture in distance runners. Therefore, these findings
are not necessarily applicable to injured runners nor to runners with relatively low magnitudes of tibial shock.
The selected group of runners had high tibial shock relative to a screened cohort. That inclusion criterion may
be a reason for the discrepancy in the absolute reduction of tibial shock (g) between studies with and without a
focus on high impact runners only4–7,13,14,35,45.
The changes in outcome cannot be fully attributed to the intervention without comparator group. The lack
of a control group raises questions about whether the reduction in tibial shock is the result of the continuous
real-time feedback or the awareness of the purpose of the feedback (i.e., shock reduction). Verbal information
was given to elicit self-discovery strategies without the provision of direct instructions (e.g., “run softer”, “land
with a toe-strike”) that may influence tibial accelerations. Although we find it unlikely that explicitly instruct-
ing people to “decrease your tibial shock” without clinician or accelerometry guided feedback would result in a
substantial shock reduction at the end of a running session, it remains unknown and unexplored.
Future directions.
The wearable system can instantaneously detect and sonify tibial shock. The next step is
to determine the effectiveness of the biofeedback system in an over-ground gait retraining program with a con-
trol group. A gait retraining program lasting multiple weeks usually involves fading of the feedback stimulus2,4,45.
Analogous to the gradual removal of the continuous and visual stream of tibial acceleration during the last four
sessions by Crowell and colleagues4, the continuous auditory feedback may be faded over time to facilitate inter-
nalization and persistence of an altered gait pattern. An assessment of motor retraining was beyond the scope
of this study, there it normally requires about six to eight sessions to enhance retention of the alterations in the
movement pattern4,7,30, but could be incorporated in gait retraining protocols.
A possibility to retrain runners in more natural environments eliminates the need of exclusive retraining in
laboratory/clinic settings. As such, runners might easily implement the auditory biofeedback-driven approach
of retraining, given some technical improvements (e.g., wireless accelerometer connected to a miniaturized pro-
cessing device) and adequate speed control. The smart music player might also benefit from a feedback protocol
that promotes motor learning in a retraining program consisting of multiple sessions.
Conclusion
Our experimental study without controls shows that a substantial reduction in tibial shock can be stimulated
with the use of continuous music-based biofeedback. If the runners are aware of the direct link between the
tibial shock and the clarity of the music, there is no need to impose a particular gait modification with the intent
of shock reduction. The proof-of-concept supports the idea that lower impact running is possible in an over-
ground environment by providing instantaneous auditory information on biomechanical data via a wearable
biofeedback system.
Data availability
The dataset used for statistical analysis and several exemplar audio fragments are available in the supplementary
materials.
Received: 7 November 2019; Accepted: 24 January 2021
References
1. Davis, I. S. & Futrell, E. Gait retraining: altering the fingerprint of gait. Phys. Med. Rehabil. Clin. N. Am. 27, 339–355 (2016).
2. Chan, Z. Y. et al. Gait retraining for the reduction of injury occurrence in novice distance runners: 1-year follow-up of a randomized
controlled trial. Am. J. Sports Med. 5, 036354651773627 (2017).
11
Vol.:(0123456789)
Scientific Reports | (2021) 11:4091 |
https://doi.org/10.1038/s41598-021-83538-w
www.nature.com/scientificreports/
3. Barton, C. J. et al. Running retraining to treat lower limb injuries: a mixed-methods study of current evidence synthesised with
expert opinion. Br. J. Sports Med. 50, 513–526 (2016).
4. Crowell, H. P. & Davis, I. S. Gait retraining to reduce lower extremity loading in runners. Clin. Biomech. 26, 78–83 (2011).
5. Crowell, H. P., Milner, C. E., Hamill, J. & Davis, I. S. Reducing impact loading during running with the use of real-time visual
feedback. J. Orthop. Sport. Phys. Ther. 40, 206–213 (2010).
6. Creaby, M. W. & Franettovich Smith, M. M. Retraining running gait to reduce tibial loads with clinician or accelerometry guided
feedback. J. Sci. Med. Sport. https:// doi. org/ 10. 1016/j. jsams. 2015. 05. 003 (2015).
7. Clansey, A. C., Hanlon, M., Wallace, E. S., Nevill, A. & Lake, M. J. Influence of Tibial shock feedback training on impact loading
and running economy. Med. Sci. Sports Exerc. 46, 973–981 (2014).
8. Milner, C. E., Ferber, R., Pollard, C. D., Hamill, J. & Davis, I. S. Biomechanical factors associated with tibial stress fracture in female
runners. Med. Sci. Sports Exerc. 38, 323–328 (2006).
9. Schütte, K. H., Seerden, S., Venter, R. & Vanwanseele, B. Influence of outdoor running fatigue and medial tibial stress syndrome
on accelerometer-based loading and stability. Gait Posture 59, 222–228 (2018).
10. Zifchock, R. A., Davis, I., Higginson, J., McCaw, S. & Royer, T. Side-to-side differences in overuse running injury susceptibility: a
retrospective study. Hum. Mov. Sci. 27, 888–902 (2008).
11. Laughton, C. A., Davis, I. M. & Hamill, J. Effect of strike pattern and orthotic intervention on tibial shock during running. J. Appl.
Biomech. 19, 153–168 (2003).
12. Van den Berghe, P., Six, J., Gerlo, J., Leman, M. & De Clercq, D. Validity and reliability of peak tibial accelerations as real-time
measure of impact loading during over-ground rearfoot running at different speeds. J. Biomech. 86, 238–242 (2019).
13. Wood, C. M. & Kipp, K. Use of audio biofeedback to reduce tibial impact accelerations during running. J. Biomech. 47, 1739–1741
(2014).
14. Morgan, A. M., Christopher, F. G., Malloy, P. J. & Kipp, K. Audio and visual biofeedback as methods of gait retraining to reduce
tibial acceleration upon foot strike. in American Society of Biomechanics congress 2015 1, (2015).
15. Schedel, M. et al. Interactive Sonification of Gait: Realtime BioFeedback for People with Parkinson’s Disease. in Proc. 5th Interact.
Sonification Work. 94–97 (2016).
16. Lorenzoni, V. et al. The sonic instructor: a music-based biofeedback system for improving weightlifting technique. PLoS ONE
https:// doi. org/ 10. 1371/ journ al. pone. 02209 15 (2019).
17. Terry, P. C., Karageorghis, C. I., Saha, A. M. & D’Auria, S. Effects of synchronous music on treadmill running among elite triathletes.
J. Sci. Med. Sport 15, 52–57 (2012).
18. Van Dyck, E. et al. Spontaneous entrainment of running cadence to music tempo. Sport. Med. ‑ Open 2, 15 (2015).
19. Styns, F., van Noorden, L., Moelants, D. & Leman, M. Walking on music. Hum. Mov. Sci. 26, 769–785 (2007).
20. Lesaffre, M., Vets, T., Moens, B. & Leman, M. Using auditory feedback for the rehabilitation of symmetrical body-weight distribu-
tion after ischemic stroke or brain trauma. in Proceedings of the Ninth Triennial Conference of the European Society for the Cognitive
Sciences of Music (eds. Ginsborg, J., Lamont, A., Philips, M. & Bramley, S.) 7 (2015).
21. Lorenzoni, V. et al. Design and validation of an auditory biofeedback system for modification of running parameters. J. Multimodal
User Interfaces https:// doi. org/ 10. 1007/ s12193- 018- 0283-1 (2018).
22. Moens, B. et al. Encouraging spontaneous synchronisation with D-Jogger, an adaptive music player that aligns movement and
music. PLoS ONE 9, e114234 (2014).
23. Alter, D. A. et al. Synchronized personalized music audio-playlists to improve adherence to physical activity among patients
participating in a structured exercise program: a proof-of-principle feasibility study. Sport. Med. Open 1, 1–13 (2015).
24. Leman, M. The Expressive Moment: How Interaction (with Music) Shapes Human Empowerment (MIT Press, Cambridge, 2016).
25. Maes, P.-J., Buhmann, J. & Leman, M. 3Mo: a model for music-based biofeedback. Front. Neurosci. 10, 548 (2016).
26. Giandolini, M., Horvais, N., Farges, Y., Samozino, P. & Morin, J. B. Impact reduction through long-term intervention in recreational
runners: Midfoot strike pattern versus low-drop/low-heel height footwear. Eur. J. Appl. Physiol. 113, 2077–2090 (2013).
27. Giandolini, M. et al. Impact reduction during running: efficiency of simple acute interventions in recreational runners. Eur. J.
Appl. Physiol. 113, 599–609 (2013).
28. Baggaley, M., Willy, R. W. & Meardon, S. A. Primary and secondary effects of real-time feedback to reduce vertical loading rate
during running. Scand. J. Med. Sci. Sport. 27, 501–507 (2017).
29. Willy, R. W. et al. In-field gait retraining and mobile monitoring to address running biomechanics associated with tibial stress
fracture. Scand. J. Med. Sci. Sports 26, 197–205 (2016).
30. Whittier, T. et al. The cognitive demands of gait retraining in runners: an EEG study. J. Mot. Behav. 52, 360–371 (2019).
31. Hafer, J. F., Brown, A. M., deMille, P., Hillstrom, H. J. & Garber, C. E. The effect of a cadence retraining protocol on running bio-
mechanics and efficiency: a pilot study. J. Sports Sci. 33, 724–731 (2015).
32. Hobara, H., Sato, T., Sakaguchi, M., Sato, T. & Nakazawa, K. Step frequency and lower extremity loading during running. Int. J.
Sports Med. 33, 310–313 (2012).
33. Clarke, T. E., Cooper, L. B., Hamill, C. L. & Clark, D. E. The effect of varied stride rate upon shank deceleration in running. J. Sports
Sci. 3, 41–49 (1985).
34. Napier, C., Cochrane, C. K., Taunton, J. E. & Hunt, M. A. Gait modifications to change lower extremity gait biomechanics in run-
ners: a systematic review. Br J Sport. Med 49, 1382–1388 (2015).
35. Townshend, A. D., Franettovich Smith, M. M. & Creaby, M. W. The energetic cost of gait retraining: a pilot study of the acute effect.
Phys. Ther. Sport 23, 113–117 (2017).
36. Yamato, T. P., Saragiotto, B. T. & Lopes, A. D. A consensus definition of running-related injury in recreational runners: a modified
delphi approach. J. Orthop. Sport. Phys. Ther. 45, 375–380 (2015).
37. Sheerin, K. R., Besier, T. F., Reid, D. & Hume, P. A. The reliability and variability of three-dimensional tibial acceleration during
running. in 34th International Conference of Biomechanics in Sport (2016).
38. Busa, M. A., Lim, J., Van Emmerik, R. E. A. & Hamill, J. Head and tibial acceleration as a function of stride frequency and visual
feedback during running. PLoS ONE 11, 1–13 (2016).
39. Saari, P., Burunat, I., Brattico, E. & Toiviainen, P. Decoding Musical Training from Dynamic Processing of Musical Features in the
Brain. Sci. Rep. 8, 708 (2018).
40. Moens, B. & Leman, M. Alignment strategies for the entrainment of music and movement rhythms. Ann. N. Y. Acad. Sci. 1337,
86–93 (2015).
41. Seiler, K. S. & Kjerland, G. Ø. Quantifying training intensity distribution in elite endurance athletes: Is there evidence for an
‘optimal’ distribution?. Scand. J. Med. Sci. Sport. https:// doi. org/ 10. 1111/j. 1600- 0838. 2004. 00418.x (2006).
42. Fritz, C. O., Morris, P. E. & Richler, J. J. Effect size estimates: Current use, calculations, and interpretation. J. Exp. Psychol. Gen.
141, 2–18 (2012).
43. Reenalda, J., Maartens, E., Buurke, J. H. & Gruber, A. H. Kinematics and shock attenuation during a prolonged run on the athletic
track as measured with inertial magnetic measurement units. Gait Posture 68, 155–160 (2019).
44. Silvetti, M. & Verguts, T. Reinforcement Learning, High-Level Cognition, and the Human Brain. in Neuroimaging ‑ Cognitive and
Clinical Neuroscience (2012). https:// doi. org/ 10. 5772/ 23471
45. Bowser, B. J., Fellin, R., Milner, C. E., Pohl, M. B. & Davis, I. S. Reducing impact loading in runners: a one-year follow-up. Med.
Sci. Sports Exerc. 50, 2500–2506 (2018).
12
Vol:.(1234567890)
Scientific Reports | (2021) 11:4091 |
https://doi.org/10.1038/s41598-021-83538-w
www.nature.com/scientificreports/
46. van der Worp, H., Vrielink, J. W. & Bredeweg, S. W. Do runners who suffer injuries have higher vertical ground reaction forces
than those who remain injury-free? A systematic review and meta-analysis. Br. J. Sports Med. 50, 450–457 (2016).
47. Davis, I. S., Bowser, B. J. & Mullineaux, D. R. Greater vertical impact loading in female runners with medically diagnosed injuries:
a prospective investigation. Br. J. Sports Med. 50, 887–892 (2016).
48. Nigg, B., Baltich, J., Hoerzer, S. & Enders, H. Running shoes and running injuries: mythbusting and a proposal for two new para-
digms: ‘preferred movement path’ and ‘comfort filter’. Br. J. Sports Med. 49, 1290–1294 (2015).
49. Gruber, A., Edwards, W. B. & H. Miller, R. Characteristics of the extracted impact component distinguish between prospective
running injury development and controls. in American Society of Biomechanics Annual Meeting 40 (2016).
50. Willems, T. M., Witvrouw, E., De Cock, A. & De Clercq, D. Gait-related risk factors for exercise-related lower-leg pain during shod
running. Med. Sci. Sport. Exerc. 39, 330–339 (2007).
51. Napier, C., MacLean, C. L., Maurer, J., Taunton, J. E. & Hunt, M. A. Kinetic risk factors of running-related injuries in female
recreational runners. Scand. J. Med. Sci. Sport. https:// doi. org/ 10. 1111/ sms. 13228 (2018).
52. Terry, P. C., Karageorghis, C. I., Curran, M. L., Martin, O. V. & Parsons-Smith, R. L. Effects of music in exercise and sport: a meta-
analytic review. Psychol. Bull. 146, 91–117 (2019).
Acknowledgements
The first two authors shared equal contribution to the presented work. This study was funded by the Research
Foundation–Flanders (FWO.3F0.2015.0048.01 and the Methusalem project titled ‘Expressive music interac-
tion’) and EU-EFRO-Interreg (Nano4Sports project 0217). The International Society of Biomechanics granted
a matching dissertation grant program to P.V.d.B. that supported this research. The authors thank the runners
who participated in any part of the study and Topsporthal Vlaanderen for the possibility of conducting a lab-
in-the-field test, and we acknowledge the assistance of PhD Bastiaan Breine for partial assistance during data
collection, Ing. Davy Spiessens for periodic maintenance of the accelerometers, and MSc Ella Haeck and Maxim
Gosseries for devoting a research internship to the present study. Some of the results have been presented at the
2018 congress of the American Society of Biomechanics as part of the doctoral student competition.
Author contributions
P.V.d.B., M.L., and D.D.C. conceived, designed and coordinated the study. P.V.d.B., J.G, and R.D. collected original
data. V.L. and J.S developed the custom software. P.V.d.B., J.G, and R.D., participated in data analysis. P.V.d.B,
J.G. and J.S. developed the figures. P.V.d.B initially drafted the manuscript and the other authors provided use-
ful suggestions in preparing the final manuscript. All authors reviewed the manuscript and gave approval for
publication.
Competing interests
The authors declare no competing interests.
Additional information
Supplementary Information The online version contains supplementary material available at https:// doi. org/
10. 1038/ s41598- 021- 83538-w.
Correspondence and requests for materials should be addressed to P.V.d.
Reprints and permissions information is available at www.nature.com/reprints.
Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and
institutional affiliations.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International
License, which permits use, sharing, adaptation, distribution and reproduction in any medium or
format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the
Creative Commons licence, and indicate if changes were made. The images or other third party material in this
article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the
material. If material is not included in the article’s Creative Commons licence and your intended use is not
permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from
the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/.
© The Author(s) 2021, corrected publication 2021
| Music-based biofeedback to reduce tibial shock in over-ground running: a proof-of-concept study. | 02-18-2021 | Van den Berghe, Pieter,Lorenzoni, Valerio,Derie, Rud,Six, Joren,Gerlo, Joeri,Leman, Marc,De Clercq, Dirk | eng |
PMC8505327 | Vol.:(0123456789)
1 3
https://doi.org/10.1007/s00421-021-04780-8
ORIGINAL ARTICLE
Steady‑state ̇VO2 above MLSS: evidence that critical speed better
represents maximal metabolic steady state in well‑trained runners
Rebekah J. Nixon1 · Sascha H. Kranen1 · Anni Vanhatalo1 · Andrew M. Jones1
Received: 1 May 2021 / Accepted: 26 July 2021
© The Author(s) 2021
Abstract
The metabolic boundary separating the heavy-intensity and severe-intensity exercise domains is of scientific and practical
interest but there is controversy concerning whether the maximal lactate steady state (MLSS) or critical power (synonymous
with critical speed, CS) better represents this boundary. We measured the running speeds at MLSS and CS and investigated
their ability to discriminate speeds at which ̇VO2 was stable over time from speeds at which a steady-state ̇VO2 could not be
established. Ten well-trained male distance runners completed 9–12 constant-speed treadmill tests, including 3–5 runs of up
to 30-min duration for the assessment of MLSS and at least 4 runs performed to the limit of tolerance for assessment of CS.
The running speeds at CS and MLSS were significantly different (16.4 ± 1.3 vs. 15.2 ± 0.9 km/h, respectively; P < 0.001).
Blood lactate concentration was higher and increased with time at a speed 0.5 km/h higher than MLSS compared to MLSS
(P < 0.01); however, pulmonary ̇VO2 did not change significantly between 10 and 30 min at either MLSS or MLSS + 0.5 km/h.
In contrast, ̇VO2 increased significantly over time and reached ̇VO2 max at end-exercise at a speed ~ 0.4 km/h above CS
(P < 0.05) but remained stable at a speed ~ 0.5 km/h below CS. The stability of ̇VO2 at a speed exceeding MLSS suggests
that MLSS underestimates the maximal metabolic steady state. These results indicate that CS more closely represents the
maximal metabolic steady state when the latter is appropriately defined according to the ability to stabilise pulmonary ̇VO2.
Keywords Endurance · Physiology · Oxygen uptake · Performance · Lactate · Threshold
Introduction
In a pioneering study, Whipp and Wasserman (1972)
reported that, for a given individual, exercise at different
constant work rates evoked distinctive response profiles
for pulmonary O2 uptake ( ̇VO2 ). It is now recognised that,
following the initial cardiopulmonary and fundamental
phases, ̇VO2 might: (1) reach a rapid (i.e. within ~ 3 min
of the onset of exercise) steady state; (2) reach a delayed
(within ~ 10–20 min) steady state; or (3) not attain a steady
state at all, but rather rise with time until the ̇VO2 max is
attained, with task failure occurring shortly thereafter (Jones
and Poole 2005; Whipp and Ward 1992). These charac-
teristic ̇VO2 response profiles are emblematic of exercise
intensity domains which have been termed moderate, heavy
and severe (Carter et al. 2002; Poole et al. 1988; Pringle
et al. 2003; Wilkerson et al. 2004). These three exercise
intensity domains have been considered to be partitioned by
the lactate threshold (LT) or gas exchange threshold for the
moderate-to-heavy intensity boundary, and the critical power
(CP) or maximal lactate steady state (MLSS) for the heavy-
to-severe intensity boundary (Jones et al. 2010; Whipp and
Wasserman 1972).
It has been established that the neuromuscular, metabolic,
blood acid–base and pulmonary gas exchange responses dif-
fer in the three intensity domains, resulting in differences
in the predominant causes of fatigue and the corresponding
limitations to exercise performance (Black et al. 2017; Burn-
ley et al. 2012; Vanhatalo et al. 2016). The heavy-to-severe
intensity boundary is particularly important for endurance
exercise in that it will determine whether a particular power
output or speed is, or is not, sustainable in a metabolic steady
state. Accurately discriminating the boundary between the
heavy-intensity and severe-intensity exercise domains is,
therefore, of both scientific and practical interest (Burnley
Communicated by Philip D. Chilibeck.
* Andrew M. Jones
[email protected]
1
Sport and Health Sciences, University of Exeter, St. Luke’s
Campus, Heavitree Road, Exeter EX12LU, UK
/ Published online: 5 August 2021
European Journal of Applied Physiology (2021) 121:3133–3144
1 3
and Jones 2018; Jones and Vanhatalo 2017; Vanhatalo et al.
2011).
As noted above, there are two principal approaches for
the determination of the heavy-to-severe exercise intensity
boundary. The first of these is based upon the well-known
hyperbolic relationship between power (or speed) and time,
with the asymptote of this relationship representing CP (or
critical speed, CS) and the curvature constant (W´ or D´,
for cycling and running, respectively) representing a finite
amount of work that can be done, or distance that can be
covered, above the CP or CS, respectively (Hughson et al.
1984; Monod and Scherrer 1965; Moritani et al. 1981; Poole
et al. 1988). Exercise performed above CP results in the
development of a pronounced ̇VO2 ‘slow component’ that
does not level off but which eventually drives ̇VO2 to its
maximum value; simultaneously, the intramuscular concen-
tration of phosphocreatine (PCr) falls and the intramuscular
concentrations of inorganic phosphate and H+ rise inexora-
bly, to reach values that are presumably limiting to muscle
function at the limit of tolerance (Black et al. 2017; Jones
et al. 2008; Vanhatalo et al. 2016). Muscle and blood lactate
concentrations are elevated and also display non-steady-state
behaviour above CP, but are elevated (compared to resting
values) but stable over time in the heavy-intensity domain
below CP (Black et al. 2017; Vanhatalo et al. 2016). The
second approach to determine the heavy-to-severe intensity
boundary is based solely upon the blood lactate responses
to continuous exercise and the identification of MLSS, with
the latter being defined as the highest power or speed that
can be sustained without a greater than 1 mM increase in
blood lactate concentration between 10 and 30 min of exer-
cise (Beneke and von Duvillard 1996; Dekerle et al. 2003;
Pringle and Jones 2002; Smith and Jones 2001).
CP and MLSS have often been considered to represent
the same phenomenon and the terms are frequently used
interchangeably. However, small but statistically significant
differences between CP and MLSS, with CP being higher,
are frequently reported (Dekerle et al. 2003, 2005; Mattioni
Maturana et al. 2016; Pringle and Jones 2002; Smith and
Jones 2001) and this has led to debate and over which of
them should be considered the ‘gold standard’ (Jones et al.
2019a). The rationale underpinning the specific defini-
tion of MLSS (i.e. less than a 1 mM increment in blood
[lactate] between 10 and 30 min) is obscure and the pro-
tocol from which MLSS is derived, it has been argued, is
methodologically biased towards an underestimation of the
‘true’ maximal metabolic steady state (Jones et al. 2019a).
Moreover, it has been proposed that using changes in blood
[lactate] alone as a proxy index of muscle metabolic and
respiratory homeostasis lacks precision and that the profile
of pulmonary ̇VO2 represents a more reasonable ‘global’
index of physiological conditions which are steady state
or non-steady state (Jones et al. 2019a). Nevertheless, it is
feasible that modification of the definition of MLSS via dif-
ferent permutations of permissible blood [lactate] increment
and the time over which lactate accumulates might result in
closer agreement between MLSS and CP and enable better
approximation of CP from measurements of blood [lactate]
derived from submaximal exercise tests.
The purpose of the present study was to investigate, in
well-trained runners, which of CS and MLSS better repre-
sent the highest speed at which ̇VO2 can be stabilised (i.e.
maximal metabolic steady state). We hypothesised that: (1)
CS would be significantly higher than MLSS; (2) CS would
more closely approximate the maximal metabolic steady
state, as defined via the behaviour of ̇VO2 ; and (3) modify-
ing the definition of MLSS would eliminate the difference
between CS and MLSS.
Methods
Participants
Ten well-trained, competitive male athletes (runners
n = 7, triathletes n = 3; mean ± SD age 22.8 ± 4.8 years,
height 1.80 ± 0.05 m, body mass 73.7 ± 5.8 kg, ̇VO2 peak
63.0 ± 4.0 mL/kg/min) volunteered to participate in this
study. The participants gave written informed consent and
completed and signed a PAR-Q form as a declaration of
their eligibility to take part in the study. Participants had no
known medical conditions that would inhibit their ability to
perform strenuous exercise or cause them harm while doing
so. The study adhered to the principles of the Declaration
of Helsinki (2013) and was approved by the University of
Exeter Sport and Health Sciences Ethics Committee.
Experimental design and general procedures
This experiment was designed to investigate the relationship
between running speeds at MLSS and CS and to explore
which of MLSS and CS provided the better approxima-
tion of the maximal metabolic steady state, as defined by
the ability to separate running speeds at which ̇VO2 can, or
cannot, be stabilised. To this end, following an initial step
incremental treadmill test, participants completed a series
of constant-speed treadmill runs of up to 30-min duration
or to the limit of tolerance for the assessment of MLSS (3–5
runs) and CS (4–6 runs). Tests for determination of MLSS
and CS were numbered consecutively from 1. The order of
tests was randomised for each participant in a counterbal-
anced manner, with five participants starting with a CS test
and the other five starting with a MLSS test. The final lab
visit involved participants running just above their CS to the
limit of tolerance.
European Journal of Applied Physiology (2021) 121:3133–3144
3134
1 3
The study had a single blind, counterbalanced and ran-
domised design and included 9–12 lab visits. Participants
were instructed to avoid strenuous exercise for 24 h before,
and caffeine and alcohol consumption for 12 h before, lab
visits. Testing was completed at the same time of day (± 2 h)
for each participant to minimise any influence of circadian
rhythm. All participants were offered a familiarisation visit.
Testing was completed within 3–6 weeks, with 2–4 tests per
week and at least 24 h between tests. The treadmill (Wood-
way PPS 55 Sport, Woodway GmbH, Weil am Rhein, Ger-
many) was set to 1% incline for all tests (Jones and Doust
1996). Pulmonary gas exchange was measured during all
tests, with participants wearing a facemask connected to a
calibrated Jaeger Oxycon Pro breath-by-breath ergospirom-
etry system (VIASYS Healthcare GmbH, Hoechberg, Ger-
many). Heart rate (HR) was also recorded during all tests
(Polar T31 Heart rate strap, Finland). Blood [lactate] was
measured during the incremental test and the MLSS deter-
mination tests. Blood was collected from a fingertip using
a single use lancet. The first drop of blood was wiped using
a clean tissue then the blood sample was collected in a cap-
illary tube and analysed enzymatically (YSI 2500 Lactate
Analyser, YSI, Letchworth, UK).
Before all testing sessions, participants completed a
standardised warm up which involved running for 5 min
at 10 km/h and the opportunity to stretch. For CS predic-
tion tests > 18 km/h, participants also completed a 30–60 s
‘strider’ at a speed above 10 km/h but below their LT. Time
was recorded for all tests to the nearest second. Timing
(Traceable Timer, Fisher Scientific, Loughborough, UK)
started as participants let go of the treadmill handrails to
transition to the moving belt and stopped when they stepped
to the sides of the treadmill to end the test.
Initial incremental test
On their first visit to the laboratory, the participants com-
pleted a multi-stage incremental step test to the limit of tol-
erance. The test commenced with 3 min of standing rest
during which pulmonary gas exchange, HR and blood [lac-
tate] were measured. The treadmill test then commenced
at a speed of 10–12 km/h, depending on the ability of the
participant. Every 3 min, the participant stopped running by
stepping to the side of the treadmill belt for 30 s to facili-
tate the collection of a fingertip blood sample. Speed was
increased by 1 km/h every 3 min until the participant could
not complete a stage or declined the opportunity to start
a new stage. The final completed stage was defined as the
peak speed attained. ̇VO2 peak was defined as the highest 30 s
mean value recorded during the test. The running speeds at
LT (i.e. the first increase in blood [lactate] above baseline
values of ~ 1 mM) and lactate turnpoint (LTP; i.e. a sudden
and sustained second threshold increase in blood [lactate]
at ~ 2.5–4 mM) were determined visually from plots of blood
[lactate] against running speed by two experienced research-
ers (Jones 2007; Jones et al. 2021).
Determination of running speed at maximal lactate
steady state
Participants ran at a constant (but different) speed for up
to 30 min on 3–5 occasions (3 runs for four participants,
4 runs for five participants and 5 runs for one participant).
Every 5 min they stopped running and stepped to the sides
of the treadmill belt for 15 s to enable the collection of a
fingertip blood sample. The selection of test speeds was
informed by the speed at LTP measured in the incremental
test. Speeds used in the tests were subsequently varied by
0.5 km/h and were continued until there was sufficient data
to determine MLSS, which was defined as the highest speed
which resulted in an increase of blood [lactate] of less than
1 mM between 10 and 30 min (Beneke and von Duvillard
1996; Jones and Doust 1998). If the increase in blood [lac-
tate] was ± 0.05 mM above or below 1 mM (an event that
occurred in two participants), the test was repeated and the
mean of the blood [lactate] at each time point during the two
tests was calculated and used in the determination of MLSS.
In addition to the conventional definition of MLSS
outlined above, different permutations of the criteria for
the change in blood [lactate] (i.e., < 1.0 mM, < 1.5 mM
and < 2.0 mM) and the time window over which changes
in blood [lactate] were measured (i.e. 5–10 min, 5–15 min,
5–20 min, 5–25 min, 5–30 min, 10–15 min, 10–20 min,
10–25 min, 10–30 min, 15–20 min, 15–25 min, 15–30 min,
20–25 min and 20–30 min), were applied to provide a modi-
fied MLSS assessment.
Determination of critical speed
The participants completed constant-speed runs to the limit
of tolerance at speeds corresponding to 90%, 95%, 100%
and 105% of the peak speed attained in the incremental test
such that test duration was ~ 2–15 min. Pulmonary ̇VO2 was
monitored throughout the tests to assess whether end-test
values exceeded 95% of the peak value determined in the
step incremental test. Test durations longer than 15 min were
only included in the CS determination if ̇VO2 was greater
than 95% of the respective peak value measured in the incre-
mental test. CS was subsequently calculated from two lin-
ear regression models, the distance–time and speed–1/time
models, and the non-linear hyperbolic speed–time model.
The standard errors (SE) associated with the CS and Dʹ
estimates for each model were calculated using regression
analysis. The coefficient of variation (CoV) for each CS and
Dʹ estimate was then calculated by expressing the SE as %
of the parameter estimate. For a model to be accepted, the
European Journal of Applied Physiology (2021) 121:3133–3144
3135
1 3
coefficient of variation (CoV) for the mathematical fit had
to be < 5% for CS and < 10% for D´. The output from the
model with the lowest error was used in subsequent analysis
(Black et al. 2015). If the CoV was too high after the initial
four tests, further tests were completed until the CoV criteria
were met. In total, five participants completed four tests,
two participants completed five tests and three participants
completed six tests.
Following the assessment of CS and calculation of the
95% confidence intervals (CI) surrounding the CS estimate
for each individual, the participants completed a final test
just above CS (i.e. CS+ test) in which they ran to the limit
of tolerance at the speed representing the upper bound of
the 95% CI. The highest running speed that was used in the
MLSS assessment but was below the lower bound of the
95% CI for the CS estimate for each participant was identi-
fied and defined as CS−.
Statistical analysis
Analyses were performed using IBM SPSS Statistics 26.0
(Chicago, IL, USA). A two-tailed paired Students t test was
used to analyse the difference between the running speeds
at MLSS and CS. One-way repeated measures ANOVA
were used to assess differences between peak values of ̇VO2
attained during the CS determination trials and the incre-
mental test, and also between the running speeds at conven-
tional MLSS, modified MLSS and CS. Two-way ANOVA
with repeated measures across condition and time (5, 10,
15, 20, 25 and 30 min) was used to assess differences in the
blood [lactate] response for running speeds at MLSS and
the speed 0.5 km/h above MLSS (termed MLSS+). Two-
way ANOVA with repeated measures across condition and
time (10 and 30 min < CS and 5 min and end-exercise > CS)
was also used to assess differences in the ̇VO2 response for
running speeds at MLSS, MLSS+, CS− and CS+. When
sphericity was violated, the significance of F-ratios was
adjusted using the Greenhouse–Geisser procedure and sig-
nificant interaction and main effects were followed up using
LSD post hoc tests. Linear regression analysis using Pearson
product moment was carried out to determine the relation-
ship between CS and D´ and the relationship between the
running speeds at conventional MLSS, modified MLSS, and
CS. Significance was set at P < 0.05 and results are reported
as mean ± SD.
Results
The LT and LTP occurred at 14.5 ± 1.2 and 16.8 ± 1.0 km/h,
respectively, and the peak speed attained in the incremental
test was 19.6 ± 1.3 km/h. The ̇VO2 peak achieved during the
incremental test was 4.65 ± 0.47 L/min or 63.0 ± 4.0 mL/kg/
min.
The times to exhaustion (s) in the CS prediction trials
performed at 90%, 95%, 100% and 105% of the peak speed
attained in the incremental treadmill test were 801 ± 219,
500 ± 146, 388 ± 112 and 185 ± 33 s. The mean ̇VO2 peak
attained during the CS trials (4.92 ± 0.58 L/min) was not
significantly different from the peak value achieved during
the step incremental test (P = 0.28). The CS and D´ were
16.4 ± 1.3 km/h and 216 ± 79 m, respectively, and the CoV
was 0.8 ± 0.7% for CS and 7 ± 3% for D´.
Following the calculation of CS and the associated 95%
CI, participants completed a final test to the limit of tolerance
at a speed just (~ 2.4%) above CS (i.e. at 16.8 ± 1.3 km/h;
CS+). The time to the limit of tolerance at this speed was
17.0 ± 4.6 min and there was a strong positive correlation
with D´ (r = 0.82, P < 0.01).
Comparison of critical speed and running speed
at maximal lactate steady state
The determination of MLSS is shown for a representa-
tive participant in Fig. 1 and the determination of CS is
shown in Fig. 2. There was a significant difference between
CS and conventionally determined MLSS (16.4 ± 1.3 vs
15.2 ± 1.0 km/h, respectively; P < 0.001). Different permu-
tations of the permitted blood [lactate] increase, and the time
window over which [lactate] increased, resulted in different
estimates for speed at MLSS (Fig. 3). Of all of the 45 per-
mutations of blood [lactate] increment and time window that
were considered, only the criterion of a < 2.0 mM increase in
blood [lactate] between 10 and 20 min produced an MLSS
value (15.9 ± 0.9 km/h) that was not significantly different
from CS (Fig. 3).
The conventional MLSS was significantly correlated
with modified MLSS (r = 0.80, P < 0.01) and CS (r = 0.96,
P < 0.001) and the modified MLSS was significantly cor-
related with CS (r = 0.80, P < 0.01).
Behaviour of blood [lactate] and oxygen uptake
in the proximity of MLSS and CS
There was a significant main effect by time (F = 13.7,
P = 0.005) and by condition (F = 38.1, P < 0.001), and a
significant interaction effect (F = 14.1, P < 0.001) on blood
[lactate] across MLSS and MLSS+ trials. Post hoc tests
showed significant differences in blood [lactate] between
MLSS and MLSS+ trials at 10, 15, 20, 25 and 30 min
(P < 0.05 for all; Fig. 4A). The change in blood [lactate]
between 10 and 30 min was significantly greater during the
run at MLSS+ compared to the run at MLSS (P < 0.001;
Fig. 4A). During the MLSS run, blood [lactate] remained
stable (2.1 ± 1.0, 2.3 ± 0.6 and 2.6 ± 1.0 mM at 10, 20 and
European Journal of Applied Physiology (2021) 121:3133–3144
3136
1 3
30 min, respectively) whereas, during the MLSS+ run,
blood [lactate] increased with time (2.7 ± 1.0, 3.4 ± 1.2 and
4.4 ± 1.3 mM at 10, 20 and 30 min, respectively).
There were significant main effects by time (F = 15.9,
P = 0.003) and by condition (F = 21.5, P < 0.001), and a
significant interaction effect (F = 4.3, P = 0.029) on ̇VO2
for running speeds at MLSS, MLSS+, CS− and CS+. Post
hoc analysis revealed that ̇VO2 during MLSS was lower
than in MLSS+, CS− and CS+ (P = 0.001, P = 0.004 and
P < 0.001, respectively), and ̇VO2 during CS+ was greater
than in MLSS, MLSS+ and CS− (P < 0.001, P = 0.012 and
P < 0.001, respectively), while there was no difference
between MLSS+ and CS− (P = 0.38). ̇VO2 did not change
between 10 min and end-exercise during the runs at MLSS,
MLSS+ or CS− (P = 0.10, P = 0.26, P = 0.78, respectively)
(Fig. 4B and C). During the run at CS+, however, ̇VO2
increased significantly between 5 min and the limit of tol-
erance (P < 0.001; Fig. 4C). The ̇VO2 peak measured at the
limit of tolerance in the CS+ run (4.67 ± 0.41 L/min) was
not significantly different from ̇VO2 peak measured in the
step incremental test; however, the end-exercise ̇VO2 in the
MLSS, MLSS+ and CS− runs were all significantly lower
than ̇VO2 peak measured in the step incremental test (P < 0.05;
Fig. 4B and C).
Discussion
The principal findings of the present study are that: (1) CS
is higher than the speed at MLSS; (2) running at a speed
0.5 km/h above MLSS (i.e. MLSS+) results in a significant
increase in blood [lactate] between 10 and 30 min, but no
significant change in ̇VO2 over the same time frame; (3) run-
ning at a speed ~ 0.4 km/h above CS (i.e. CS+), but not at a
speed ~ 0.5 km/h below CS (i.e. CS−), results in a significant
increase in ̇VO2 over time with peak ̇VO2 attained at the limit
of tolerance; and (4) in the current data set, defining the
MLSS as a < 2 mM increment in blood [lactate] between 10
and 20 min, as opposed to a < 1 mM increment between 10
and 30 min as per the conventional definition, eliminates the
difference between MLSS and CS. These findings are con-
sistent with our experimental hypotheses. We interpret the
results to indicate that, while MLSS differentiates running
speeds for which the blood [lactate] response is steady state
vs. non-steady state, it underestimates the maximal meta-
bolic steady state as represented by the ability, or inability, to
stabilise ̇VO2 during exercise. In contrast, running at a speed
just above, but not just below, CS results in a rising ̇VO2
profile until the limit of tolerance is reached, indicating that
CS provides a more appropriate representation of maximal
metabolic steady state.
Consistent with our hypothesis, CS was significantly
higher than the speed at MLSS. This finding is in agreement
with several other studies which have found CS or CP to be
higher than MLSS, both for running and cycling (Dekerle
et al. 2003, 2005; Mattioni Maturana et al. 2016; Pringle
and Jones 2002). In the present study, CS was ~ 8% higher
than MLSS, which is broadly consistent with previous com-
parisons: for example, 4% in Smith and Jones (2001), 9% in
Pringle and Jones (2002), 16% in Dekerke et al. (2003), 5%
in Dekerke et al. (2005) and 1% in Keir et al (2015). While
differences in experimental protocol including the number
and duration of prediction trials, the sensitivity of MLSS
determination (which is a function of the power or speed
increments between trials), and the mathematical models
used to calculate CP or CS may explain some of the discrep-
ancy (Bishop et al. 1998; Black et al. 2015; Mattioni Matu-
rana et al. 2018), it is now clear that while CP and MLSS
Fig. 1 Maximal lactate steady-
state (MLSS) assessment in a
representative individual. MLSS
was identified as the highest
speed where the increase in
blood [lactate] did not exceed
1 mM between 10 and 30 min
0
5
10
15
20
25
30
0.0
0.5
1.0
1.5
2.0
2.5
15.0 km/h
15.5 km/h
16.0 km/h
Time (min)
Blood [lactate] (mmol/L)
MLSS
European Journal of Applied Physiology (2021) 121:3133–3144
3137
1 3
have historically been considered to represent broadly the
same phenomenon, there is limited agreement in practice.
The tendency for CP to be higher than MLSS has led to
the interpretation that CP overestimates the maximal meta-
bolic steady state, with the assumption that MLSS represents
the ‘gold standard’ and that a blood [lactate] steady state
reflects a ̇VO2 steady state (e.g. Iannetta et al. 2018; Pringle
and Jones 2002). However, the physiological rationale for
the accepted definition of MLSS (i.e. the highest power or
speed at which blood [lactate] does not increase by more
than 1 mM between 10 and 30 min of exercise, equivalent to
a 0.05 mM increment in blood [lactate] per min) is obscure
and apparently arbitrary (Jones et al. 2019a). It has been
pointed out that reliance on blood [lactate] alone as a proxy
for the existence of muscle metabolic and systemic homeo-
stasis is hazardous; that absolute blood [lactate] is influenced
by exercise-induced haemoconcentration and modifications
to substrate metabolism (Tanaka 1991); and that human,
technical and instrument error in the collection and analy-
sis of capillary blood samples for [lactate] (Morton et al.
2012; Tanner et al. 2010) at just two discrete time points
(10 and 30 min), along with poor day-to-day reproducibility
Fig. 2 Pulmonary ̇VO2 responses to four severe-intensity prediction
trials at speeds ranging from 17.0 to 21.0 km/h (Panel A), and the
critical speed (CS) and D´ estimation in a representative participant
using distance–time (Panel B), speed–1/time (Panel C) and speed–
time models (Panel D). The model with the lowest sum of coefficients
of variation (CoV) for CS and D´ for each participant was selected
for analysis. The dashed line indicates the ̇VO2 peak measured in the
initial step incremental test in panel A, and the speed-asymptote (CS)
in panel D
European Journal of Applied Physiology (2021) 121:3133–3144
3138
1 3
of [lactate] during MLSS assessment (Hauser et al. 2013),
could result in either false positives or false negatives (Jones
et al. 2019a). Moreover, the use of discrete powers or speeds
in the MLSS assessment procedure will inevitably result in
an underestimation of the actual maximal metabolic steady
state (Jones et al. 2019a), the extent of which will depend
on the sensitivity of the measurements, with tests typically
differing by 20–30 W for cycling and 0.5 or 1.0 km/h for
running.
Methodological strengths of the present study included
that: at least three (and frequently 4 or 5) 30-min trials
were used in the assessment of MLSS; the trials were
separated by relatively small (0.5 km/h) running speed
increments; and, where the increase in blood [lactate]
was within 0.05 mM of meeting the criterion of a 1 mM
increase (an event that occurred in two participants), the
test was repeated and the mean response was used in
subsequent analysis. These elements of the study design
provide a high level of confidence in the precision of
MLSS determination. An important finding in the pre-
sent study was that, while MLSS partitioned a running
speed at which blood [lactate] did not change between 10
and 30 min from a running speed at which blood [lactate]
increased significantly over the same time frame, it did not
separate steady state from non-steady-state ̇VO2 responses.
Specifically, at the speed immediately (0.5 km/h) above
that which was identified as representing MLSS, ̇VO2 did
not change significantly between 10 and 30 min. Other
recent studies also indicate that MLSS does not reflect the
maximum power or speed at which ̇VO2 can be stabilised.
For example, Bräuer and Smekal (2020) measured MLSS
from 4 to 6 30-min cycle exercise tests in 45 participants
and reported stable ̇VO2 over the last 10 min of exercise at
both MLSS and at a power output above MLSS. Similarly,
Iannetta et al (2018) found that when participants cycled at
10 W above the power output established as representing
MLSS, a ̇VO2 steady state was manifest. Such results are
insightful because it is known that, in the steady state, pul-
monary ̇VO2 closely reflects skeletal muscle ̇VO2 (Grassi
et al. 1996; Krustrup et al. 2009). Moreover, pulmonary
̇VO2 and intramuscular [PCr] profiles are closely related
both when steady states can be attained and when slow
components in the responses are manifest (Rossiter et al.
2002). Collectively, these findings indicate that MLSS,
as conventionally defined, does not represent the maxi-
mal metabolic steady state, which is more appropriately
defined in relation to the ability to stabilise pulmonary ̇VO2
and thus skeletal muscle ̇VO2 and [PCr] (Grassi et al. 1996;
Rossiter et al. 2002).
In contrast, when the athletes in the present study ran
at a speed just above CS (CS+, 16.8 ± 1.3 km/h, calculated
according to the 95% CI surrounding the estimate of CS),
̇VO2 increased significantly beyond 5 min, and the end-
exercise ̇VO2 was not different from the ̇VO2 peak measured
Speed (km/h)
10-30 min
10-20 min
15-30 min
20-30 min
10-30 min
10-20 min
15-30 min
20-30 min
10-30 min
10-20 min
15-30 min
20-30 min
>1.0 mM
>1.5 mM
>2.0 mM
*
*
*
*
*
*
*
*
*
*
*
Fig. 3 Different permutations of the maximal lactate steady state
(MLSS) definition including < 1.0, < 1.5 and < 2.0 mM blood
[lactate] increase, over time intervals of 10–30 min, 10–20 min,
15–30 min and 20–30 min, and the critical speed (dashed bar).
Note that, for clarity, not all of the MLSS permutations are shown.
All MLSS permutations were lower than CS (*P < 0.05), except
for the < 2.0 mM increase in blood [lactate] between 10 and 20 min
(MLSS = 15.9 ± 0.9 km/h)
European Journal of Applied Physiology (2021) 121:3133–3144
3139
1 3
Fig. 4 Panel A: blood [lactate]
responses at the established
maximal lactate steady state
(MLSS, 15.2 ± 0.9 km/h)
and at the speed immedi-
ately above MLSS (MLSS+;
15.7 ± 0.9 km/h). Panel B:
pulmonary ̇VO2 responses
during exercise at MLSS and
MLSS+; note that ̇VO2 did not
change significantly between
10 and 30 min. Panel C: ̇VO2 at
the speeds immediately below
CS (CS−, 15.9 ± 0.9 km/h)
and above CS (CS+,
16.8 ± 1.3 km/h). ̇VO2 did not
change significantly between 10
and 30 min at CS− but increased
between 5 min and end-exercise
for CS+ (P < 0.05). The dashed
line in panels B and C indicates
the group mean ̇VO2 peak meas-
ured in the step incremental
test. Error bars indicate standard
deviations. *End-exercise ̇VO2
significantly different from
̇VO2 peak measured in the step
incremental test (P < 0.05)
European Journal of Applied Physiology (2021) 121:3133–3144
3140
1 3
in the maximal incremental test. When the highest speed
below CS the athletes ran at (CS−, 15.9 ± 0.9 km/h) was
considered, ̇VO2 was not significantly different between 10
and 30 min. Therefore, when the athletes ran at a speed
that was ~ 6 s per km slower than CS, ̇VO2 was in steady
state and the prescribed 30 min of exercise was completed,
whereas when the athletes ran at a speed that was ~ 5 s
per km faster than CS, a ̇VO2 steady state could not be
achieved and exercise tolerance was limited to ~ 17 min,
indicative of exercise within the severe-intensity domain
(Black et al. 2017; Poole et al. 1988).
These data indicate that CS provides a rather precise
demarcation of the highest running speed at which ̇VO2 can
be stabilised. This observation is consistent with several pre-
vious studies which indicate that CP or CS is the metabolic
threshold which partitions severe-intensity exercise, which,
by definition, is characterised by an inexorable increase in
̇VO2 to its peak value at the limit of tolerance, from heavy-
intensity exercise, during which a ̇VO2 steady state can still
be achieved (Jones et al. 2010; Poole et al. 2016; Whipp
1994). It has been established from muscle biopsy studies
that these characteristic ̇VO2 profiles during exercise per-
formed above and below CP are associated with correspond-
ing steady-state or non-steady-state responses in skeletal
muscle PCr and lactate concentrations (Black et al. 2017;
Vanhatalo et al. 2016). These findings are reinforced by non-
invasive 31P-magnetic resonance spectroscopy assessment of
the skeletal muscle metabolic responses, which demonstrate
striking differences in the profiles of PCr, inorganic phos-
phate and pH for exercise performed just above, compared
to just below, CP (Jones et al. 2008). These differences in
the rates of substrate utilisation and metabolite accumula-
tion likely underpin observations that the rate and nature
of neuromuscular fatigue development also differ according
to the intensity of the exercise task relative to CP (Black
et al. 2017; Burnley et al. 2012; Dinyer et al. 2020; Pethick
et al. 2020). Finally, it is pertinent to note that simultaneous
assessment of the responses of muscle [lactate] and blood
[lactate] during heavy-intensity and severe-intensity exercise
reveal that the former may be stable while the latter rises
(Jones et al. 2019b), suggesting differences in the dynamics
of lactate accumulation in the muscle and blood compart-
ments. These observations clearly indicate that a maximum
blood [lactate] steady state will likely underestimate the
maximal metabolic steady state as determined by muscle
[lactate], as well as the responses of other muscle ions and
metabolites, and ̇VO2.
The evaluation of CP or CS is not without its challenges,
ideally requiring 3–5 maximal efforts on separate days,
although this burden can be alleviated in athletes through
the use of recent training or competition data (Jones and
Vanhatalo 2017; Karsten et al. 2015; Smyth and Muniz-
Pumares 2020). In some situations, estimating CP or CS
from submaximal exercise tests may be considered prefer-
able to direct assessment. While it is clear from both the
present study and from earlier studies that the conventional
protocol and criteria for MLSS assessment underestimates
the maximal metabolic steady state, it is possible that adjust-
ments to these factors might enable a closer approximation
of CP or CS. In the present study, we calculated the running
speed at MLSS using a variety of permutations of absolute
increments in blood [lactate] (e.g. 1.0, 1.5 and 2.0 mM) and
the time frame over which such increments were consid-
ered. Of these permutations, we found that modifying the
criteria to a 2 mM increment in blood [lactate] between
10 and 20 min increased the group mean running speed at
MLSS from 15.2 to 15.9 km/h and eliminated the difference
between MLSS and CS. This approach does not, however,
circumvent other limitations to relying solely on measure-
ments of blood [lactate] for the assessment of the maximal
metabolic steady state and might be considered to be just
as obscure and arbitrary as the conventional definition of
MLSS. At the present time, therefore, we favour direct
assessment of CP or CS if precision is required in scientific
studies or for training prescription.
By definition, exercise in the severe-intensity domain
should result in the attainment of ̇VO2 peak at or shortly before
the limit of tolerance is reached (Hill et al. 2002; Jones et al.
2010; Poole et al. 1988; Whipp 1994). It should be appreci-
ated, however, that CP and CS are estimated mathematically
from several prediction trials and there will inevitably be
some error, both computational and biological (e.g. day-
to-day variability), surrounding the estimates (Black et al.
2015; Mattioni Maturana et al. 2018). For this reason, asking
participants to exercise to the limit of tolerance exactly at the
computed CP or CS can result in wide variability in both the
physiological responses and the time to the limit of tolerance
due to some participants being below and others being above
the CP or CS (Pethick et al. 2020; see Jones et al. 2019a
for review). Indeed, the notion of exercising at the CP (or
CS) is vacuous because the asymptote of the power–time
relationship represents the power that lies exactly between
those powers at which W´ is utilised and those powers at
which it is not; that is, it defines the threshold separating the
heavy-intensity and severe-intensity domains and the inher-
ent steady-state or non-steady-state physiological behaviour
that defines those domains; and therefore, it is erroneous
to define CP as the highest power at which steady-state
responses are observed. In the present study, we employed
several approaches to minimise the error surrounding the CS
estimate including: having the athletes complete at least 4
and up to 6 prediction trials; ensuring that the athletes ran to
the limit of tolerance during all prediction trials, as validated
by there being no significant difference in ̇VO2 peak achieved
in the prediction trials compared to the maximal incremental
test; and applying all three standard mathematical models
European Journal of Applied Physiology (2021) 121:3133–3144
3141
1 3
and choosing the output from the model with the least error
for each individual (Black et al. 2015). Together, these
approaches resulted in CoV that were appreciably lower for
both CS (0.8 ± 0.7%) and D´ (7 ± 3%) than the degree of
error which has been suggested to be acceptable (< 5% for
CS and < 10% for D´; Hill 1993).
The experimental procedures employed in the present
study also ensured that the 95% CI surrounding CS were
relatively narrow (i.e. group mean of ± 0.4 km/h). Our study
participants, therefore, ran very close to, but very slightly
above, their CS to the limit of tolerance as a validation that
CS represents the heavy-to-severe exercise intensity bound-
ary. It should be acknowledged that, while the group mean
CS was estimated to be 16.4 km/h, when the 95% CI is taken
into account, the ‘actual’ CS could have occurred anywhere
between 16.0 and 16.7 km/h. We took appropriate measures
to minimise errors arising from biological factors and math-
ematical modelling, but it is important to note that it is not
possible to entirely eliminate the error margin surrounding
any physiological threshold estimate (Pethick et al. 2020). It
should also be appreciated that this range of speeds within
which the CS resides represents an error of only ± 4–5 s per
km (~ 2%).
The time to the limit of tolerance at CS+ (17.0 ± 4.6 min)
was closely correlated with the athletes’ D´ (r = 0.82). This
indicates that when an exercise task is relativized to ath-
letes’ CS values, then the energetic reserve or work capacity
above CS becomes an important factor determining exercise
tolerance. Within the limitations of the present study (i.e.
the error margin surrounding the estimation of CS), these
results also reveal that the longest an athlete can run at a con-
stant speed and still attain ̇VO2 peak is approximately 17 min.
Therefore, assuming an even pace is employed throughout
the race to minimise the time taken to complete the distance
(Fukuba and Whipp, 1999), it appears likely that athletes
will attain ̇VO2 peak during a 5000 m race (since this will
be at the lower end of the severe-intensity domain) but not
during a 10,000 m race (which is positioned at the upper end
of the heavy-intensity domain). These results highlight an
important conceptual issue: it is inappropriate to consider
peri-CS (or CP) exercise to be ‘fatigueless’, or to be sustain-
able indefinitely or for some arbitrary time period such as
60 min; indeed, this might be considered a misinterpreta-
tion of the original descriptions of the concept (Monod and
Scherrer 1965). Rather, contemporary understanding is (or,
at least, in our view, should be) that CS separates exercise
domains within which: (1) physiological (including muscle
metabolic and cardiorespiratory) responses are differenti-
ated by steady-state vs. non-steady-state behaviour; and (2)
the predominant determinants of fatigue are altered, with
exercise tolerance > CS being predictable as a function of
CS and D´.
In conclusion, this study affirms that CS occurs at a
higher speed than MLSS in well-trained runners. An impor-
tant novel finding was that a ̇VO2 steady state was elicited
when the athletes ran at a speed which was above MLSS but
below CS, whereas a ̇VO2 steady state could not be attained
when they ran at a speed which was just above CS. These
results indicate that CS, rather than MLSS, provides a better
representation of the maximal metabolic steady state.
Author contributions AMJ conceived and designed the research. RJN
and SHK conducted the experiments. RJN and AV analysed the data.
RJN and AMJ wrote the manuscript. All the authors read and approved
the manuscript.
Declarations
Conflict of interests The authors have no conflicts of interest to de-
clare.
Open Access This article is licensed under a Creative Commons Attri-
bution 4.0 International License, which permits use, sharing, adapta-
tion, distribution and reproduction in any medium or format, as long
as you give appropriate credit to the original author(s) and the source,
provide a link to the Creative Commons licence, and indicate if changes
were made. The images or other third party material in this article are
included in the article’s Creative Commons licence, unless indicated
otherwise in a credit line to the material. If material is not included in
the article’s Creative Commons licence and your intended use is not
permitted by statutory regulation or exceeds the permitted use, you will
need to obtain permission directly from the copyright holder. To view a
copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/.
References
Beneke R, von Duvillard SP (1996) Determination of maximal lactate
steady state response in selected sports events. Med Sci Sports
Exerc 28(2):241–246
Bishop D, Jenkins DG, Howard A (1998) The critical power function is
dependent on the duration of the predictive exercise tests chosen.
Int J Sports Med 19(2):125–129
Black MI, Jones AM, Bailey SJ, Vanhatalo A (2015) Self-pacing
increases critical power and improves performance during severe-
intensity exercise. Appl Physiol Nutr Metab 40(7):662–670
Black MI, Jones AM, Blackwell JR, Bailey SJ, Wylie LJ, McDonagh
ST, Thompson C, Kelly J, Sumners P, Mileva KN, Bowtell JL,
Vanhatalo A (2017) Muscle metabolic and neuromuscular deter-
minants of fatigue during cycling in different exercise intensity
domains. J Appl Physiol 122(3):446–459
Bräuer EK, Smekal G (2020) VO2 steady state at and just above
maximum lactate steady state intensity. Int J Sports Med
41(09):574–581
Burnley M, Jones AM (2018) Power-duration relationship: physiol-
ogy, fatigue, and the limits of human performance. Eur J Sport
Sci 18(1):1–12
Burnley M, Vanhatalo A, Jones AM (2012) Distinct profiles of neuro-
muscular fatigue during muscle contractions below and above the
critical torque in humans. J Appl Physiol 113(2):215–223
European Journal of Applied Physiology (2021) 121:3133–3144
3142
1 3
Carter H, Pringle JS, Jones AM, Doust JH (2002) Oxygen uptake kinet-
ics during treadmill running across exercise intensity domains.
Eur J Appl Physiol 86(4):347–354
Dekerle J, Baron B, Dupont L, Vanvelcenaher J, Pelayo P (2003) Maxi-
mal lactate steady state, respiratory compensation threshold and
critical power. Eur J Appl Physiol 89(3):281–288
Dekerle J, Pelayo P, Clipet B, Depretz S, Lefevre T, Sidney M (2005)
Critical swimming speed does not represent the speed at maximal
lactate steady state. Int J Sports Med 26(7):524–530
Dinyer TK, Byrd MT, Succi PJ, Bergstrom HC (2020) The time course
of changes in neuromuscular responses during the performance of
leg extension repetitions to failure below and above critical resist-
ance in women. J Strength Cond Res. https:// doi. org/ 10. 1519/ JSC.
00000 00000 003529
Fukuba Y, Whipp BJ (1999) A metabolic limit on the ability to make up
for lost time in endurance events. J Appl Physiol 87(2):853–861
Grassi B, Poole DC, Richardson RS, Knight DR, Erickson BK, Wagner
PD (1996) Muscle O2 uptake kinetics in humans: implications for
metabolic control. J Appl Physiol 80(3):988–998
Hauser T, Bartsch D, Baumgärtel L, Schulz H (2013) Reliability of
maximal lactate-steady-state. Int J Sports Med 34(3):196–199
Hill DW (1993) The critical power concept. Sports Med 16(4):237–254
Hill DW, Poole DC, Smith JC (2002) The relationship between
power and the time to achieve VO2max. Med Sci Sports Exerc
34(4):709–714
Hughson RL, Orok CJ, Staudt LE (1984) A high velocity treadmill
running test to assess endurance running potential. Int J Sports
Med 5(1):23–25
Iannetta D, Inglis EC, Fullerton C, Passfield L, Murias JM (2018) Met-
abolic and performance-related consequences of exercising at and
slightly above MLSS. Scand J Med Sci Sports 28(12):2481–2493
Jones AM (2007) Middle and long distance running. In: Winter EM,
Jones AM, Davison RRC, Bromley P, Mercer T (eds) Sport and
exercise science testing guidelines: the British association of sport
and exercise sciences guide. Routledge, London, pp 147–154
Jones AM, Doust JH (1996) A 1% treadmill grade most accurately
reflects the energetic cost of outdoor running. J Sports Sci
14(4):321–327
Jones AM, Doust JH (1998) The validity of the lactate minimum test
for determination of the maximal lactate steady state. Med Sci
Sports Exerc 30(8):1304–1313
Jones AM, Poole DC (2005) Oxygen uptake dynamics: from muscle to
mouth—an introduction to the symposium. Med Sci Sports Exerc
37(9):1542–1550
Jones AM, Vanhatalo A (2017) The “critical power” concept: applica-
tions to sports performance with a focus on intermittent high-
intensity exercise. Sports Med 47(Suppl 1):65–78
Jones AM, Wilkerson DP, DiMenna F, Fulford J, Poole DC (2008)
Muscle metabolic responses to exercise above and below the
“critical power” assessed using 31P-MRS. Am J Physiol Regul
Integr Comp Physiol 294(2):R585-593
Jones AM, Vanhatalo A, Burnley M, Morton RH, Poole DC (2010)
Critical power: implications for determination of V˙O2max and
exercise tolerance. Med Sci Sports Exerc 42(10):1876–1890
Jones AM, Burnley M, Black MI, Poole DC, Vanhatalo A (2019a) The
maximal metabolic steady state: redefining the ‘gold standard.’
Physiol Rep 7(10):e14098
Jones AM, Burnley M, Black MI, Poole DC, Vanhatalo A (2019b)
Response to considerations regarding maximal lactate steady state
determination before redefining the gold-standard. Physiol Rep
7(22):e14292
Jones AM, Kirby BS, Clark IE, Rice HM, Fulkerson E, Wylie LJ,
Wilkerson DP, Vanhatalo A, Wilkins BW (2021) Physiological
demands of running at 2-hour marathon race pace. J Appl Physiol
130:369–379
Karsten B, Jobson SA, Hopker J, Stevens L, Beedie C (2015) Validity
and reliability of critical power field testing. Eur J Appl Physiol
115(1):197–204
Keir DA, Fontana FY, Robertson TC, Murias JM, Paterson DH, Kow-
alchuk JM (2015) Exercise intensity thresholds: identifying the
boundaries of sustainable performance. Med Sci Sports Exerc
47(9):1932–1940
Krustrup P, Jones AM, Wilkerson DP, Calbet JA, Bangsbo J (2009)
Muscular and pulmonary O2 uptake kinetics during moderate- and
high-intensity sub-maximal knee-extensor exercise in humans. J
Physiol 587(Pt 8):1843–1856
Mattioni Maturana F, Keir DA, McLay KM, Murias JM (2016) Can
measures of critical power precisely estimate the maximal meta-
bolic steady-state? Appl Physiol Nutr Metab 41(11):1197–1203
Mattioni Maturana F, Fontana FY, Pogliaghi S, Passfield L, Murias JM
(2018) Critical power: how different protocols and models affect
its determination. J Sci Med Sport 21(7):742–747
Monod H, Scherrer J (1965) The work capacity of a synergic muscular
group. Ergonomics 8(3):329–338
Moritani T, Nagata A, deVries HA, Muro M (1981) Critical power
as a measure of physical work capacity and anaerobic threshold.
Ergonomics 24(5):339–350
Morton RH, Stannard SR, Kay B (2012) Low reproducibility of many
lactate markers during incremental cycle exercise. Br J Sports
Med 46(1):64–69
Pethick J, Winter SL, Burnley M (2020) Physiological evidence that
the critical torque is a phase transition, not a threshold. Med Sci
Sports Exerc 52(11):2390–2401
Poole DC, Ward SA, Gardner GW, Whipp BJ (1988) Metabolic and
respiratory profile of the upper limit for prolonged exercise in
man. Ergonomics 31(9):1265–1279
Poole DC, Burnley M, Vanhatalo A, Rossiter HB, Jones AM (2016)
Critical power: an important fatigue threshold in exercise physiol-
ogy. Med Sci Sports Exerc 48(11):2320–2334
Pringle JS, Jones AM (2002) Maximal lactate steady state, criti-
cal power and EMG during cycling. Eur J Appl Physiol
88(3):214–226
Pringle JS, Doust JH, Carter H, Tolfrey K, Campbell IT, Sakkas GK,
Jones AM (2003) Oxygen uptake kinetics during moderate, heavy
and severe intensity “submaximal” exercise in humans: the influ-
ence of muscle fibre type and capillarisation. Eur J Appl Physiol
89(3–4):289–300
Rossiter HB, Ward SA, Kowalchuk JM, Howe FA, Griffiths JR, Whipp
BJ (2002) Dynamic asymmetry of phosphocreatine concentration
and O(2) uptake between the on- and off-transients of moderate-
and high-intensity exercise in humans. Jphysiol 541:991–1002
Smith CG, Jones AM (2001) The relationship between critical velocity,
maximal lactate steady-state velocity and lactate turnpoint veloc-
ity in runners. Eur J Appl Physiol 85(1–2):19–26
Smyth B, Muniz-Pumares D (2020) Calculation of critical speed from
raw training data in recreational marathon runners. Med Sci
Sports Exerc 52(12):2637–2645
Tanaka K (1991) Cardiorespiratory and lactate responses to a 1-hour
submaximal running at the lactate threshold. Ann Physiol Anthro-
pol 10(3):155–162
Tanner RK, Fuller KL, Ross MLR (2010) Evaluation of three portable
blood lactate analysers: Lactate pro, lactate scout and lactate plus.
Eur J Appl Physiol 109(3):551–559
Vanhatalo A, Jones AM, Burnley M (2011) Application of critical
power in sport. Int J Sports Physiol Perform 6(1):128–136
Vanhatalo A, Black MI, DiMenna FJ, Blackwell JR, Schmidt JF,
Thompson C, Wylie LJ, Mohr M, Bangsbo J, Krustrup P, Jones
AM (2016) The mechanistic bases of the power-time relationship:
muscle metabolic responses and relationships to muscle fibre type.
J Physiol 594(15):4407–4423
European Journal of Applied Physiology (2021) 121:3133–3144
3143
1 3
Whipp BJ (1994) The slow component of O2 uptake kinetics during
heavy exercise. Med Sci Sports Exerc 26(11):1319–1326
Whipp BJ, Ward SA (1992) Pulmonary gas exchange dynamics and
the tolerance to muscular exercise: effects of fitness and training.
Ann Physiol Anthropol 11(3):207–214
Whipp BJ, Wasserman K (1972) Oxygen uptake kinetics for various
intensities of constant-load work. J Appl Physiol 33:351–356
Wilkerson DP, Koppo K, Barstow TJ, Jones AM (2004) Effect of work
rate on the functional “gain” of phase II pulmonary O2 uptake
response to exercise. Respir Physiol Neurobiol 142(2–3):211–223
Publisher’s Note Springer Nature remains neutral with regard to
jurisdictional claims in published maps and institutional affiliations.
European Journal of Applied Physiology (2021) 121:3133–3144
3144
| Steady-state [Formula: see text] above MLSS: evidence that critical speed better represents maximal metabolic steady state in well-trained runners. | 08-05-2021 | Nixon, Rebekah J,Kranen, Sascha H,Vanhatalo, Anni,Jones, Andrew M | eng |
PMC4076256 | 1
List of abbreviations, definitions and symbols
Abbreviations
CoM
center of mass
SLIP
spring loaded inverted pendulum model
TD
touch down, referring to landing conditions at the swing-stance transition
Definitions
Model
A reduced-order mathematical description of the physical sys-
tem. Here we use the highly reductionist spring-mass model
with massless leg.
Landing conditions
Initial conditions of the CoM (position and velocity) at the be-
ginning of stance. These are directly influenced by the swing-
leg trajectory during flight.
Passive dynamics
Synonymous with intrinsic dynamics—the response of the
physical model. Here, the stance dynamics of the model are
fully determined by landing conditions and leg stiffness.
Control policy
Active control applied to the model with a specific target per-
formance goal. Here the only applied control is late-swing leg
angular trajectory.
Peak force control
Late-swing leg trajectory optimized to target landing condi-
tions for constant peak force of the SLIP model in the drop
step (equal to the peak force of the previous step).
Impulse control
Late-swing leg trajectory optimized to target landing condi-
tions for constant axial impulse of the SLIP model in the drop
step (equal to the impulse of the previous step).
Equilibrium gait control
Late-swing leg trajectory optimized to target landing condi-
tions for perfect disturbance rejection of the SLIP model in
the drop step, resulting in a steady, symmetric gait cycle.
2
Parameters
SI Units
g
gravitational acceleration [m/s2]
m
body mass [kg]
L0
resting leg length [m]
BW = mg
body weight [N]
T =
p
L0/g
periodic time of a pendulum [s]
Non-dimensional
α
leg angle [deg]
αPolicy
angle of the virtual leg (CoM to foot) predicted by a swing-leg control strategy [deg]
αSLIP
angle of the virtual leg for the SLIP model during stance [deg]
˙α
leg angular velocity [deg/T]
∆ECoM
net CoM work [BW L0]
Faxial
axial leg force [BW]
Iaxial
axial leg impulse [BW T]
Ix
fore-aft impulse [BW T]
kLeg
effective linear leg stiffness [BW/L0]
L
leg length [L0]
˙L
leg length velocity [L0/T]
r = (x, y)T
CoM position [L0]
˙r = ( ˙x, ˙y)T
CoM velocity [L0/T]
¨r = (¨x, ¨y)T
CoM acceleration [L0/T 2]
t
time [T]
| Swing-leg trajectory of running guinea fowl suggests task-level priority of force regulation rather than disturbance rejection. | 06-30-2014 | Blum, Yvonne,Vejdani, Hamid R,Birn-Jeffery, Aleksandra V,Hubicki, Christian M,Hurst, Jonathan W,Daley, Monica A | eng |
PMC8128237 | Dear authors and editor,
It is a pleasure the opportunity to act as a reviewer for PLOS ONE. So, thank you very
much for it. I have written my comments into two sections, the general comments with
the main three ideas, and specific comments. With them, I hope to help the authors to
improve the manuscript, encouraging them to perform due to the relevance of this topic.
General comments
-
The lack of a deep literature review makes that introduction and discussion are
overviews, but not further contextualization of “state of the art” or discussion of
this topic.
-
I am worried with the sampling rate considered. In sport literature, two
systematic reviews have recently highlighted that tracking technologies are
limited in high intensity short efforts such as CODs [1,2]. However, the authors
considered 20 Hz for data raw and then they down sampling. Are these sampling
rates suitable for this analysis? or is it a limitation?
-
In my opinion, the results section shows a deep analysis, that does not meet with
the rest of the article. The results showed:
1. The difference between COD and set angle.
2. Number and duration of COD during matches.
3. Number of CODs per playing position.
Regarding to the results 1 and 2, the aim should be: the validity of a new approach
to set angles and duration of CODs. These aims are contextualized and fit with the
rest of the sections (introduction and discussion), however, the reason to present
data per playing position is not contextualized. Why do the authors consider it?
Specific comments
Title:
-
Re-consider changing “locomotion” by “set angles and time of CODs”.
Abstract:
-
“CODs” is not defined the first time in it was mentioned.
-
The “background” of abstract not contextualize the main objective. It should be
something like: “Soccer players frequently change of directions (CODs) at
various speeds during matches”. However, tracking systems have shown
limitations to measure these efforts. Therefore, the aim of the present study was
to propose a new approach to measure CODs using a local positioning system
(LPS).”
Introduction:
-
The introduction lacks from relevant literature, making non suitable the
contextualization of the objective. I suggest:
o Paragraph 1: Explain the importance of change of directions. Maybe,
different studies performed using PCA are suitable option to highlight
the importance of change of directions and other high-intensity short
efforts [3]. (The idea has been written, but in my opinion, further
information is needed).
o Paragraph 2: It should explain:
▪ What are the tracking technologies (the main basis to use the new
proposal that the authors presented).
▪ What it has been found about tracking systems and change of
directions.
▪ What has it been the main problem of accuracy in this regard (e.g
sampling rate).
o Paragraph 3: it should explain “the state of the art” about all studies
published in this topic (LPS and change of direction). Consider, at least,
these references: [4–8]. Following this recently publicized systematic
review [1], these all the validity proposals using LPS in sport setting.
*In general, the introduction has interesting ideas, but it should be deeply rewritten.
Methodology:
Separate “participants” from “experimental design”.
Participants:
Add:
-
Where do the players selected? What inclusion/exclusion criteria were
considered?
-
Participants characteristics.
-
See any scale for risk of bias, and follow it.
Experimental design:
-
Line 77-78: CODs in directions determined by 13 set angles. My question is: 3
or 13 set angles?
Players´ coordinate data:
-
Sampling frequency: the authors mentioned 20 Hz (and less with filtering
processed). However, the use of 20 Hz has not been enough for high intensity
short efforts using LPS. Why do the authors think that in this study could be
suitable sampling frequency? Could be a limitation? What was the criteria to use
these sampling rates?
-
The authors mentioned that an LPS was used, but, several factors could affect
the outcome that they could not be related with the proposed approach.
Therefore, a higher precision about “the use of technology” should be made,
mainly to avoid the following principle: “A good experimental design is one in
which the only explanation for the change in the dependent variable is due to the
treatment applied”. I suggest the use of recently published survey [9].
Results
The results are well-conducted, but they have different ways ((i) the difference between
COD and set angle, (ii) number and duration of COD during matches, and (iii) number
of CODs per playing position)) that are not reflected in the rest of the article. In my
opinion, the article should be adapted to performed results (see general comments).
Consider explain the three protocols in the abstract.
Discussion:
-
Re-write the discussion, considering the suggested articles for the introduction
(paragraph 3).
-
Reconsider separate a discussion for each of the results presented: (i) the
difference between COD and set angle, (ii) number and duration of COD during
matches, and (iii) number of CODs per playing position. If all of them are
relevant for the aim of this study (see general comments).
-
Add limitations.
Conclusion:
See the comments mentioned above. If following the results the aim was different folds,
the conclusions should have different folds too.
References:
A further revision of the literature is needed. Different articles have been published
about the use of LPS to assess CODs, and some of them were not considered.
Bibliography:
1. Rico-González M, Arcos AL, Clemente FM, Rojas-Valverde D, Pino-Ortega J.
Accuracy and Reliability of Local Positioning Systems for Measuring Sport Movement
Patterns in Stadium-Scale: A Systematic Review. Applied Sciences. 2020;10:5994.
2. Pino-Ortega J, Oliva-Lozano JM, Gantois P, Nakamura FY, Rico-González M.
Comparison of the validity and reliability of local positioning systems against other
tracking technologies in team sport: A systematic review. Proc IMechE Part P: J Sports
Engineering and Technology. 2021;
3. Oliva-Lozano JM, Rojas-Valverde D, Gómez-Carmona CD, Fortes V, Pino-Ortega J.
Impact of contextual variables on the representative external load profile of Spanish
professional soccer match-play: A full season study. European Journal of Sport Science.
2020;1–10.
4. Frencken WGP, Lemmink KAPM, Delleman NJ. Soccer-specific accuracy and
validity of the local position measurement (LPM) system. Journal of Science and
Medicine in Sport. 2010;13:641–5.
5. Ogris G, Leser R, Horsak B, Kornfeind P, Heller M, Baca A. Accuracy of the LPM
tracking system considering dynamic position changes. Journal of Sports Sciences.
2012;30:1503–11.
6. Stevens TGA, de Ruiter CJ, van Niel C, van de Rhee R, Beek PJ, Savelsbergh GJP.
Measuring Acceleration and Deceleration in Soccer-Specific Movements Using a Local
Position Measurement (LPM) System. International Journal of Sports Physiology and
Performance. 2014;9:446–56.
7. Linke D, Link D, Lames M. Validation of electronic performance and tracking
systems EPTS under field conditions. Ardigò LP, editor. PLOS ONE.
2018;13:e0199519.
8. Luteberget LS, Spencer M, Gilgien M. Validity of the Catapult ClearSky T6 Local
Positioning System for Team Sports Specific Drills, in Indoor Conditions. Front
Physiol. 2018;9:115.
9. Rico-González M, Arcos AL, Rojas-Valverde D, Clemente FM, Pino-Ortega J. A
Survey to Assess the Quality of the Data Obtained by Radio-Frequency Technologies
and Microelectromechanical Systems to Measure External Workload and Collective
Behavior Variables in Team Sports. Sensors. 2020;16.
| A new approach to quantify angles and time of changes-of-direction during soccer matches. | 05-17-2021 | Kai, Tomohiro,Hirai, Shin,Anbe, Yuhei,Takai, Yohei | eng |
PMC10011548 | 1
Vol.:(0123456789)
Scientific Reports | (2023) 13:4167
| https://doi.org/10.1038/s41598-023-30798-3
www.nature.com/scientificreports
Progressive daily hopping exercise
improves running economy
in amateur runners: a randomized
and controlled trial
Tobias Engeroff 1,6, Kristin Kalo 2,6*, Ryan Merrifield 3, David Groneberg 4 & Jan Wilke 4,5
This study investigated the effects of a daily plyometric hopping intervention on running economy
(RE) in amateur runners. In a randomized, controlled trial, thirty-four amateur runners (29 ± 7 years, 27
males) were allocated to a control or a hopping exercise group. During the six-week study, the exercise
group performed 5 min of double-legged hopping exercise daily. To progressively increase loading, the
number of hopping bouts (10 s each) was steadily increased while break duration between sets was
decreased. Pre- and post-intervention, RE, peak oxygen uptake (VO2peak), and respiratory exchange
ratio (RER) were measured during 4-min stages at three running speeds (10, 12, and 14 km/h).
ANCOVAs with baseline values and potential cofounders as cofactors were performed to identify
differences between groups. ANCOVA revealed an effect of hopping on RE at 12 km/h (df = 1; F = 4.35;
p < 0.05; η2 = 0.072) and 14 km/h (df = 1; F = 6.72; p < 0.05; η2 = 0.098), but not at 10 km/h (p > 0.05).
Exercise did not affect VO2peak (p > 0.05), but increased RER at 12 km/h (df = 1; F = 4.26; p < 0.05;
η2 = 0.059) and 14 km/h (df = 1; F = 36.73; p < 0.001; η2 = 0.520). No difference in RER was observed
at 10 km/h (p > 0.05). Daily hopping exercise is effective in improving RE at high running speeds in
amateurs and thus can be considered a feasible complementary training program.
Clinical trial registration German Register of Clinical Trials (DRKS00017373).
Competitive runners are on the constant quest for maximal performance. However, after decades of significant
improvements, a trend towards more marginal changes has been observed in several disciplines including endur-
ance running1. Consequently, a “marginal gains” approach, which aims at combining multiple small performance
improvements in different areas to create a significant advantage, is becoming increasingly popular in amateur
and professional competitive sports. Aiming at such marginal gains, amateur runners adopt complementary
training approaches such as plyometrics from professional sports with the aim to maximize performance and
minimize injury risk2.
Running performance is inherently dependent on the efficiency of locomotion, which is often referred to as
running economy (RE). How efficient a human moves over the ground is not only influenced by metabolic fac-
tors, but also by the quality of movement patterns and the mechanical characteristics of the locomotor system3.
The concept of RE takes these metabolic, neural and tissue-specific factors into account. It is defined as the oxygen
uptake required per distance at a given running speed and represents one of the key parameters to quantify the
ability to transform aerobic capacity into endurance running performance3,4. Studies in both, amateurs5 and
elite athletes6, confirmed the relevance of RE as a crucial factor for endurance running performance. Therefore,
strategies to improve RE are sought after by coaches, athletes, and sports scientists7.
Resistance training, plyometrics, and stretching represent three popular methods used to target RE7. A
common facet of these interventions is that they all influence metabolic, biomechanical and neuromuscular
OPEN
1Division Health and Performance, Institute of Occupational, Social and Environmental Medicine, Goethe
University Frankfurt, Frankfurt Am Main, Germany. 2Department of Sports Medicine, Disease Prevention
and Rehabilitation, Johannes Gutenberg University Mainz, Albert-Schweitzer-Straße 22, 55128 Mainz,
Germany. 3Department of Sports Medicine and Exercise Physiology, Institute of Sports Sciences, Goethe
University Frankfurt, Frankfurt Am Main, Germany. 4Institute of Occupational, Social and Environmental Medicine,
Goethe University, Frankfurt, Frankfurt Am Main, Germany. 5Departement of Movement Sciences, University of
Klagenfurt, Klagenfurt, Austria. 6These authors contributed equally: Tobias Engeroff and Kristin Kalo. *email:
[email protected]
2
Vol:.(1234567890)
Scientific Reports | (2023) 13:4167 |
https://doi.org/10.1038/s41598-023-30798-3
www.nature.com/scientificreports/
efficiency7. Since simply replacing endurance training by an RE intervention would limit the ability to main-
tain or increase maximal aerobic performance, a more promising approach to maximize running performance
requires the maintenance of endurance training routines to which additional methods that may improve RE are
added (e.g. explosive strength exercises3). Popular contents of such additional training strategies include jumps,
hops, or sprints and are suggested to improve muscle/tendon stiffness or to modify movement mechanics and
the stretch-shorten cycle (SSC)3.
According to the available evidence, the Achilles tendon (AT) has been demonstrated to play a significant role
in RE8–10. Kunimasa et al.11 compared a variety of anatomical characteristics of Kenyan and Japanese elite distance
runners. The Kenyans, superior in running performance, exhibited higher relative AT lengths and greater AT
tendon moment arms. This is of importance because both factors are positively associated with RE9,10. In addi-
tion to morphological features, RE is also influenced by the mechanical properties of the AT. Arampatzis et al.8
showed that RE is positively associated with normalized AT stiffness. Following a 14-week resistance exercise
intervention, a 7%-increase in plantar flexor strength and a 16%-increase in tendon stiffness resulted in a 4%
reduction of oxygen consumption12.
In a pioneering study, Kawakami et al.13 found that muscle fibers, contrary to earlier beliefs, act almost iso-
metrically during stretch–shortening cycles (SSC). Conversely, the tendon undergoes significant length changes,
storing and releasing kinetic energy. While the AT contributes more than 50% of the positive work even at low
running speeds of ~ 2 m/s, this proportion increases to about 75% during sprinting14. In view of the accumulat-
ing evidence supporting the importance of the AT in RE, numerous studies have investigated the effectiveness
of related exercise interventions. Plyometric training, often using reactive jumps, hops or, bounces when applied
in the lower limb, is a popular strategy aiming to improve SSC performance. It hence seems particularly suited
to trigger morphological and functional adaptation of the tendon. However, so far, only a limited number of
studies examined the effect of explosiveness training interventions on RE, and only a few of them used exclu-
sively plyometric exercises3. Furthermore, most available trials used one to three sessions per week, but none
studied higher frequencies3. This is of importance because it has been shown that collagen production, which
is paramount for tendon stiffness, is most effectively triggered by intermittent, progressive loading paradigms
with relatively short durations and intervals15,16. Finally, most of the available evidence of exercise interventions
and RE focusses on athletes with moderate to high-performance levels17. The present study therefore aimed to
investigate the effects of a daily plyometric hopping intervention on RE in amateur runners. We defined amateur
runners as persons who compete in sports without striving for financial reward (as opposed to professional
athletes), thus do running as a hobby.
Methods
Study design and survey procedure.
A two-arm randomized controlled trial was performed. Active
amateur runners were allocated to a hopping exercise (HE) or control (CON) group. Randomization was coun-
terbalanced and conducted using BiAS for Windows version 11.10 (Goethe University Frankfurt, Germany).
The study was approved by the local ethics committee (Ethikkommission FB 05, Goethe University Frankfurt;
reference number: 2018-17b) and registered at the German Register of Clinical Trials (DRKS00017373, date of
registry: 03/09/2019). All participants provided written informed consent.
Healthy adults were recruited using word of mouth, printed flyers, and social media advertising. To ensure a
specific fitness level and thus that participants are able to complete the running protocol, individuals had to be
amateur runners with a 10 km time < 55 min and younger than 40 years of age. Exclusion criteria encompassed
contraindications for engagement in physical activity (tested by means of the Physical Activity Readiness Ques-
tionnaire), severe cardiovascular, metabolic, endocrine, neural, and psychiatric diseases, unhealed orthopaedic
injuries and overuse disorders (particularly with regard to the knee and ankle region), local inflammation,
pregnancy, self-reported use of supplements containing stimulants and anabolic–androgenic steroids.
Intervention.
The CON and the HE groups continued their regular exercise regimes. While the CON group
did not engage in an additional specific exercise intervention, the HE group completed a six-week plyometric
hopping protocol. Each day, the individuals randomized to HE performed a variable amount of double-legged
10-s hopping bouts (Table 1). While total session duration (5 min) was constant, the number of sets (and with
this, net training time) was increased weekly in order to ensure safe functional and mechanical adaptation.
When hopping, participants were instructed to start with both feet no wider than hip width apart and to hop
as high as possible with both legs, keeping the knees extended and aiming to minimize ground contact time.
Table 1. Protocol of the hopping intervention. s seconds.
Week
Sets
Set duration [s]
Net hopping duration [s]
Rest between sets [s]
1
5
10
50
50
2
6
10
60
40
3
8
10
80
30
4
10
10
100
20
5
15
10
150
10
6
15
10
150
10
3
Vol.:(0123456789)
Scientific Reports | (2023) 13:4167 |
https://doi.org/10.1038/s41598-023-30798-3
www.nature.com/scientificreports/
To ensure safe and correct execution, participants received a 1-to-1 explanation by a coach holding a bachelor’s
degree in Sports Science, who additionally monitored the first three training sessions of each individual. When
later exercising alone, the HE participants provided the instructor with video recordings in order to allow super-
vision and, if needed, correction. Moreover, participants completed a hopping diary. If at least 70% of the jumps
were completed, the subjects were considered compliant.
All participants completed a training diary, documenting weekly running activity (number of sessions, hours
per session, pace) as well as other exercises in hours/ week (see Table 2) and (in case of the HE group) adherence
to the hopping intervention.
Measures.
Before and after the intervention period, all exercise tests were performed on an electronically
driven treadmill (mercury® med, h/p/cosmos sports & medical gmbh, Traunstein, Germany) without using a
safety belt. To reduce intrasubject variability, the time of the day, the test equipment, as well as the type of run-
ning shoes worn were standardized for both, baseline and post-exercise tests. Participants were instructed to
avoid exercise for 24 h and strenuous exercise for 48 h prior to testing. In addition, participants were asked to eat
about 1.5 to 2 h before exercise testing, not to drink alcohol the day before, and not to consume any alcohol or
cigarettes on the day of the test. Temperature was controlled by air conditioning and the difference in tempera-
ture between the baseline and post measurement did not exceed ± 0.5° Celsius.
In accordance with Saunders et al.18, RE was determined by measuring submaximal oxygen uptake (VO2)
during 4-min stages at three constant running speeds and an inclination of 0°. After a standardized warm-up of
3 min walking at 5 km/h, participants ran at 10, 12, and 14 km/h, respectively7. Peak oxygen uptake (VO2peak) was
determined during a ramp protocol performed 2 min after the last submaximal running stage. For this purpose,
speed was increased by 1 km/h every minute from 12 km/h and to 20 km/h, treadmill inclination was increased
by 1% until volitional exhaustion19.
A breath-by-breath gas analyser was used to monitor gas exchange during exercise testing (Metalyzer, COR-
TEX Biophysik GmbH, Leipzig, Germany). Calibration of the gas as well as the flow sensor was performed
according to manufacturer recommendations. Meyer et al.20 showed an excellent test–retest reliability for the
used system (VO2: 0.969, VCO2: 0.964, VE 0.953). Recorded data was stored and processed using a spiroergometry
software (MetaSoft® Studio, CORTEX Biophysik GmbH, Leipzig, Germany). At the end of each running stage, the
participants prepared themselves to step off the treadmill by gripping the side handles of the treadmill. Therefore,
for each step, the last 10 s of VO2 data was cut off and RE was determined as the V̇O2 collected during the last
valid 60 s (i. e., seconds 170 to 230) of each 4-min running stage. Respiratory exchange ratio (RER, VCO2/VO2)
was calculated for these 60 s as well. During the ramp protocol, the highest individual V̇O2 recorded over a 30 s
period within the testing time (floating mean) was defined as VO2peak.
Statistical analysis.
All analyses were performed using Jamovi 1.8 (The jamovi project, https:// www. jam-
ovi. org) and the significance level was set to α = 0.05. After variance homogeneity was confirmed using Levene´s
Test, analyses of covariance (ANCOVA) with baseline values as a cofactor were performed to test for inter-
vention effects (between subjects/groups) on running economy and secondary outcomes (VO2peak, respiratory
exchange ratio). Sex was analysed as potential confounder. Further confounders (weight, frequency of regular
running sessions and other exercises) were tested for between group differences using Kruskal–Wallis Tests. In
case of significance, a second ANCOVA for intervention effects on running economy including baseline values
for running economy and the potential confounding outcomes was carried out. For the estimates of effect sizes,
eta squared (η2) was used and interpreted according to Cohen21: 0.01 (small effect), 0.06 (medium effect) and
0.14 (large effect).
Ethics approval and patient consent.
The study was conducted according to the ethical guidelines of
the Helsinki Declaration and was approved by the local ethical review board (Ethikkommission FB 05, Goethe
University Frankfurt), number: 2018-17b. All participants provided written informed consent.
Table 2. Description of the sample (mean values and standard deviations). n number, kg kilograms, m meter,
h hours, km kilometers.
HE group (n = 15)
CON group (n = 19)
Total (n = 34)
Sex
11♂ ♀4
16♂ ♀3
27♂ ♀7
Age [years]
29.1 (7.6)
28.2 (5.9)
28.6 (6.6)
Weight [kg]
73.8 (10.4)
78.6 (9.01)
76.5 (9.8)
Height [m]
1.78 (0.07)
1.80 (0.07)
1.80 (0.08)
Exercise [h/week]
8.0 (3.0)
8.2 (3.4)
8.1 (3.2)
Running duration [h/week]
3.3 (2.4)
2.37 (2.2)
2.8 (2.3)
Running frequency [n/week]
2.7 (1.5)
1.89 (1.3)
2.2 (1.4)
Running experience [years]
6.2 (4.7)
8.21 (5.6)
7.3 (5.3)
Running speed [km/h]
10.9 (1.8)
11.15 (1.5)
11.0 (1.6)
4
Vol:.(1234567890)
Scientific Reports | (2023) 13:4167 |
https://doi.org/10.1038/s41598-023-30798-3
www.nature.com/scientificreports/
Results
From n = 46 recruited individuals, a total of n = 34 adults (29 ± 7 years, 27 males) completed the study. Overall,
n = 12 participants dropped out of our study. Reasons were illness (n = 4), injury (n = 4), lack of time to follow-up
(n = 2), and < 70% compliant with the hopping protocol (n = 2). A detailed indication of the dropouts per group
is depicted in Fig. 1.
Descriptive data of sample characteristics are presented in Table 2. None of the potential confounders includ-
ing weight (df = 1; F = 0.0468; p = 0.830; η2 = 0.001), running frequency (df = 1; χ2 = 0.0305; p = 0.861), and general
training volume (df = 1; χ2 = 0.8605; p = 0.354) showed significant differences between the two groups. Pre and
post values of VO2peak, RE and RER per stage and group are presented in Table 3.
Levene´s test indicated variance homogeneity for primary and secondary outcomes. ANCOVA revealed that
hopping significantly improves running economy at 12 km/h (df = 1; F = 4.35; p = 0.045; η2 = 0.072) and 14 km/h
(df = 1; F = 6.72; p = 0.015; η2 = 0.098) running speed. In contrast, no difference between the HE and CON group
was found at a low (10 km/h) running speed (df = 1; F = 3.11; p = 0.088; η2 = 0.043). HE did not lead to higher
Figure 1. Flow chart of the study.
Table 3. Spiroergometric values (mean values and standard deviations). n number, kg kilograms, h hours, km
kilometers, ml milliliters, min minutes, VO2 oxygen intake, RE running economy, RER respiratory exchange
ratio, VCO2 carbon dioxide production.
HE group (n = 15)
CON group (n = 19)
Total (n = 34)
Pre
Post
Pre
Post
Pre
Post
VO2peak [ml/min/kg]
51.4 (6.0)
51.1 (5.6)
48.26 (4.3)
50.2 (4.5)
49.7 (5.2)
50.6 (5.0)
RE (ml/min/kg), 10 km/h
35.2 (2.97)
33.9 (2.95)
34.8 (3.07)
34.8 (2.79)
35.0 (2.99)
34.4 (2.86)
RE (ml/min/kg), 12 km/h
41.1 (2.57)
40.2 (2.67)
40.4 (3.35)
41.3 (2.86)
40.7 (3.01)
40.9 (2.79)
RE (ml/min/kg), 14 km/h
47.1 (3.04)
46.1 (3.31)
45.3 (3.33)
46.9 (2.79)
46.1 (3.29)
46.5 (3.01)
RER (VCO2/VO2), 10 km/h
0.90 (0.04)
0.93 (0.05)
0.92 (0.05)
0.93 (0.05)
0.91 (0.05)
0.93 (0.05)
RER (VCO2/VO2), 12 km/h
0.95 (0.04)
0.98 (0.05)
0.97 (0.05)
0.97 (0.05)
0.96 (0.05)
0.98 (0.05)
RER (VCO2/VO2), 14 km/h
1.01 (0.06)
1.05 (0.07)
1.02 (0.05)
1.02 (0.05)
1.02 (0.06)
1.03 (0.06)
5
Vol.:(0123456789)
Scientific Reports | (2023) 13:4167 |
https://doi.org/10.1038/s41598-023-30798-3
www.nature.com/scientificreports/
VO2peak values in the HE group compared to the CON group (df = 1; F = 2.83; p = 0.102; η2 = 0.025). However, after
six weeks of training, the respiratory exchange ratios during 12 km/h (df = 1; F = 4.26; p = 0.047; η2 = 0.059) and
14 km/h (df = 1; F = 36.73; p < 0.001; η2 = 0.520) running speed were significantly higher in the HE group. RER
at 10 km/h running speed remained unchanged (df = 1; F = 0.490; p = 0.489; η2 = 0.011). Sex showed no impact
on ANCOVA results. Figure 2 shows group differences in the estimated marginal means of the post-values for
RE and RER (generated considering the pre-values) at all three running speeds.
Discussion
The present study yielded three key findings. Firstly, six weeks of daily hopping exercise improve running econ-
omy and increase respiratory exchange ratio at higher running speeds (12 and 14 km/h) in amateur runners.
Secondly, maximal aerobic capacity remains unaltered by hopping if regular running and exercise habits are
maintained. Our findings are in line with earlier studies examining plyometric interventions in amateur athletes22
and highly trained runners18. They also corroborate the observation of Saunders et al.18 that the effects of plyomet-
rics on running economy are more pronounced at higher running speeds. However, in contrast to the previous
trials which used three weekly sessions with durations of up to 30 min as well as multiple jump exercises18,22, this
study applied short daily bouts (net duration 5 min) consisting of one simple and easy-to-learn hopping exercise
only. Our results suggest that regular endurance and concurrent plyometric training can be performed jointly by
amateur runners, without complex programs and at a high frequency without leading to adverse events such as
overuse injuries or pain. Finally, third, performing a progressive hopping protocol with a high exercise density
seems to be safe in amateur athletes as no injury or other side effects related to the interventions were reported.
Although the present trial strengthens the evidence that plyometric exercise improves running economy, the
relative contributions of factors driving the observed changes are still a matter of debate. Tendon stiffness has
been shown to significantly correlate with running economy23. Earlier findings reported that the contribution
of AT energy storage to the positive work performed during running increases with locomotion speed14. Our
data show significant improvements in RE at 12 and 14 km/h but not at 10 km/h which would provide support
for the hypothesis that tendon stiffness was one of the driving factors for improvements in RE.
Our hypothesis assumed that repeated hopping would elicit morphological and functional adaptation in the
tissue, which in turn improve energy storage capacity. The daily loading paradigm was chosen based on research
elucidating optimal loading paradigms for collagenous connective tissues15,16. For instance, Paxton et al.15 used
engineered ligaments to study the effect of cyclic stretch on the phosphorylation of the extracellular signal-
regulated kinase 1/2 (ERK1/2), an enzyme associated with collagen synthesis. Maximal ERK 1/2 levels were
observed after 10 min of cyclic stretch and up to 6 h, cells were refractory to new stretching bouts. As these data
suggest that repeated short loading periods may be optimal for tissue adaptation, the authors compared a daily
intermittent stretch program (10 min each 6 h) to continuous tissue lengthening and confirmed the superiority
of the physiology-specific paradigm. In our trial, tissue loading was of similar duration (8–14 min). Yet, as three
training bouts per day with 6-h-intervals are not feasible due to nocturnal sleep, we decided for one exercise
Figure 2. Group differences in the estimated marginal means of the post intervention values of RE and RER at
three running speeds. Means (dots) and 95% confidence intervals (vertical lines) are displayed.
6
Vol:.(1234567890)
Scientific Reports | (2023) 13:4167 |
https://doi.org/10.1038/s41598-023-30798-3
www.nature.com/scientificreports/
session per day. The intervention was well tolerated and just two participants report pain (shin and ball of the
foot), which might be related to the intervention. Upcoming trials may therefore consider further increasing
the frequency, e.g., to twice daily. Also, it has to be noted that we did not measure tendon stiffness (e.g., using an
instrumented treadmill). Future studies should hence be geared to combine both, assessments of stiffness and RE.
In addition to adaptations in the tendon, metabolic factors may have caused the alterations of RE3. One often
discussed adaptation is increased aerobic carbohydrate utilisation for oxidative phosphorylation24. A shift in
substrate use could have led to better running economy based on a lower oxygen cost for adenosine triphosphate
(ATP) synthesis if a greater share of carbohydrates is used instead of fatty acids. Since RER values in our study
exceeded 1.0 at higher running speed, another explanation for decreased oxygen uptake might be an increase in
anaerobic metabolism5. The assumption that a change in substrate use might have caused alterations in running
economy is supported by an increase in respiratory exchange ratios in the hopping training group. Due to the
applied method, which is based on oxygen uptake, running economy in our analysis reflects only aerobic energy
metabolism. Based on our data, we thus are not able to delineate which of the aforementioned mechanisms led
to the increase in RER. Furthermore, we did not control if all participants were able to maintain a metabolic
steady state over a larger timeframe than 4 min during all tested running speeds and did not assess running
performance using a time trial or a fixed distance. Consequently, further studies are necessary to prove, that
differences in oxygen costs of running govern improvements in running performance. A third pathway for hop-
ping induced improvements in RE could be based on changes in movement patterns3. Alterations in running
mechanics are discussed as a possible mechanism for RE improvements induced by plyometric and explosive
resistance training7. However, evidence on specific adaptations is scarce so far and further studies investigating
running mechanics are needed. Moreover, we did not investigate the subjective acceptance and feasibility of our
hopping protocol to daily (exercise) routines using structured interviews.
Our findings have implications for clinical practice and may open new avenues for future research. As indi-
cated earlier, previous training paradigms predominantly focussed on complex and more time consuming inter-
ventions which are particularly used by professional athletes3. This trial provided first evidence that short daily
regimes consisting of just one jump exercise are effective in improving RE in amateurs. Such regimes become
more relevant as many people run and compete for fun and social aspects, but also report musculoskeletal pain
caused by running25. In addition, the “marginal gains” approach, is becoming increasingly popular in amateur
sports2. Therefore, hopping can be considered as short and feasible program for amateur runners to optimize
performance and minimize the risk for musculoskeletal pain and injuries. Although direct comparison of a) short
daily and b) longer but less frequently applied training approaches are lacking, based on current evidence, the
choice could be left to individual preference. In regard to applicability, further studies should analyse if plyometric
exercises can be executed in a fatigued and non-fatigued state. As an example, it is unknown whether hopping
interventions performed in direct proximity to regular training (immediately before or after) would thwart the
beneficial effects achieved by endurance or hopping exercises. Finally, besides endurance runners, athletes in
other sports might also benefit from increased RE. Team sports such as basketball or soccer are characterized
by periods of moderate intensity running which are disrupted by short bouts of high-intensity actions26. Con-
sequently, RE is discussed as a relevant factor for athletic performance26 and training regimes might contribute
to enhanced endurance and running speed in team sport athletes.
Perspective.
This study provides first evidence that 5 min of daily hopping improve RE at moderate and high
running speed without compromising maximal aerobic capacity in amateur runners. This is in line with previous
studies using less frequent jump exercises with higher duration18,22. The more frequent units nevertheless appear
to be feasible and safe. However, the factors that lead to an improvement in RE are still debated. Upcoming trials
should focus on comparisons of amateur and elite athletes, further increases of exercise frequency (e.g., twice
daily), and the specific mechanisms underlying hopping induced RE improvements, inter alia including altered
tendon stiffness and substrate utilisation.
Data availability
The data that support the findings of this study are available from the corresponding author upon reasonable
request.
Received: 25 November 2022; Accepted: 1 March 2023
References
1. Weiss, M., Newman, A., Whitmore, C. & Weiss, S. One hundred and fifty years of sprint and distance running—Past trends and
future prospects. Eur. J. Sport Sci. 16(4), 393–401. https:// doi. org/ 10. 1080/ 17461 391. 2015. 10425 26 (2016).
2. García-Pinillos, F. et al. Strength training habits in amateur endurance runners in Spain: Influence of athletic level. Int. J. Environ.
Res. Pub. Health 17(21), 8184. https:// doi. org/ 10. 3390/ ijerp h1721 8184 (2020).
3. Denadai, B. S., de Aguiar, R. A., de Lima, L. C. R., Greco, C. C. & Caputo, F. Explosive training and heavy weight training are
effective for improving running economy in endurance athletes: A systematic review and meta-analysis. Sports Med. (Auckland,
N.Z.) 47(3), 545. https:// doi. org/ 10. 1007/ s40279- 016- 0604-z (2017).
4. Jones, A. M. & Carter, H. The effect of endurance training on parameters of aerobic fitness. Sports Med. 29(6), 373–386. https://
doi. org/ 10. 2165/ 00007 256- 20002 9060- 00001 (2000).
5. Engeroff, T. et al. Intensity related changes of running economy in recreational level distance runners. J. Sports Med. Phys. Fitness
57(9), 1111–1118. https:// doi. org/ 10. 23736/ S0022- 4707. 16. 06403-3 (2017).
6. Conley, D. L. & Krahenbuhl, G. S. Running economy and distance running performance of highly trained athletes. Med. Sci. Sports
Exerc. 12(5), 357–360 (1980).
7
Vol.:(0123456789)
Scientific Reports | (2023) 13:4167 |
https://doi.org/10.1038/s41598-023-30798-3
www.nature.com/scientificreports/
7. Barnes, K. R. & Kilding, A. E. Strategies to improve running economy. Sports Med. 45(1), 37–56. https:// doi. org/ 10. 1007/ s40279-
014- 0246-y (2015).
8. Arampatzis, A. et al. Influence of the muscle-tendon unit’s mechanical and morphological properties on running economy. J. Exp.
Biol. 209(17), 3345–3357. https:// doi. org/ 10. 1242/ jeb. 02340 (2006).
9. Hunter, G. R. et al. Tendon length and joint flexibility are related to running economy. Med. Sci. Sports Exerc. 43(8), 1492–1499.
https:// doi. org/ 10. 1249/ MSS. 0b013 e3182 10464a (2011).
10. Kovács, B., Kóbor, I., Sebestyén, Ö. & Tihanyi, J. Longer Achilles tendon moment arm results in better running economy. Phys.
Int. 107(4), 527–541. https:// doi. org/ 10. 1556/ 2060. 2020. 10000 (2021).
11. Kunimasa, Y. et al. Specific muscle-tendon architecture in elite Kenyan distance runners: Achilles tendon moment arm for Kenyan
runners. Scand. J. Med. Sci. Sports 24(4), e269–e274. https:// doi. org/ 10. 1111/ sms. 12161 (2014).
12. Albracht, K. & Arampatzis, A. Exercise-induced changes in triceps surae tendon stiffness and muscle strength affect running
economy in humans. Eur. J. Appl. Physiol. 113(6), 1605–1615. https:// doi. org/ 10. 1007/ s00421- 012- 2585-4 (2013).
13. Kawakami, Y., Muraoka, T., Ito, S., Kanehisa, H. & Fukunaga, T. In vivo muscle fibre behaviour during counter-movement exercise
in humans reveals a significant role for tendon elasticity. J. Physiol. 540(2), 635–646. https:// doi. org/ 10. 1113/ jphys iol. 2001. 013459
(2002).
14. Lai, A., Schache, A. G., Lin, Y.-C. & Pandy, M. G. Tendon elastic strain energy in the human ankle plantar-flexors and its role with
increased running speed. J. Exp. Biol. https:// doi. org/ 10. 1242/ jeb. 100826 (2014).
15. Paxton, J. Z., Hagerty, P., Andrick, J. J. & Baar, K. Optimizing an intermittent stretch paradigm using ERK1/2 phosphorylation
results in increased collagen synthesis in engineered ligaments. Tissue Eng. Part A 18(3–4), 277–284. https:// doi. org/ 10. 1089/ ten.
tea. 2011. 0336 (2012).
16. Schmidt, J. B., Chen, K. & Tranquillo, R. T. Effects of intermittent and incremental cyclic stretch on ERK signaling and collagen
production in engineered tissue. Cell. Mol. Bioeng. 9(1), 55–64. https:// doi. org/ 10. 1007/ s12195- 015- 0415-6 (2016).
17. Ramirez-Campillo, R. et al. Effects of jump training on physical fitness and athletic performance in endurance runners: A meta-
analysis. J. Sports Sci. 39(18), 2030–2050. https:// doi. org/ 10. 1080/ 02640 414. 2021. 19162 61 (2021).
18. Saunders, P. U. et al. Short-term plyometric training improves running economy in highly trained middle and long distance run-
ners. J. Strength Cond. Res. 20(4), 947. https:// doi. org/ 10. 1519/R- 18235.1 (2006).
19. Saunders, P. U., Pyne, D. B., Telford, R. D. & Hawley, J. A. Reliability and variability of running economy in elite distance runners.
Med. Sci. Sports Exerc. 36(11), 1972–1976. https:// doi. org/ 10. 1249/ 01. MSS. 00001 45468. 17329. 9F (2004).
20. Meyer, T., Georg, T., Becker, C. & Kindermann, W. Reliability of gas exchange measurements from two different spiroergometry
systems. Int. J. Sports Med. 22(8), 593–597. https:// doi. org/ 10. 1055/s- 2001- 18523 (2001).
21. Cohen J. (2013) Statistical power analysis for the behavioral sciences: Routledge
22. Turner, A. M., Owings, M. & Schwane, J. A. Improvement in running economy after 6 weeks of plyometric training. J. Strength
Cond. Res. 17(1), 60. https:// doi. org/ 10. 1519/ 1533- 4287(2003) 017% 3c0060: IIREAW% 3e2.0. CO;2 (2003).
23. Fletcher, J. R., Esau, S. P. & MacIntosh, B. R. Changes in tendon stiffness and running economy in highly trained distance runners.
Eur. J. Appl. Physiol. 110(5), 1037–1046. https:// doi. org/ 10. 1007/ s00421- 010- 1582-8 (2010).
24. Saunders, P. U., Pyne, D. B., Telford, R. D. & Hawley, J. A. Factors affecting running economy in trained distance runners. Sports
Med. 34(7), 465–485. https:// doi. org/ 10. 2165/ 00007 256- 20043 4070- 00005 (2004).
25. Wilke, J., Vogel, O. & Vogt, L. Why are you running and does it hurt? Pain, motivations and beliefs about injury prevention among
participants of a large-scale public running event. Int. J Environ. Res Pub Health 16(19), 3766. https:// doi. org/ 10. 3390/ ijerp h1619
3766 (2019).
26. Boone, J., Deprez, D. & Bourgois, J. Running economy in elite soccer and basketball players: Differences among positions on the
field. Int. J. Perform. Anal. Sport 14(3), 775–787. https:// doi. org/ 10. 1080/ 24748 668. 2014. 11868 757 (2014).
Author contributions
Conception and design: K.K., J.W.; Acquisition of data: K.K., R.M.; Analysis and interpretation of the data: T.E.,
J.W.; Drafting of the article: T.E., K.K., J.W.; Critical revision of the article for important intellectual content:
R.M., D.G.; Final approval of the article: all authors.
Funding
Open Access funding enabled and organized by Projekt DEAL.
Competing interests
The authors declare no competing interests.
Additional information
Correspondence and requests for materials should be addressed to K.K.
Reprints and permissions information is available at www.nature.com/reprints.
Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and
institutional affiliations.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International
License, which permits use, sharing, adaptation, distribution and reproduction in any medium or
format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the
Creative Commons licence, and indicate if changes were made. The images or other third party material in this
article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the
material. If material is not included in the article’s Creative Commons licence and your intended use is not
permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from
the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/.
© The Author(s) 2023
| Progressive daily hopping exercise improves running economy in amateur runners: a randomized and controlled trial. | 03-13-2023 | Engeroff, Tobias,Kalo, Kristin,Merrifield, Ryan,Groneberg, David,Wilke, Jan | eng |
PMC10382573 | 1
Vol.:(0123456789)
Scientific Reports | (2023) 13:12244
| https://doi.org/10.1038/s41598-023-38328-x
www.nature.com/scientificreports
Relationship between
biomechanics and energy cost
in graded treadmill running
Marcel Lemire 1,3, Robin Faricier 2, Alain Dieterlen 1,3, Frédéric Meyer 4,5* &
Grégoire P. Millet 4*
The objective of this study was to determine whether the relationships between energy cost of
running (Cr) and running mechanics during downhill (DR), level (LR) and uphill (UR) running could
be related to fitness level. Nineteen athletes performed four experimental tests on an instrumented
treadmill: one maximal incremental test in LR, and three randomized running bouts at constant
speed (10 km h−1) in LR, UR and DR (± 10% slope). Gas exchange, heart rate and ground reaction forces
were collected during steady-state. Subjects were split into two groups using the median Cr for all
participants. Contact time, duty factor, and positive external work correlated with Cr during UR (all,
p < 0.05), while none of the mechanical variables correlated with Cr during LR and DR. Mechanical
differences between the two groups were observed in UR only: contact time and step length were
higher in the economical than in the non-economical group (both p < 0.031). This study shows that
longer stance duration during UR contributes to lower energy expenditure and Cr (i.e., running
economy improvement), which opens the way to optimize specific running training programs.
Although complex physiological and biomechanical factors play important roles in level running (LR), the energy
cost of running (Cr) appears as one of the three main predictive factors determining performance1. The Cr rep-
resents the amount of energy required per unit of kilometer at a given submaximal running velocity allowing to
maintain a physiological steady state2. Though, the influence of LR Cr on graded running performance remains
unclear3–6. It has been reported that LR cost of locomotion is a poor indictor of performance in short distance
trail races3,4 but the importance of energy cost on ultramarathon remains debated5,6. A relationship has been
observed between the oxygen cost (amount of oxygen consumed per distance unit) in uphill running (UR) and
LR in elite ultra-trail runners7. However, additional specific parameters such as knee extensor muscle endurance
and UR Cr may play a role on inclined running performance4.
It has been suggested that changes in the running pattern from negative to positive slopes explain the positive
linear increase in Cr with positive slope8–11, but not in downhill running (DR), where the relationship between
the slopes and Cr has an U-shape with the lowest Cr value at approximately − 10 to − 20% slope9,12. While each
individual naturally develops their optimal running pattern (i.e., spatiotemporal parameters of stride, running
gait) according to their personal characteristics in order to lower their Cr13,14, it is well known that changing this
self-selected running pattern may alter the Cr15–18. Therefore, one may suggest that the most economical runners
efficiently adapt their running mechanics to the slope condition. Though, which biomechanical adaptations are
associated with a lower Cr remains an open question. According to experimental data, DR involves braking
muscle actions of the lower limbs and is considered a predominantly eccentric exercise modality19. In contrast,
UR predominantly involves concentric propulsive muscle contractions19. While the physiological adaptations
to hilly terrain are currently being widely investigated, the main performance determinants for LR, UR and DR
may differ, with a greater contribution of biomechanical parameters in DR performance20.
Compared to LR, UR induces a decrease of aerial time and step length, whereas DR increases the aerial time
and the step length. However, the contact phase is less affected by slope, leading to an increase of step frequency
during UR9,10,21,22. Furthermore, the ratio of positive to negative work is another important biomechanical factor
that may explain the slope-dependent variations in Cr9,23. The mechanical positive and negative external works
(Wext
+ and Wext
−, respectively) represent the work performed at each step to support the upward and downward
OPEN
1Faculty of Sport Sciences, University of Strasbourg, Strasbourg, France. 2School of Kinesiology, The University of
Western Ontario, London, ON, Canada. 3Institut de Recherche en Informatique, Mathématiques, Automatique Et
Signal, Université de Haute-Alsace, 68070 Mulhouse, France. 4Institute of Sport Sciences, University of Lausanne,
Lausanne, Switzerland. 5Digital Signal Processing Group, Department of Informatics, University of Oslo, Oslo,
Norway. *email: [email protected]; [email protected]
2
Vol:.(1234567890)
Scientific Reports | (2023) 13:12244 |
https://doi.org/10.1038/s41598-023-38328-x
www.nature.com/scientificreports/
movements of the center of mass of the body (CoM), respectively24. As the downward movement of the CoM
decreases with positive slopes, Wext
− decreases and the Wext
+ increases, and conversely with negative slopes the
downward movement increases whereas the upward decreases24. Since the energy required to perform Wext
− is
less than that of Wext
+ 25, the more positive the slope, the higher the concentric muscle actions for the elevation
of the CoM, leading to an increase in energy expenditure8. Conversely, the energy demand in DR is lowered due
to the increased part of the eccentric muscle activation and the gravity effect, saving energy8. Nevertheless, the
direct relationship between running mechanics and Cr in UR or DR is under investigated.
The stride kinematic adaptations in graded running may also influence Cr, as debated in LR13. One study
has investigated the relationship between running economy and spatiotemporal running parameters within
specific slope conditions in a homogeneous group of well-trained runners and reported correlations between
spatiotemporal parameters only in DR12. The Cr in DR was negatively correlated with both step frequency and
step length while positively correlated with contact time12. The step length and frequency, and the vertical stiffness
were negatively correlated whereas ground contact time was positively correlated with Cr in DR12. Conversely,
Lussiana et al.14 showed that minimal shoes reduced contact time and increased aerial phase whatever the slope
condition (± 8% slope), while Cr was not affected. Taken together these results tend to show that biomechanical
responses may affect Cr in graded running. However, to the best of our knowledge, no study examined these
relationships, including ground reaction forces, in a large heterogeneous group.
Moreover, the running pattern appears to be dependent on the fitness level in LR. Lower vertical forces were
observed during the stance phase in a group of runners with the lowest oxygen consumption for a given speed
(~ 13 km h−1)17. The magnitude of peak vertical force determines the work performed by the leg muscles to sup-
port the running motion. During incline running, these peak vertical forces have been shown to increase and
decrease during DR and UR, respectively21. Nevertheless, to our knowledge, no previous study has attempted to
determine whether peak vertical forces are also a factor of Cr in incline running (DR and/or UR).
Identification of key running pattern parameters associated with low Cr may have direct practical applications
such as developing grade-specific training methods to improve running technique and potentially performance.
Thus, it appears interesting to investigate whether specific biomechanical responses can distinguish economical
and less economical runners.
Therefore, the objectives of this study were, first, to determine if there was a relationship between Cr and
mechanical responses associated with the running pattern; and second, to determine if these biomechanical
responses were different between two groups of different Cr levels (economical vs. non-economical). Our hypoth-
eses were first that Cr values would correlate with their biomechanical responses; and second that biomechanical
responses would be different between two groups of different economy levels in each slope condition.
Methods
Participants.
Nineteen volunteer athletes took part in this study (Table 1) and were informed of the benefits
and risks of this investigation before giving their written informed consent. They performed between one and
five session per week of running training but were not trail specialists. The experiment was previously approved
by our Institutional Review Board (CCER-VD 2015-00006) and complied with the Declaration of Helsinki.
Experimental setup.
All participants completed (1) a level running (0% slope) incremental test to exhaus-
tion; and (2) three randomized running bouts at constant velocity (10 km h−1) with different slope conditions,
LR, UR (+ 10%) and DR (− 10%). The running speed of 10 km h−1 was selected to ensure that subjects were below
the second ventilatory threshold in each slope condition. Participants performed all the sessions on a treadmill
(T-170-FMT, Arsalis, Belgium) at the same time of the day with 1 week of recovery allocated. The subjects were
instructed to not perform any eccentric and/or strenuous exercises in this time interval.
Maximal incremental level running test.
The first session was an incremental running test until
exhaustion. The test began at 8 km h−1 for 4 min and then the speed increased by 1 km h−1 every min. During
each session, V̇O2, carbon dioxide output (V̇CO2), and respiratory exchange ratio (RER) were collected breath-
by-breath through a facemask with an open-circuit metabolic cart with rapid O2 and CO2 analyzers (Quark
Table 1. Participant characteristics (n = 19). vV̇O2max velocity associated to V̇O2max, VT1 and VT2 V̇O2 at the
first and the second ventilatory thresholds, respectively, HRmax maximal heart rate.
Age (years)
34
±
10
Height (cm)
175
±
10
Body mass (kg)
68.5
±
12.2
BMI (kg m−2)
22.2
±
2.3
vV̇O2max (km h−1)
17.3
±
2.3
V̇O2max (mlO2 kg−1 min−1)
58.3
±
7.7
VT1 (mlO2 kg−1 min−1)
40.8
±
4.9
VT2 (mlO2 kg−1 min−1)
54.0
±
7.3
HRmax (bpm)
179
±
12
3
Vol.:(0123456789)
Scientific Reports | (2023) 13:12244 |
https://doi.org/10.1038/s41598-023-38328-x
www.nature.com/scientificreports/
CPET, Cosmed, Rome, Italy) in order to calculate the Cr. Heart rate (HR) was continuously measured (Polar
Electro, Kempele, Finland). The highest V̇O2 value over 30 s during the maximal incremental test represented
the V̇O2max. The speed associated with V̇O2max (vV̇O2max) was determined as the speed of the step that elicited
V̇O2max
26. The first ventilatory threshold was determined as a breakpoint in the plot of V̇CO2 as a function of
V̇O2. At that point, the ventilatory equivalent for O2 (V̇E/V̇O2) increases without an increase in ventilatory equiv-
alent for CO2 (V̇E/V̇CO2)27. The second ventilatory threshold was located between the first ventilatory threshold
and V̇O2max, when V̇E/V̇CO2 starts to increase while V̇E/V̇O2 continues to rise28. These thresholds were blind
assessed by two accustomed experimenters. The average value was kept, and in case of a difference above 30 s, a
third experimenter was involved, and the average of the two closest values was used. The rate of perceived exer-
tion was obtained by using a designed scale29 to assess the exercise intensity about 30 s after the end of the test.
During the second session, after a short warm-up participant performed three randomized constant velocity
running bout of 4 min. As for the maximal incremental test, V̇O2, V̇CO2 and RER continuously recorded. Before
each session, the O2 and CO2 analyzers were calibrated according to the manufacturer’s instructions.
Metabolic power during constant velocity bouts in level, uphill, and downhill running.
Mean
Cr values were recorded between 3:15 and 3:45 (min:s) of each running bout. The Cr was computed as following12:
where Cr is expressed in J kg−1 m−1, ΔV̇O2 for the difference between oxygen consumption at steady-state and
oxygen consumption at baseline in mlO2 kg−1 min−1 30, v corresponded to the velocity of the trial (10 km h−1), and
E(O2) for O2 energy equivalent determined with RER. As the V̇O2 response is slope-dependent in running31, for
each slope condition (i.e., LR, UR and DR), the subjects were arbitrarily divided into two groups (i.e., economi-
cal vs. non-economical) based on the absolute Cr median value (2.42, 3.83, and 6.09 J kg−1 m−1 for DR, LR, and
UR, respectively), to obtain equal proportion of runners within each group17.
Biomechanical data collection and processing.
An instrumented treadmill equipped with a three-
dimensional force platform sampling 1000 Hz was used in this study. To reduce the noise inherent to the tread-
mill’s vibrations, we first applied, a second order stop-band Butterworth filter with edge frequencies set at 25 and
65 Hz, on the vertical ground reaction force signal. The filter configuration was chosen empirically to obtain
a satisfactory reduction of the oscillations observed during flight phases while minimizing its widening effect
during ground contact time. Further data analysis was conducted using MATLAB software version R2021a
(MathWorks Inc., Natick, MA, USA). The instants of initial contact and terminal contact were identified using
a threshold of 7% of bodyweight on the filtered vertical ground reaction force signal32, and ~ 80 steps were ana-
lyzed for each condition. The contact time (in ms) is the time between initial and terminal contacts of the same
leg, the aerial time (in ms) is the time between the terminal contact of one leg and the initial contact of the oppo-
site leg. Duty factor (expressed in %) was computed as the ratio between the contact time and the stride time (i.e.,
contact time + aerial time). The step frequency (in Hz) is the reciprocal of the time required for one step (time
between two consecutive initial contacts). The step length (m) is the quotient of the treadmill belt speed divided
by step frequency. Peak vertical ground reaction force (GRF) was computed over the entire stance phase. The
Wext was determined using the method proposed by Saibene and Minetti33 and is defined as the sum of potential,
and horizontal and vertical kinetic works associated with the displacement of the CoM. The Wext
− and Wext
+
represent the work done due to decelerate and accelerate, respectively, the body’s CoM with respect to the envi-
ronment. The percentage of negative work is the ratio between the Wext
− and the total external work. These data
were continuously recorded during 30 s between 3:15 and 3:45 (min:s) of each constant velocity running bouts.
Statistical analysis.
Jamovi statistical software (Jamovi 1.6.23, Sydney; Australia) was used for all statisti-
cal analyses. All variables were examined for normality using a Shapiro–Wilk. A repeated measures ANOVA
was performed to compare the effect of the slope’s condition on the Cr and the biomechanical data, after using
Mauchly’s test to assess sphericity. Bonferroni’s correction was applied on the alpha level to account for repeated
univariate testing. When significant effects were observed, Bonferroni’s post-hoc tests were used to localize the
significant differences. For each condition of slope, scale intercept and Pearson’s product–moment correlation
coefficients (r) were used to assess the intensity of the relations between Cr and the selected biomechanical vari-
ables, with Bonferroni’s multiplicity correction33. A one-way ANOVA was used to compare the biomechanical
responses on the treadmill between efficiency groups. For all these analyses, data are expressed as mean ± SD and
a p value inferior to 0.05 was considered statistically significant.
Ethics approval.
This study was performed in line with the principles of the Declaration of Helsinki.
Approval was granted by the Ethics Committee by our Institutional Review Board (CCER-VD 2015-00006).
Consent to participate.
Informed consent was obtained from all individual participants included in the
study.
Results
Cost of locomotion and biomechanics.
Values of Cr and biomechanical parameters in the different
slope conditions are presented in Table 2. The contact time was negatively correlated with the Cr in UR only
(r = − 0.54; p = 0.017; Fig. 1). For both UR and LR only, the aerial time was positively correlated (r = 0.54 and
r = 0.57, respectively; both p ≤ 0.018; Fig. 1), while the duty factor was negatively correlated with the Cr (r = − 0.50
Cr = V·O2/(v × 1000) × 60 × E(O2)
4
Vol:.(1234567890)
Scientific Reports | (2023) 13:12244 |
https://doi.org/10.1038/s41598-023-38328-x
www.nature.com/scientificreports/
and r = − 0.57, respectively; both p ≤ 0.029; Fig. 1). The relative peak force was correlated with the Cr in LR and
UR only (r = 0.50 and r = 0.56, respectively; both p ≤ 0.031) but not in DR. Regarding mechanical work param-
eters, the Cr correlated to Wext
+ in UR (r = 0.49; p = 0.035), but none of the mechanical work parameters corre-
lates with Cr in DR and UR. In addition, Cr was significantly positively correlated between each slope condition:
DR-LR (r = − 0.57; p = 0.011), DR-UR (r = − 0.53; p = 0.020), and LR-UR (r = − 0.72; p < 0.001).
Economical and non-economical runners.
All running pattern parameters were similar between eco-
nomical and non-economical groups in both DR and LR (Table 3). The contact time, step length, and mass-
specific peak vertical GRF (Fig. 2A, B, D) were higher while the step frequency (Fig. 2C) was lower in the eco-
nomical than in the non-economical group in UR (all p < 0.031). However, aerial time, GRF, or mechanical work
values were not different between the two groups in UR (all p > 0.05).
Discussion
This study provides new insights into biomechanical factors according to the level of metabolic economy of
runners on different running slopes. The main findings of the present study are that (1) aerial time, peak verti-
cal ground reaction force, positive mechanical work, and duty factor were associated with Cr during LR and
UR; while contact time correlated only with Cr during UR; (2) no relationship was observed between Cr and
biomechanical responses during DR; and (3) most economical runners tend to have specific running pattern
adaptations (i.e., longer contact time and greater step length as well as a lower step frequency) during UR only.
These results partially support the hypothesis since specific running pattern responses were characterized by a
lower metabolic cost of running on flat or positive slopes but not on negative slope.
During UR, Cr positively correlated with several running pattern parameters such as Wext
+, aerial time, and
mass specific GRF, and negatively correlated with contact time and duty factor. These correlations demonstrate
that UR is related to specific running pattern parameters and highlight that the running pattern may influence
Cr, especially during UR. Our results revealed that Cr, Wext
−, Wext
+, duty factor, and step frequency were lower;
whereas areal time, step length (in absolute and relative to the height), GRF (expressed in both absolute and rela-
tive values), and the percentage of the Wext
− were higher during LR than during UR, which is rather consistent
with the literature23. Furthermore, as already observed on similar slopes8,9,34, the running economy was reduced
when running on the positive slope.
Most of the mechanical work performed comes from positive external mechanical work during UR (~ 69% of
Wext was provided by Wext
+ and only ~ 31% by Wext
−; Table 2). Comparable distribution on equivalent slope and
speed was reported9. These results confirm that UR is primarily a concentric muscle contraction that is energy-
consuming25. Indeed, the Wext
+ represents the amount of mechanical energy spent during the pushing phase to
elevate and move forward the CoM24.
It has been highlighted that the increase in Wext
+ was caused by the elevation of CoM related to the upward
movement of the body during UR24. The external work is the product of vertical displacement of the CoM and
the step frequency. It was observed that lower vertical displacements of the CoM and/or a higher step frequency
in LR were associated with better running economy35. Therefore, one could potentially expect a similar relation-
ship during UR. However, no correlation was observed between step frequency and Cr, possibly because runners
choose their own optimal step frequency and step length, whatever the slope level16, close to their minimal Cr36.
Nevertheless, other spatiotemporal parameters such as the contact time and aerial time were correlated with Cr
during UR. Contact time was negatively correlated to the Cr, meaning that a longer contact was associated with
Table 2. Cost of locomotion and biomechanical parameters in downhill, level and uphill running (n = 19).
GRF peak ground reaction force. a p < 0.05 versus downhill running. b p < 0.05 versus level running.
Downhill running
Level running
Uphill running
F-STATISTICS
Energy cost of running
Energy cost (J kg−1 m−1)
2.56 ± 0.51
3.87 ± 0.43a
6.21 ± 0.51ab
674
Running kinematics and GRF
Contact time (s)
0.28 ± 0.03
0.29 ± 0.03
0.29 ± 0.03
3.73
Aerial time (s)
0.09 ± 0.03
0.08 ± 0.02a
0.07 ± 0.02ab
25.9
Duty factor (%)
74.9 ± 7.6
78.1 ± 5.8a
81.5 ± 5.2ab
22.6
Step frequency (Hz)
2.66 ± 0.10
2.71 ± 0.12
2.81 ± 0.11ab
26.6
Step length (m)
1.05 ± 0.04
1.03 ± 0.05
0.99 ± 0.04ab
26.2
Relative step length (% of height)
60.0 ± 3.8
58.8 ± 3.8
56.7 ± 3.3ab
25.1
Peak vertical GRF (kN)
1.41 ± 0.20
1.35 ± 0.19a
1.29 ± 0.18ab
21.8
Mass-specific vertical GRF (N kg−1)
20.8 ± 2.1
19.8 ± 1.5a
19.0 ± 1.2ab
22.1
Mechanical work
External work (J kg1)
3.06 ± 0.44
2.75 ± 0.29a
2.76 ± 0.19a
17.5
Positive external work (J kg−1)
1.07 ± 0.23
1.42 ± 0.15a
1.91 ± 0.11ab
367
Negative external work (J kg−1)
− 1.99 ± 0.22
− 1.34 ± 0.14a
− 0.85 ± 0.09ab
737
Percentage negative external work (%)
65.3 ± 2.3
48.5 ± 0.2a
30.8 ± 1.1ab
2270
5
Vol.:(0123456789)
Scientific Reports | (2023) 13:12244 |
https://doi.org/10.1038/s41598-023-38328-x
www.nature.com/scientificreports/
a better running economy. This result is in agreement with the observation made by Vernillo et al.37 after an
ultramarathon event (330 km with 24,000 m elevation gain). The UR Cr (4 min at 6 km h−1 and + 15% incline)
was negatively correlated with contact time, duty factor, and as well as rate of force application (characterized
by inverse contact time: tc
−1). For instance, shorter contact time reduces the time allowed to generate force into
the ground and increases the rate at which the muscle fibers shorten38,39, so more fast muscle fibers or muscle
mass should be required for a given applied force37,38. The step frequency depends on the interrelationship
Figure 1. Relationships between the cost of locomotion and contact time (A), aerial time (B), duty factor (C),
mass-specific peak vertical ground reaction force (D), and positive external mechanical work (E) in different
slope conditions (DR downhill running—unfiled circles, LR level running—filled diamonds, UR uphill
running—*p < 0.05).
6
Vol:.(1234567890)
Scientific Reports | (2023) 13:12244 |
https://doi.org/10.1038/s41598-023-38328-x
www.nature.com/scientificreports/
Table 3. Comparison of the cost of locomotion and biomechanical responses for economical versus non-
economical runners in the three slope conditions (n = 19). GRF peak ground reaction force. Significant values
are in [bold].
Downhill running
Level running
Uphill running
Economical N = 9
Non-economical
N = 10
p
Economical N = 9
Non-economical
N = 10
p
Economical N = 9
Non-economical
N = 10
p
Energy cost of running
Energy cost of run-
ning (J kg−1 m−1)
2.15 ± 0.20
2.93 ± 0.42
< .001
3.52 ± 0.18
4.20 ± 0.32
< .001
5.82 ± 0.27
6.55 ± 0.42
< .001
Running kinematics and GRF
Contact time (s)
0.28 ± 0.03
0.29 ± 0.03
0.697
0.30 ± 0.04
0.28 ± 0.02
0.199
0.30 ± 0.02
0.28 ± 0.02
0.013
Aerial time (s)
0.10 ± 0.03
0.09 ± 0.02
0.621
0.07 ± 0.02
0.09 ± 0.02
0.063
0.06 ± 0.01
0.07 ± 0.02
0.130
Duty factor (%)
74.1 ± 8.9
75.7 ± 6.5
0.654
80.5 ± 5.8
76.0 ± 5.3
0.093
83.7 ± 4.1
79.5 ± 5.5
0.080
Step frequency (Hz)
2.65 ± 0.09
2.66 ± 0.12
0.900
2.71 ± 0.16
2.70 ± 0.09
0.864
2.75 ± 0.10
2.86 ± 0.10
0.026
Step length (m)
1.05 ± 0.04
1.05 ± 0.05
0.932
1.03 ± 0.06
1.03 ± 0.03
0.947
1.01 ± 0.04
0.97 ± 0.03
0.038
Relative step length
(% of height)
60.1 ± 2.4
59.8 ± 4.8
0.871
58.3 ± 3.3
59.3 ± 4.4
0.560
56.5 ± 3.5
56.8 ± 3.2
0.854
Peak vertical GRF
(kN)
1.43 ± 0.14
1.39 ± 0.2
0.619
1.36 ± 0.16
1.34 ± 0.22
0.875
1.34 ± 0.14
1.25 ± 0.2
0.252
Mass-specific verti-
cal GRF (N kg−1)
21.1 ± 2.5
20.5 ± 1.7
0.573
19.2 ± 1.5
20.4 ± 1.4
0.102
18.5 ± 0.9
19.5 ± 1.3
0.075
Mechanical work
External work
(J kg−1)
3.13 ± 0.47
3.00 ± 0.43
0.537
2.69 ± 0.23
2.81 ± 0.34
0.368
2.71 ± 0.14
2.80 ± 0.23
0.317
Positive external
work (J kg−1)
1.11 ± 0.25
1.04 ± 0.22
0.502
1.38 ± 0.12
1.45 ± 0.18
0.363
1.87 ± 0.08
1.93 ± 0.13
0.255
Negative external
work (J kg−1)
− 2.02 ± 0.23
− 1.97 ± 0.21
− 1.30 ± 0.11
− 1.36 ± 0.17
0.375
− 0.84 ± 0.06
− 0.87 ± 0.11
0.468
Percentage negative
external work (%)
64.9 ± 2.5
65.7 ± 2.1
0.448
48.5 ± 0.2
48.5 ± 0.2
0.652
30.8 ± 0.8
30.8 ± 1.3
0.996
Figure 2. Contact time (A), step length (B), step frequency (C) and mass-specific ground reaction force (D) in
economical (white box-plots) and non-economical runners (gray box-plots) during uphill running.
7
Vol.:(0123456789)
Scientific Reports | (2023) 13:12244 |
https://doi.org/10.1038/s41598-023-38328-x
www.nature.com/scientificreports/
between contact time and step length. Thus, for a given step frequency, increasing the aerial time will shorten
the contact time and lead to Cr deterioration, since the metabolic cost of force generation increases as the con-
tact time shortens38. In addition, as it exists an inverse relationship between the step frequency and the vertical
oscillation of CoM in running35, Furthermore, reducing aerial time and vertical displacement (which will occur
together) are the result of less external positive work (resulting in a reduced vertical velocity at takeoff) during
UR. However, an excessive increase of the stride frequency during running to reduce mechanical work could
be disadvantageous as it causes a Cr raise16. The negative correlation observed between Cr and the duty factor
might underline the importance of optimizing energy transfer (from metabolic to mechanics) to reduce energy
demands in UR. Altogether, increasing the stance phase and decreasing the aerial time may be an appropriate
strategy for improving running economy during UR. As such, patterns of locomotion may play a decisive role
in lowering the cost of locomotion by walking compared to running on a steep positive slope40. Indeed, walking
pattern is characterized by a longer contact time, a higher duty factor, and a lower stride frequency associated
with reduced muscle activation compared to the running pattern on a 30° slope41.
The present data showed a relationship between the Cr and GRF. Normalized GRF to body weight was
positively correlated with Cr during UR. Since GRF is the result of the forces produced by all the muscles in the
vertical direction during the stance phase, an excessive GRF value in this orientation is a waste of energy. Thus,
minimizing the vertical GRF seems to be more economical during UR. Increasing step frequency could lower
the vertical GRF and might be a useful strategy in UR. Therefore, the adoption of strategies to reduce vertical
GRF forces should be incorporated in training programs in order to improve running economy during UR as
well as during LR. However, such adjustments must be individually adapted.
In the present study, we confirm that negative slope has a significant effect on the running pattern and
decreases Cr compared to LR42. Total mechanical work, Wext
−, proportion of Wext
−, aerial time, and GRF
(expressed in absolute and mass-specific values), increased in DR while the Wext
+ decreased, in agreement with
the literature9,23,24,43. However, conversely to UR, downhill Cr was not correlated with any mechanical aspects,
suggesting that, at least for the present velocity and slope, there was not a more economical running pattern
during DR. These results are not in agreement with previous results12 which reported a significant correlation
between Cr and several spatiotemporal parameters such as step length, step frequency, and contact time during
DR (− 15%). According to these later results, it was suggested that the ability to store and restitute elastic energy
had an important role in DR Cr12. The difference observed in the literature may come from differences in the
experimental design and the fitness level of the participants. As observed by Minetti et al.9 more than half (~ 65%)
of the total external work is provided by Wext
−. The latter represents the work done during the braking phase of
the stance phase. During this phase, the knee extensor muscles forcibly lengthen (i.e., eccentric muscle action)
under the potential effect of gravity to limit the drop-down of the CoM. From an energetical point of view, this
eccentric muscle’s action requires less energy than a concentric muscle contraction25, and part of the potential
energy from the vertical oscillation of the CoM is either dissipated as wasted heat (mostly) or stored in the mus-
cle–tendon units during the braking phase prior its restitution during the pushing phase. The stretch–shorten-
ing cycle is mainly involved during DR16,44 and is known to be less energy consuming than purely concentric
actions45, saving energy and reducing the Cr as well45. For moderate negative slopes (~ 15%), the elastic energy
stored in the muscle–tendon units can supply almost all the energy demand for the push phase44. However, no
correlation was observed between Wext
− and the Cr.
Runners have their own running style based on ability and experience which implies that a similar Cr from
one individual to another can be associated with different biomechanical parameters. Indeed, there was only a
relatively small influence of each of the parameters measured on CR, even when the relationship was significant
(Fig. 1). Therefore, we have to be cautious when suggesting potential gait modifications for performance enhance-
ment. Changing one parameter of the running pattern can alter the overall mechanics and potentially the running
economy46. For example, ± 15% changes in preferred step frequency increased Cr by ~ 20% in DR16. Running in
negative slopes is demanding for the body, as greater braking force must be applied on the ground to maintain
a constant speed, which can generate muscle damage47. Furthermore, since force absorption is less energetically
demanding, runners may neglect their running mechanics. A more protective running pattern is privileged by
runners based on their experience in DR9. Running on negative slopes where the fear of falling is higher could
also exacerbate emotional aspect, when compared to LR and UR. During treadmill running, irrespective of the
slope, it is well-known that the mechanics is different than during overground running: the influence of the
motion belt that affects both potential and kinetic works remains difficult to be accurately assessed48. Therefore,
we have to be cautious for translating the present findings to field running.
The present study compared the metabolic and biomechanical responses of economical and non-economical
runners at different slopes. Individual Cr values in the three slope conditions were used to split participants into
two groups. We showed that the most economical runners remained the same ones, independently of the slope
(i.e., in DR, LR or UR). This result is rather consistent with the literature: Willis et al.7 reported a strong cor-
relation between Cr measured in LR and in UR (12% slope) in a group of elite ultra-trail runners (6 males and
5 females), while Balducci et al.49 found no correlation between LR and UR (12.5 or 25% slope) in trained trail
runners49. Moreover, in the present study, there were differences in biomechanical responses between the two
groups, but only in UR economical runners had longer contact time and step length compared to less economical
runners, while their step frequency was smaller. Ultimately, the longer duration of the stance phase may allow
runners to optimize the direction of propulsive force and the time allowed to apply force to the ground50. Indeed,
mass-specific GRF tends to be lower in the economical runners than in the non-economical group (p = 0.070;
Table 3), suggesting that lower mass-specific GRF in UR may allow to reduce the metabolic cost of running. A
lower vertical force during the stance phase was observed for the group of runners who had the lowest oxygen
consumption for a given speed (~ 13 km h−1) on flat terrain17.
8
Vol:.(1234567890)
Scientific Reports | (2023) 13:12244 |
https://doi.org/10.1038/s41598-023-38328-x
www.nature.com/scientificreports/
No difference for all the biomechanical variables was observed in DR and LR between the groups of economi-
cal and non-economical runners, i.e., with different Cr. This result is rather in line with the literature13,14, and
may be partly explained by the heterogeneity of the population in the present study. Less experienced runners
tend to have a greater stride to stride variability than experienced runners51. The Cr may be influenced by many
other factors, such as anthropometry, flexibility, and joint kinematics but also by physiological differences (e.g.,
metabolic efficiency) or equipment (e.g., running shoes)18. Runners naturally chose their optimal running pat-
tern themselves to minimize their Cr13. Each one having its own specificity, the number of mechanical combina-
tions is likely very important. Nevertheless, even if two groups use different running strategies, there were no
significant differences in Cr13,14.
Conclusions
The present study reported that Cr was related to few key running pattern parameters (i.e., contact time, aerial
time, mass specific GRF and positive mechanical external work) mainly in UR, but not in DR. Moreover, all run-
ning pattern parameters were similar between economical and non-economical runners in DR and LR, but not
in UR. Interestingly, the contact time and the step length were longer, whereas the step frequency was lower in
the group of economical runners compared to the group of non-economical runners in UR. These results provide
interesting insights concerning an optimal running pattern to reduce the cost of locomotion, and consequently
improve performance during graded running. In practice, it may be preferable to reduce step frequency, or even
to shift to walking, on a positive slope to increase step length and slow down the knee extension during the pro-
pulsive phase. On steep slopes, poles could facilitate this mechanism52. Overall, the present study emphasizes
that the mechanics of LR and UR are fundamentally different. Future investigations are needed to deepen the
knowledge with a heterogeneous population (trained or untrained runners) to improve the training protocols
of mountain or trail runners.
Data availability
The datasets used and/or analyzed during the current study are available from the corresponding author on
reasonable request.
Received: 13 October 2022; Accepted: 6 July 2023
References
1. di Prampero, P. E., Atchou, G., Brückner, J.-C. & Moia, C. The energetics of endurance running. Eur. J. Appl. Physiol. 55, 259–266
(1986).
2. Fletcher, J. R., Esau, S. P. & MacIntosh, B. R. Economy of running: Beyond the measurement of oxygen uptake. J. Appl. Physiol.
107, 1918–1922 (2009).
3. Björklund, G., Swarén, M., Born, D.-P. & Stöggl, T. Biomechanical adaptations and performance indicators in short trail running.
Front. Physiol. 10, 506 (2019).
4. Ehrström, S. et al. Short trail running race: Beyond the classic model for endurance running performance. Med. Sci. Sports Exerc.
50, 580–588 (2018).
5. Millet, G. P. Economy is not sacrificed in ultramarathon runners. J. Appl. Physiol. 113, 686–686 (2012).
6. Millet, G. Y., Hoffman, M. D. & Morin, J. B. Sacrificing economy to improve running performance—A reality in the ultramarathon?.
J. Appl. Physiol. 113, 507–509 (2012).
7. Willis, S. J. et al. Level versus uphill economy and mechanical responses in elite ultratrail runners. Int. J. Sports Physiol. Perform.
14, 1001–1005 (2019).
8. Minetti, A. E., Moia, C., Roi, G. S., Susta, D. & Ferretti, G. Energy cost of walking and running at extreme uphill and downhill
slopes. J. Appl. Physiol. (Bethesda Md) 1985(93), 1039–1046 (2002).
9. Minetti, A., Ardigò, L. & Saibene, F. Mechanical determinants of the minimum energy cost of gradient running in man. J. Exp.
Biol. 195, 211–225 (1994).
10. Lussiana, T., Fabre, N., Hébert-Losier, K. & Mourot, L. Effect of slope and footwear on running economy and kinematics. Scand.
J. Med. Sci. Sports 23, e246–e253 (2013).
11. Vernillo, G. et al. Biomechanics and physiology of uphill and downhill running. Sports Med. Auckl. NZ 47, 615–629 (2017).
12. Lemire, M. et al. Energy cost of running in well-trained athletes: Toward slope-dependent factors. Int. J. Sports Physiol. Perform.
https:// doi. org/ 10. 1123/ IJSPP. 2021- 0047 (2021).
13. Patoz, A., Lussiana, T., Breine, B., Gindre, C. & Hébert-Losier, K. There is no global running pattern more economic than another
at endurance running speeds. Int. J. Sports Physiol. Perform. 1, 1–4 (2022).
14. Lussiana, T. et al. Similar running economy with different running patterns along the aerial-terrestrial continuum. Int. J. Sports
Physiol. Perform. 12, 481–489 (2017).
15. Tseh, W., Caputo, J. L. & Morgan, D. W. Influence of gait manipulation on running economy in female distance runners. J. Sports
Sci. Med. 7, 91–95 (2008).
16. Snyder, K. L. & Farley, C. T. Energetically optimal stride frequency in running: the effects of incline and decline. J. Exp. Biol. 214,
2089–2095 (2011).
17. Williams, K. R. & Cavanagh, P. R. Relationship between distance running mechanics, running economy, and performance. J. Appl.
Physiol. (Bethesda Md) 1985(63), 1236–1245 (1987).
18. Saunders, P. U., Pyne, D. B., Telford, R. D. & Hawley, J. A. Reliability and variability of running economy in elite distance runners.
Med. Sci. Sports Exerc. 36, 1972–1976 (2004).
19. Lindstedt, S. L., LaStayo, P. C. & Reich, T. E. When active muscles lengthen: Properties and consequences of eccentric contractions.
Physiology 16, 256–261 (2001).
20. Ehrström, S. et al. Acute and delayed neuromuscular alterations induced by downhill running in trained trail runners: Beneficial
effects of high-pressure compression garments. Front. Physiol. 9, 1627 (2018).
21. Gottschall, J. S. & Kram, R. Ground reaction forces during downhill and uphill running. J. Biomech. 38, 445–452 (2005).
22. Padulo, J. et al. Uphill running at iso-efficiency speed. Int. J. Sports Med. 33, 819–823 (2012).
23. Vernillo, G. et al. Biomechanics of graded running: Part I—Stride parameters, external forces, muscle activations. Scand. J. Med.
Sci. Sports 30, 1632–1641 (2020).
9
Vol.:(0123456789)
Scientific Reports | (2023) 13:12244 |
https://doi.org/10.1038/s41598-023-38328-x
www.nature.com/scientificreports/
24. Dewolf, A. H., Peñailillo, L. E. & Willems, P. A. The rebound of the body during uphill and downhill running at different speeds.
J. Exp. Biol. 219, 2276–2288 (2016).
25. Abbott, B. C., Bigland, B. & Ritchie, J. M. The physiological cost of negative work. J. Physiol. 117, 380–390 (1952).
26. Billat, L. V. & Koralsztein, J. P. Significance of the velocity at VO2max and time to exhaustion at this velocity. Sports Med. Auckl.
NZ 22, 90–108 (1996).
27. Beaver, W. L., Wasserman, K. & Whipp, B. J. A new method for detecting anaerobic threshold by gas exchange. J. Appl. Physiol.
60, 2020–2027 (1986).
28. Whipp, B. J., Davis, J. A. & Wasserman, K. Ventilatory control of the ‘isocapnic buffering’ region in rapidly-incremental exercise.
Respir. Physiol. 76, 357–367 (1989).
29. Borg, G. Perceived exertion as an indicator of somatic stress. J. Rehabil. Med. 2, 92–98 (1970).
30. Lemire, M., Falbriard, M., Aminian, K., Millet, G. P. & Meyer, F. Level, uphill, and downhill running economy values are correlated
except on steep slopes. Front. Physiol. 12, 697315 (2021).
31. Lemire, M. et al. Trail runners cannot reach V˙O2max during a maximal incremental downhill test. Med. Sci. Sports Exerc. 52,
1135–1143 (2020).
32. Falbriard, M., Meyer, F., Mariani, B., Millet, G. P. & Aminian, K. Accurate estimation of running temporal parameters using foot-
worn inertial sensors. Front. Physiol. 9, 610 (2018).
33. Saibene, F. & Minetti, A. E. Biomechanical and physiological aspects of legged locomotion in humans. Eur. J. Appl. Physiol. 88,
297–316 (2003).
34. Margaria, R., Cerretelli, P., Aghemo, P. & Sassi, G. Energy cost of running. J. Appl. Physiol. 18, 367–370 (1963).
35. Halvorsen, K., Eriksson, M. & Gullstrand, L. Acute effects of reducing vertical displacement and step frequency on running
economy. J. Strength Cond. Res. 26, 2065–2070 (2012).
36. de Ruiter, C. J., Verdijk, P. W. L., Werker, W., Zuidema, M. J. & de Haan, A. Stride frequency in relation to oxygen consumption in
experienced and novice runners. Eur. J. Sport Sci. 14, 251–258 (2014).
37. Vernillo, G. et al. Influence of the world’s most challenging mountain ultra-marathon on energy cost and running mechanics. Eur.
J. Appl. Physiol. 114, 929–939 (2014).
38. Roberts, T. J., Kram, R., Weyand, P. G. & Taylor, C. R. Energetics of bipedal running. I. Metabolic cost of generating force. J. Exp.
Biol. 201, 2745–2751 (1998).
39. Cavanagh, P. R. & Kram, R. Mechanical and muscular factors affecting the efficiency of human movement. Med. Sci. Sports Exerc.
17, 326–331 (1985).
40. Ortiz, A. L. R., Giovanelli, N. & Kram, R. The metabolic costs of walking and running up a 30-degree incline: Implications for
vertical kilometer foot races. Eur. J. Appl. Physiol. 117, 1869–1876 (2017).
41. Whiting, C. S., Allen, S. P., Brill, J. W. & Kram, R. Steep (30°) uphill walking vs. running: COM movements, stride kinematics, and
leg muscle excitations. Eur. J. Appl. Physiol. 120, 2147–2157 (2020).
42. Bontemps, B., Vercruyssen, F., Gruet, M. & Louis, J. Downhill running: What are the effects and how can we adapt? A narrative
review. Sports Med. (Auckl. NZ) 50, 2083–2110 (2020).
43. Vernillo, G. et al. Biomechanics and physiology of uphill and downhill running. Sports Med. 47, 615–629 (2017).
44. Snyder, K. L., Kram, R. & Gottschall, J. S. The role of elastic energy storage and recovery in downhill and uphill running. J. Exp.
Biol. 215, 2283–2287 (2012).
45. Belli, A. & Bosco, C. Influence of stretch-shortening cycle on mechanical behaviour of triceps surae during hopping. Acta Physiol.
Scand. 144, 401–408 (1992).
46. Moore, I. S. Is there an economical running technique? A review of modifiable biomechanical factors affecting running economy.
Sports Med. (Auckl. Nz) 46, 793–807 (2016).
47. Giandolini, M. et al. Fatigue associated with prolonged graded running. Eur. J. Appl. Physiol. 116, 1859–1873 (2016).
48. Firminger, C. R. et al. Joint kinematics and ground reaction forces in overground versus treadmill graded running. Gait Posture
63, 109–113 (2018).
49. Balducci, P. et al. Comparison of level and graded treadmill tests to evaluate endurance mountain runners. 9 (2016).
50. Kugler, F. & Janshen, L. Body position determines propulsive forces in accelerated running. J. Biomech. 43, 343–348 (2010).
51. Mo, S. & Chow, D. H. K. Stride-to-stride variability and complexity between novice and experienced runners during a prolonged
run at anaerobic threshold speed. Gait Posture 64, 7–11 (2018).
52. Giovanelli, N., Sulli, M., Kram, R. & Lazzer, S. Do poles save energy during steep uphill walking?. Eur. J. Appl. Physiol. 119,
1557–1563 (2019).
Acknowledgements
The authors thank the subjects for their enthusiastic participation in the present study.
Author contributions
G.P.M. and F.M. participated in the design of the study and contributed to data collection, M.L., A.D. and R.F.
contributed to the data reduction/analysis; All authors contributed to the manuscript writing. All authors have
read and approved the final version of the manuscript and agree with the order of presentation of the authors.
Funding
The authors did not receive support from any organization for the submitted work.
Competing interests
The authors declare no competing interests.
Additional information
Correspondence and requests for materials should be addressed to F.M. or G.P.M.
Reprints and permissions information is available at www.nature.com/reprints.
Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and
institutional affiliations.
10
Vol:.(1234567890)
Scientific Reports | (2023) 13:12244 |
https://doi.org/10.1038/s41598-023-38328-x
www.nature.com/scientificreports/
Open Access This article is licensed under a Creative Commons Attribution 4.0 International
License, which permits use, sharing, adaptation, distribution and reproduction in any medium or
format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the
Creative Commons licence, and indicate if changes were made. The images or other third party material in this
article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the
material. If material is not included in the article’s Creative Commons licence and your intended use is not
permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from
the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/.
© The Author(s) 2023
| Relationship between biomechanics and energy cost in graded treadmill running. | 07-28-2023 | Lemire, Marcel,Faricier, Robin,Dieterlen, Alain,Meyer, Frédéric,Millet, Grégoire P | eng |
PMC4154741 | Is the COL5A1 rs12722 Gene Polymorphism Associated
with Running Economy?
Roˆ mulo Bertuzzi1,2*, Leonardo A. Pasqua2, Saloma˜o Bueno2, Adriano Eduardo Lima-Silva3,
Monique Matsuda4, Monica Marquezini1, Paulo H. Saldiva1
1 Laboratory of Experimental Air Pollution, Department of Pathology, School of Medicine, University of Sa˜o Paulo, Sa˜o Paulo, Brazil, 2 Endurance Sports Research Group,
School of Physical Education and Sport, University of Sa˜o Paulo, Sa˜o Paulo, Brazil, 3 Sports Science Research Group, Academic Center of Vitoria, Federal University of
Pernambuco, Vitoria de Santo Anta˜o, Brazil, 4 Faculty of Medicine, University of Sa˜o Paulo, Sa˜o Paulo, Brazil
Abstract
The COL5A1 rs12722 polymorphism is considered to be a novel genetic marker for endurance running performance. It has
been postulated that COL5A1 rs12722 may influence the elasticity of tendons and the energetic cost of running. To date,
there are no experimental data in the literature supporting the relationship between range of motion, running economy,
and the COL5A1 rs12722 gene polymorphism. Therefore, the main purpose of the current study was to analyze the influence
of the COL5A1rs12722 polymorphism on running economy and range of motion. One hundred and fifty (n = 150) physically
active young men performed the following tests: a) a maximal incremental treadmill test, b) two constant-speed running
tests (10 kmNh21 and 12 kmNh21) to determine the running economy, and c) a sit-and-reach test to determine the range of
motion. All of the subjects were genotyped for the COL5A1 rs12722 single-nucleotide polymorphism. The genotype
frequencies were TT = 27.9%, CT = 55.8%, and CC = 16.3%. There were no significant differences between COL5A1 genotypes
for running economy measured at 10 kmNh21 (p = 0.232) and 12 kmNh21 (p = 0.259). Similarly, there were no significant
differences between COL5A1 genotypes for range of motion (p = 0.337). These findings suggest that the previous
relationship reported between COL5A1 rs12722 genotypes and running endurance performance might not be mediated by
the energetic cost of running.
Citation: Bertuzzi R, Pasqua LA, Bueno S, Lima-Silva AE, Matsuda M, et al. (2014) Is the COL5A1 rs12722 Gene Polymorphism Associated with Running
Economy? PLoS ONE 9(9): e106581. doi:10.1371/journal.pone.0106581
Editor: Mikko Lammi, University of Eastern Finland, Finland
Received January 30, 2014; Accepted August 8, 2014; Published September 4, 2014
Copyright: 2014 Bertuzzi et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits
unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Funding: Romulo Bertuzzi is supported by CAPES (grant 23038.000486/2011-15). Leonardo A. Pasqua is supported by FAPESP (grant 2010/13913-6). Saloma˜o
Bueno is supported by CAPES (grant 1182744). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the
manuscript.
Competing Interests: The authors have declared that no competing interests exist.
* Email: [email protected]
Introduction
It is well known that athletic performance is dependent on a
multifactorial phenotype resulting from a complex interaction
between environmental [1,2] and genetic [3,4] factors. In the past
few years, researchers have analyzed the influence of some genes
that encode proteins that are involved in metabolic pathways [5]
or skeletal muscle structure [4,6], which are important to athletic
performance. More recently, Posthumus et al. [7] reported that
endurance running performance is associated with a gene, named
COL5A1, which encodes a structural protein of the extracellular
matrix. These authors observed that the TT genotype was
overrepresented in runners with faster performances during the
42.2-km running portion of the Ironman triathlon, suggesting
thatCOL5A1 might be a novel genetic marker for endurance
running performance.
It is believed that the relationship between COL5A1 and
endurance running performance might be mediated by the energy
storage capacity of the connective tissues [7,8]. COL5A1 encodes
the a1 (V) chain of type V collagen [9], which plays a crucial role
in the regulation of the size and configuration of other abundant
fibrillar collagens supporting many tissues in the body, such as
tendons, ligaments, and muscles [10]. A mutation in the COL5A1
gene, which causes haploinsufficiency, results in a 50% reduction
of type V collagen, and leads to poorly organized fibrils, decreased
tensile strength, and reduced stiffness of connective [9]. It has been
demonstrated that a common C to T single nucleotide polymor-
phism within the COL5A1 39 untranslated region (rs12722) may
alter COL5A1 mRNA stability and thereby reduce the production
of type V collagen [11]. Thus, it was hypothesised that individuals
with a TT genotype of this variant would have increased type V
collagen production and thus favorably altered mechanical
properties of tendons, which enhances endurance running ability
[7,8].
Running economy (RE), which has been defined as the energy
cost or oxygen uptake for a given submaximal running speed
[12,13], is considered to be an important determinant of
endurance performance [14,15]. It has been indicated that the
inter-individual variation in RE is approximately 10–25% between
homogeneous subjects for maximal oxygen uptake ( :VO2 max )
[15]. The cause of this variation is not well understood, but it
might be partially explained by the energy storage capacity of the
tendons [14]. Indeed, elastic energy is generated by the passive
stretch of the muscle elastic elements, and this energy is converted
to kinetic energy free of metabolic cost during a running bout [16].
Thus, the most economical runners have a higher compliance of
the tendons and aponeuroses compared to their less economical
PLOS ONE | www.plosone.org
1
September 2014 | Volume 9 | Issue 9 | e106581
counterparts [14]. Because of this reported association between
RE and the energy storage capacity of the tendons, it has been
proposed that individuals with a TT genotype would have less
extensible tendon structures, resulting in a lower range of motion
(ROM) and greater RE when compared with individuals with at
least one copy of the C allele [8]. However, to date there are no
experimental data in the literature supporting this relationship
among RE, ROM, and the COL5A1 rs12722 gene polymorphism.
Consequently, it is difficult to determine whether the association
previously reported between the COL5A1 gene and endurance
performance was due to a reduced running energy cost.
Therefore, the main objective of the present study was to
analyze the influence of the COL5A1 gene polymorphism on RE.
In light of past studies that reported a relation between the
COL5A1 gene (rs12722) and endurance running performance
[7,8], it was hypothesized that individuals with a TT genotype
would have a greater RE and a lower ROM when compared to
individuals with the TC and CC genotypes.
Materials and Methods
Participants
One hundred and fifty (n = 150) physically active men (age
25.264.0 years, body mass 77.8613.9 kg, height 173.9621.4 cm,
and body fat 13.364.2%) volunteered to participate in this study.
All subjects were medication-free, nonsmokers, and free of
neuromuscular disorders and cardiovascular dysfunctions. The
subjects had been involved in recreational sports in the past year,
but they had not engaged in flexibility and strength training for at
least six months before the study. The participants received a
verbal
explanation
about
the
possible
benefits,
risks,
and
discomforts associated with the study and signed a written
informed consent before participation in the study. The proce-
dures were in accordance with the Helsinki declaration of 1975,
and the study was approved by the Ethics Committee for Human
Studies of the School of Physical Education and Sport of the
University of Sa˜o Paulo.
Experimental design
All of the subjects were required to visit the laboratory on three
occasions separated by at least 72 h over a 2-week period. In the
first visit, the subjects were asked to perform mouthwashes for
genomic DNA extraction, to perform a sit-and-reach test for
ROM measurement, and to fill out a short version of the physical
activity questionnaire (IPAQ-short version) to estimate their
physical activity levels. In the second visit, anthropometric
measurements (height, body mass, and body composition) were
recorded and a maximal incremental treadmill test was performed.
A constant-speed, treadmill running familiarization test was
conducted at the end of the first and second visits after a 20-min
passive recovery. In the third session, the subjects performed two
submaximal constant-speed tests. All of the tests were performed at
the same time of day in a controlled-temperature room (20–24uC)
and 2–3 h after the last meal. All of the subjects were asked to
refrain from any exhaustive or unaccustomed exercise for 48 h
preceding the test. They also were instructed to wear standard
running shoes and asked from taking off nutritional supplements
six months before the experimental period.
Anthropometric measurements
All of the anthropometric measurements were made according
to the procedures described by Norton and Olds [17]. The
subjects were weighed using an electronic scale to the nearest
0.1 kg. Height was measured with a stadiometer to the nearest
0.1 cm. Skinfold thickness was measured at seven sites (chest,
axilla, triceps, subscapular, abdominal, suprailiac, and thigh) with
a Harpenden caliper (West Sussex, UK) to the nearest 0.2 mm.
The seven skinfold thickness values were obtained on the right side
of the body in a serial fashion, and the median of three values was
used for data analysis. When the difference between these three
values was higher than 10%, a fourth measurement was obtained.
All measurements were made by the same experienced investiga-
tor. Body density was estimated using the generalized equation of
Jackson and Pollock [18], and body fat was estimated using the
equation of Brozek et al. [19].
Maximal incremental treadmill test
The subjects performed a maximal incremental test on a motor-
driven treadmill (model TK35, CEFISE, Nova Odessa) to
determine maximal oxygen uptake ( :VO2 max ). After a 3-min
warm-up at 8 km?h21, the velocity was increased by 1 km?h21
every minute until exhaustion. The participants received strong
verbal encouragement to ensure attainment of maximal values.
Gas exchange was measured breath-by-breath using a gas analyzer
(Cortex Metalyzer 3B, Cortex Biophysik, Leipzig, Germany) and
was subsequently averaged over 20 s intervals throughout the test.
Before each test, the gas analyzer was calibrated according to the
recommendations of the manufacturer. Maximal heart rate
(HRMAX) was defined as the highest value obtained at the end
of the test. Blood samples (25 ml) were collected from the ear lobe
one, three, and five minutes after the test to determine the peak
blood lactate. Lactate concentrations were measured spectropho-
tometrically (EONC, Biotek Instruments, USA) using a wave-
length of 546 nm. :VO2 max was determined when two or more of
the following criteria were met: an increase in oxygen uptake of
less than 2.1 ml?kg21?min21 between two consecutive stages, a
respiratory exchange ratio greater than 1.1, and a 610 bpm of the
predicted maximal heart rate (i.e., 220-age) [20].
Running economy
The subjects performed two constant-speed running tests on a
treadmill to determine RE. Because it has been reported that a
subject who is economical at a given speed will not be necessarily
economical at other speeds [21], we measured the RE at10
km?h21 and 12 km?h21 speeds. These intensities corresponded to
78.866.7% and 89.767.9% of the :VO2 max , respectively. Due
the different percentage values required of the :VO2 max , it was
assumed that these intensities represented the RE at low
(RELW = 10 km?h21) and high (REHG = 12 km?h21) intensities.
Before the tests, the participants performed a standardized warm-
up consisting of a 5 min run at 8 km?h21 followed by a 5-min of
passive recovery. RE was determined by measuring breath-by-
breath oxygen uptake for 10 minutes at each running speed. RE
was defined by averaging the oxygen uptake values during the last
20 s for each running speed. Recovery time between these two
constant-speeds running tests was 10 min.
Range of motion
Range of motion (ROM) was measured using a sit-and-reach
test [22,23]. This test has been used in previous studies
investigating the effects of the COL5A1 genotypes on lower limb
flexibility because it represents an indirect measure of both
hamstring musculotendinous unit length and lumbar ROM
[25,26]. The subjects sat with their bare feet pressed against the
sit-and-reach box. The knees were extended and the right hand
was positioned over the left. Then, the subjects were asked to push
a ruler transversally located over the box as far as possible on the
COL5A1 and Running Economy
PLOS ONE | www.plosone.org
2
September 2014 | Volume 9 | Issue 9 | e106581
fourth bobbing movement. Three trials were performed, and the
best trial was used for statistical analysis.
Physical activity level determination
Because the RE may be influenced by the training status of
the subjects [27], the short version of the International Physical
Activity Questionnaire (IPAQ-SV) was used to identify the
physical activity level of the participants. This questionnaire was
developed with a multi-cultural adaptation and is considered to
be one of the most widely utilized instruments due to the
quickness of data collection, low operating cost, and non-
invasive characteristics [28]. In addition, it has recently been
shown that IPAQ-SV outcomes are associated with flexibility
and cardiorespiratory fitness levels in healthy men [29]. The
participants answered the questionnaire in a classroom setting
after a detailed description of the IPAQ-SV. An assistant
remained in the classroom setting for eventual doubts. The
major aim of the IPAQ-SV is to sum up walking, moderate and
vigorous PA and to generate a total PA score for weekly
expenditure, expressed in metabolic equivalent task units (METs
min/wk). We used the following recommended METs min/wk
estimates of the IPAQ-SF: walking PA = 3.3 METs min/wk,
moderate PA = 4.0 METs min/wk, vigorous PA = 8.0 METs
min/wk. The total PA was calculated assuming: 3.36walking
PA+4.06moderate PA+8.06vigorous PA. The PA level was
classified as low, moderate or high. Low activity represented
individuals who do not meet the criteria for moderate and
vigorous intensity categories (,599 METs min/wk). Moderate
activity represented moderate or vigorous intensity activities
achieving a minimum of at least 600 METs min/wk. High
activity represented participants achieving a minimum of at least
3000 METs min/wk (http://www.ipaq.ki.se/scoring.htm).
Genotype assessment
Cells from the mouthwashes were submitted to an overnight
digestion with proteinase K. Nucleotides were separated from
the cellular debris by density gradient centrifugation using
chloroform. Genomic DNA was then precipitated with isopropyl
alcohol, isolated by centrifugation and resuspended in TE
buffer. DNA quantification was performed using a spectropho-
tometer (NanoDrop, ND 2000, USA), and the concentration
was adjusted to 1 mg/mL for subsequent storage in 220uC.
COL5A1 rs12722 gene polymorphisms were determined by
conventional 2-primer PCR (F: 59-GCAGTCAGCAGCGTGG-
GTCTGGTTATCT-39 and R: 59-TTTGGGGTGGCACTTG-
CAGCACT GGTCG-39). This assay resulted in the amplifica-
tion of a 637-bp fragment of the COL5A1 gene that includes
the polymorphic region. The reaction conditions were as follow:
initial hold at 94uC for 3 min, 35 cycles of denaturation at
94uC for 60 s, annealing at 53uC for 60 s and extension at
72uC for 60 s, and a final extension step of 8 min at 72uC. The
amplified fragment was subsequently digested using BstUI (New
England, Biolabs, Beverly, MA, USA), following the supplier’s
recommendations. The digested products were then separated
on a 3% agarose gel. To ensure proper internal control, for
each batch of analysis, we used positive and negative controls
from different DNA aliquots that were previously genotyped by
the same method, according to recent recommendations for
replicating genotype-phenotype association studies [30]. The
restriction fragment length polymorphism (RFLP) results were
scored by three experienced and independent investigators who
were blinded to the participant’s data.
Statistical analysis
Data normality was assessed using the Kolmogorov-Smirnov
test, and all of the variables showed a normal distribution. The
results are reported as means and standard deviations (6SD). A
X2 test was used to verify that the genotype frequencies were in
Hardy-Weinberg
equilibrium.
The
effects
of
the
COL5A1
genotypes in the analyzed variables were tested using a one-way
analysis of variance (ANOVA). The significance level was set at
p,0.05. All of the statistical analyses were performed using
Statistica 8 (StataSoft Inc., Tulsa, OK, USA).
Results
Genotype distribution and sample characteristics
The
genotype
distribution
attained
the
Hardy-Weinberg
Equilibrium,
as
evidenced
by
the
X2
test.
The
genotype
frequencies were TT = 27.9%, CT = 55.8% and CC = 16.3%.
The
genotype
distribution
of
the
COL5A1
rs12722
gene
polymorphism was similar to the distribution reported in the
public databases for Caucasian populations (http://opensnp.org/
snps/rs12722). In accordance with the mean values of the IPAQ-
SV outcomes, the subjects of the three groups were classified as
moderately active [31]. There were no differences in age,
anthropometric characteristics, or physical activity levels among
the CC, CT, and TT genotypes of the COL5A1 gene (p.0.05)
(table 1).
Maximal incremental, running economy, and range of
motion tests
Table 2 shows the mean values of the physiological variables
measured during the maximal incremental test. There were no
significant differences in
:VO2 max , HRMAX, and [La]peak
between individuals with different COL5A1 genotypes (p.0.05).
Figure 1 shows the results of the constant-speed running and sit-
and-reach
tests.
The
RELW
(p = 0.232)
(panel
A),
REHG
(p = 0.259) (panel B), and ROM (p = 0.337) (panel C) were not
Table 1. Anthropometric characteristics and physical activity levels of the subjects in the three genotype groups.
CC (n = 24)
CT (n = 84)
TT (n = 42)
P values
Age (years)
25.763.7
21.364.2
23.865.0
0.873
Body mass (kg)
80.5610.7
76.169.9
75.2610.4
0.230
Height (cm)
177.865.4
172.666.5
174.265.8
0.243
Body fat (%)
14.363.8
13.264.2
13.964.6
0.584
IPAQ-SV (score)
12916237
13566289
13336301
0.605
Data are means 6 standard deviations. IPAQ-SV: short version of the international physical activity questionnaire. There were no differences between the groups.
doi:10.1371/journal.pone.0106581.t001
COL5A1 and Running Economy
PLOS ONE | www.plosone.org
3
September 2014 | Volume 9 | Issue 9 | e106581
statistical
different
between
the
CC
(RELW = 35.976
3.18 ml?kg21?min21; REHG=40.9864.59 ml?kg21?min21; ROM=
26.1468.34 cm), CT (RELW=36.0362.93 ml?kg21?min21; REHG=
41.8063.36 ml?kg21?min21;
ROM=27.8568.22 cm),
and
TT
(RELW=37.6563.37 ml?kg21?min21; REHG=42.6663.79 ml?kg21?min21;
ROM = 27.0866.88 cm) COL5A1 genotypes groups.
Discussion
Previous findings suggested that the COL5A1 gene might be a
marker of endurance running performance [24,7]. It has been
speculated that the superior running ability of individuals with a
TT genotype could be explained by a greater RE when compared
with individuals with at least a C allele [8]. However, it is
important to notice that these studies did not investigate the
relationship between the energetic cost of running and COL5A1-
genotypes because the subjects did not perform constant-speed
tests to measure the RE. To the best of our knowledge, this is the
first study designed to analyze the relationship between a common
C to T single-nucleotide polymorphism gene within COL5A1 gene
(rs12722) and RE. The major findings of this study show that
theCOL5A1 genotypes were not associated with either RE or
ROM.
It has been proposed that mechanical properties of connective
tissues are responsible for converting elastic energy to kinetic
energy free of metabolic cost [14]. It was demonstrated that more
economical subjects have a higher contractile strength and
compliance of the tendons and aponeuroses when compared with
their less economical counterparts [14]. Some studies have
suggested that the mechanical properties of tendons and ligaments,
which largely consists of collagen fibrils, may be genetically
determined [32,10]. In particular, the genetic variation in the
COL5A1 gene, which encodes type V collagen, may affect the
mechanical properties of tendons and ligaments through altering
the mechanical properties. It has previously been shown that the
COL5A1 gene variant investigated in this study was associated
with endurance running ability. Although the specific mechanisms
remain largely unknown, we hypothesised in this study that this
gene variant improves endurance running ability through an
improving running economy. However, the present findings did
not confirm the relationship between COL5A1 genotypes and RE.
Our results showed no significant differences in RELW between
individuals with different COL5A1 genotypes. This finding
suggests that the superior performance observed in individuals
with a TT genotype in the COL5A1gene (rs12722) cannot be
explained by a lower energy cost during running at a low
percentage of :VO2 max.
In the present study, we considered that a subject who is
economical at low relative running intensity might not be
economical at high relative running intensity. Our results
demonstrated that COL5A1 genotypes were not associated with
the energetic cost of running regardless of running intensity. An
alternative explanation for the relationship between COL5A1
genotypes and endurance running performance might be the
ability to produce force rather than a reduced energetic cost.
Recently, Kubo et al. [33] demonstrated that individuals with a T
allele had a higher stiffness of the knee extensors compared to
those individuals with at least one copy of the C allele. It was
previously
demonstrated
that
dynamic
muscle
actions
are
positively related to stiffness, possibly due a more effective force
transmission from the contractile elements to the bone [34]. In
turn, the ability to produce force has been considered an essential
determinant of endurance performance without being necessarily
related to RE [35,36]. This occurs because the horizontal
component of ground reaction force is fundamental for endurance
runners to attain high-intensity running speeds [35]. Thus,
individuals with a TT genotype may be able to maintain higher
running speeds during long-distance events than their counterparts
with a CT or CC genotype by applying greater forces to the
ground. Nevertheless, this hypothesis should be analyzed with
caution because we were unable to obtain ground reaction forces
Figure 1. Running economy, range of motion, and COL5A1
(rs12722) genotypes. Panel A: running economy at low intensity,
Panel B: running economy at high intensity, Panel C: range of motion.
There were no differences between the groups.
doi:10.1371/journal.pone.0106581.g001
COL5A1 and Running Economy
PLOS ONE | www.plosone.org
4
September 2014 | Volume 9 | Issue 9 | e106581
in the present study. Thus, further research is needed to examine
the underlying mechanisms determining the relationship between
COL5A1 genotypes and endurance running performance.
It has also been postulated that COL5A1 rs12722 genotypes
could alter the elasticity of tendons [10] and contribute to
explaining the inter-individual variation in ROM [36] additionally
to other polymorphisms (i.e. COL5A1 rs71746744) of this gene
[37]. However, conflicting results were reported in the literature
regarding the relationship between COL5A1 gene polymorphisms
and ROM. Some studies found that the CC genotype was
overrepresented in individuals with greater ROM [24], while
others found no relationship [25]. In the present study, the ROM
values between individuals with CC, CT, and TT genotypes were
not significantly different. This result is in agreement with the
earlier findings of Brown et al. [25] who found that COL5A1
genotypes in a South African cohort were not associated with
ROM in young subjects (,35 years). However, these authors
found a positive correlation between COL5A1genotypes and
ROM in older subjects (.38 years old) [26]. It is important to note
that besides our cohort consisting of younger subjects (25.264.0
years), a significant difference was not detected for age between
individuals with different COL5A1 genotypes. Taken together,
these findings reinforce the suggestion that COL5A1 rs12722
genotypes might interact with age for ROM in physically active
subjects [26].
It is important to acknowledge some of the limitations of the
present study. First, the subjects in the present investigate on were
characterized as physically active, as evidenced by the IPAQ-SV
outcomes. This would imply that our subjects might have a lower
capacity for endurance exercise than the trained endurance
runners that were previously studied [24,7]. Thus, caution should
be exercised in extrapolating these findings for highly-trained
endurance runners. Second, the present study was conducted on a
relatively small sample size. Therefore, our findings need to be
confirmed in a larger cohort of subjects. On the other hand, it is
important to notice that in the present study no differences for
physical activity levels, age, and anthropometric measurements
were observed among the three COL5A1 (rs12722) genotypes
(Table 1). In addition, our cohort was composed exclusively of
men. This seems to be especially important because training status
[38], age [39], body weight [27], and sex [36] all influence RE.
Therefore, it is reasonable to assume that most of the potential
confounding variables were controlled in the current study.
In conclusion, the results of the current study demonstrated that
variants within the COL5A1gene were not associated with RE and
ROM. This indicates that the previous relationship reported
between COL5A1 genotypes and endurance running performance
may not be mediated by the energetic cost of running. Therefore,
further studies are needed to examine the causal relationship
between COL5A1 gene and endurance running performance.
Author Contributions
Conceived and designed the experiments: RB LAP SB AEL-S PHS.
Performed the experiments: RB SB LAP M. Matsuda M. Marquezini.
Analyzed the data: M. Matsuda M. Marquezini RB SB LAP. Contributed
reagents/materials/analysis tools: M. Matsuda M. Marquezini PHS.
Wrote the paper: RB LAP SB AEL-S M. Matsuda M. Marquezini PHS.
References
1. Onywera VO, Scott RA, Boit MK, Pitsiladis YP (2006) Demographic
characteristics of elite Kenyan endurance runners. J Sports Sci 24(4): 415–22.
2. Scott RA, Georgiades E, Wilson RH, Goodwin WH, Wolde B, et al. (2003)
Demographic characteristics of elite Ethiopian endurance runners. Med Sci
Sports Exerc 35(10): 1727–32.
3. Go´mez-Gallego F, Ruiz JR, Buxens A, Artieda M, Arteta D, et al. (2009) The 2
786 T/C polymorphism of the NOS3 gene is associated with elite performance
in power sports. Eur J Appl Physiol 107(5): 565–9.
4. Eynon N, Hanson ED, Lucia A, Houweling PJ, Garton F, et al. (2013) Genes for
elite power and sprint performance: ACTN3 leads the way. Sports Med 43(9):
803–17.
5. Cupeiro R, Gonza´lez-Lamun˜o D, Amigo T, Peinado AB, Ruiz JR, et al. (2010)
Influence of the MCT1-T1470A polymorphism (rs1049434) on blood lactate
accumulation during different circuit weight trainings in men and women. J Sci
Med Sport 15(6): 541–7.
6. Eynon N, Duarte JA, Oliveira J, Sagiv M, Yamin C, et al. (2009) ACTN3
R577X polymorphism and Israeli top-level athletes. Int J Sports Med 30(9):
695–8.
7. Posthumus M, Schwellnus MP, Collins M (2011) The COL5A1 gene: a novel
marker of endurance running performance. Med Sci Sports Exerc 43(4): 584–9.
8. Collins M, Posthumus M (2011) Type V collagen genotype and exercise-related
phenotype relationships: a novel hypothesis. Exerc Sport Sci Rev 39(4): 191–8.
9. Wenstrup RJ, Florer JB, Davidson JM, Phillips CL, Pfeiffer BJ, et al. (2006)
Murine model of the Ehlers-Danlos syndrome. col5a1 haploinsufficiency
disrupts collagen fibril assembly at multiple stages. J Biol Chem 281(18):
12888–95.
10. Wenstrup RJ, Smith SM, Florer JB, Zhang G, Beason DP, et al. (2011)
Regulation of collagen fibril nucleation and initial fibril assembly involves
coordinate interactions with collagens V and XI in developing tendon. J Biol
Chem 286(23): 20455–65.
11. Laguette MJ, Abrahams Y, Prince S, Collins M (2011) Sequence variants within
the 39-UTR of the COL5A1 gene alters mRNA stability: implications for
musculoskeletal soft tissue injuries. Matrix Biol 30(5–6): 338–45.
12. Helgerud J, Støren O, Hoff J (2010) Are there differences in running economy at
different velocities for well-trained distance runners? Eur J Appl Physiol 108(6):
1099–105.
13. Williams TJ, Krahenbuhl GS, Morgan DW (1991) Daily variation in running
economy of moderately trained male runners. Med Sci Sports Exerc 23(8): 944–
8.
14. Arampatzis A, De Monte G, Karamanidis K, Morey-Klapsing G, Stafilidis S, et
al. (2006) Influence of the muscle-tendon unit’s mechanical and morphological
properties on running economy. J Exp Biol 209(Pt 17): 3345–57.
15. Ramsbottom R, Nute MG, Williams C (1987) Determinants of five kilometre
running performance in active men and women. Br J Sports Med 21(2): 9–13.
16. Komi PV, Bosco B (1978) Utilization of stored elastic energy in leg extensor
muscles by men and women. Med Sci Sports 10: 261–265.
17. Norton Olds (1996) (eds) Anthropometrica. University of New South Wale Press,
Sydney.
Table 2. Physiological variables measured during the maximal incremental test in the three genotype groups.
CC (n = 24)
CT (n = 84)
TT (n = 42)
P values
:VO2 max
(mL.kg21.min21)
47.165.2
47.466.1
47.665.8
0.966
HRMAX (bpm)
190610
18969
18967
0.940
[La]peak
(mmol.L21)
9.963.7
9.463.9
9.664.1
0.781
Data are means 6 standard deviations. :VO2 max: maximal oxygen uptake, HRMAX = maximal heart rate, [La]peak: peak of blood lactate accumulation. There were no
differences between the groups.
doi:10.1371/journal.pone.0106581.t002
COL5A1 and Running Economy
PLOS ONE | www.plosone.org
5
September 2014 | Volume 9 | Issue 9 | e106581
18. Jackson AS, Pollock ML (1985) Practical assessment of body composition. The
Physican and Sportsmedicine. 19: 76–90.
19. Brozek J, Grande F, Anderson J, Keys A (1963) Densitometric analysis of body
composition: revision of some quantitative assumptions. Ann NY Acad Sci 110:
113–140.
20. Howley ET, Bassett DR Jr, Welch HG (1995) Criteria for maximal oxygen
uptake: review and commentary. Med Sci Sports Exerc 27(9): 1292–301.
21. Kyrolainen H, Kivela R, Koskinen S, McBride J, Andersen JL, et al. (2003)
Interrelationships between muscle structure, muscle strength, and running
economy. Med Sci Sports Exerc 35(1): 45–9.
22. Samogin Lopes FA, Menegon EM, Franchini E, Tricoli V, de M Bertuzzi RC
(2010) Is acute static stretching able to reduce the time to exhaustion at power
output corresponding to maximal oxygen uptake? J Strength Cond Res 24(6):
1650–6.
23. Baltaci G, Un N, Tunay V, Besler A, Gerc¸eker S (2003) Comparison of three
different sit and reach tests for measurement of hamstring flexibility in female
university students. Br J Sports Med 37(1): 59–61.
24. Brown JC, Miller CJ, Posthumus M, Schwellnus MP, Collins M (2011) The
COL5A1 gene, ultra-marathon running performance, and range of motion.
Int J Sports Physiol Perform 6(4): 485–96.
25. Brown JC, Miller CJ, Schwellnus MP, Collins M (2011) Range of motion
measurements diverge with increasing age for COL5A1 genotypes.
Scand J Med Sci Sports 21(6): 266–72.
26. Saunders PU, Pyne DB, Telford RD, Hawley JA (2004) Factors affecting
running economy in trained distance runners. Sports Med 34(7): 465–85.
27. Craig CL, Marshall AL, Sjostrom M, Bauman AE, Booth ML, et al. (2003)
International physical activity questionnaire: 12-country reliability and validity.
Med Sci Sports Exerc 35: 1381–1395, 2003.
28. Silva-Batista C, Urso RP, Lima Silva AE, Bertuzzi R (2013) Associations
between fitness tests and the international physical activity questionnaire-short
form in healthy men. J Strength Cond Res 27(12): 3481–7.
29. Fogelholm M, Malmberg J, Suni Santilla M, Kyro¨la¨inen H, Ma¨ntysaari M, et al.
(2006) International Physical Activity Questionnaire: Validity against fitness.
Med Sci Sports Exerc 38(4): 753–60.
30. Chanock SJ, Thomas G (2007) The devil is in the DNA. Nat Genet 39(3): 283–4.
31. Ficek K, Cieszczyk P, Kaczmarczyk M, Maciejewska-Karłowska A, Sawczuk M,
et al. (2013) Gene variants within the COL1A1 gene are associated with reduced
anterior cruciate ligament injury in professional soccer players. J Sci Med Sport
16(5): 396–400.
32. Kubo K, Yata H, Tsunoda N (2013) Effect of gene polymorphisms on the
mechanical properties of human tendon structures. Springerplus 25(2): 343.
33. Bojsen-Møller J, Magnusson SP, Rasmussen LR, Kjaer M, Aagaard P (2005)
Muscle performance during maximal isometric and dynamic contractions is
influenced by the stiffness of the tendinous structures. J Appl Physiol 99(3): 986–
94.
34. Nummela A, Kera¨nen T, Mikkelsson LO (2007) Factors related to top running
speed and economy. Int J Sports Med 28(8): 655–61.
35. Kyro¨la¨inen H, Belli A, Komi PV (2001) Biomechanical factors affecting running
economy. MedSci Sports Exerc 33(8): 1330–7.
36. Collins M, Mokone GG, September AV, van der Merwe L, Schwellnus MP
(2009) The COL5A1 genotype is associated with range of motion measurements.
Scand J Med Sci Sports 19(6): 803–10.
37. Abrahams S1, Posthumus M, Collins M (2014) A polymorphism in a functional
region of the COL5A1 gene: association with ultraendurance-running
performance and joint range of motion. Int J Sports Physiol Perform 9(3):
583–90.
38. Bransford DR, Howley ET (1977) Oxygen cost of running in trained and
untrained men and women. Med Sci Sports 9(1): 41–4.
39. Arie¨ns GA, van Mechelen W, Kemper HC, Twisk JW (1997) The longitudinal
development of running economy in males and females aged between 13 and 27
years: the Amsterdam Growth and Health Study. Eur J Appl Physiol Occup
Physiol 76(3): 214–20.
COL5A1 and Running Economy
PLOS ONE | www.plosone.org
6
September 2014 | Volume 9 | Issue 9 | e106581
| Is the COL5A1 rs12722 gene polymorphism associated with running economy? | 09-04-2014 | Bertuzzi, Rômulo,Pasqua, Leonardo A,Bueno, Salomão,Lima-Silva, Adriano Eduardo,Matsuda, Monique,Marquezini, Monica,Saldiva, Paulo H | eng |
PMC7959157 | Footwear designed to enhance energy return
improves running economy compared to a minimalist
footwear: does it matter for running performance?
R.C. Dinato1
00 , R. Cruz1,2
00 , R.A. Azevedo1,3
00 , J.S. Hasegawa1
00 , R.G. Silva1
00 , A.P. Ribeiro4
00 ,
A.E. Lima-Silva5
00 , and R. Bertuzzi1
00
1Grupo de Estudo em Desempenho Aeróbio, Escola de Educac¸ão Física e Esporte, Universidade de São Paulo,
São Paulo, SP, Brasil
2Centro de Desportos, Departamento de Educac¸ão Física, Universidade Federal de Santa Catarina, Florianópolis, SC, Brasil
3Human Performance Laboratory, Faculty of Kinesiology, University of Calgary, Calgary, Canada
4Departamento de Pós-Graduac¸ão em Ciências da Saúde, Laborátorio de Biomecânica e Reabilitac¸ão Musculoesquelética,
Universidade de Santo Amaro, São Paulo, SP, Brasil
5Grupo de Pesquisa em Performance Humana, Departamento Acadêmico de Educac¸ão Física, Universidade Tecnológica Federal
do Paraná, Curitiba, PR, Brasil
Abstract
The present study compared the effects of a footwear designed to enhance energy return (thermoplastic polyurethane, TPU) vs
minimalist shoes on running economy (RE) and endurance performance. In this counterbalanced and crossover design study,
11 recreational male runners performed two submaximal constant-speed running tests and two 3-km time-trials with the two
shoe models. Oxygen uptake was measured during submaximal constant-speed running tests in order to determine the RE at
12 km/h and oxygen cost of running (CTO2) at individual average speed sustained during the 3-km running time-trials wearing
either of the two shoes. Our results revealed that RE was improved (2.4%) with TPU shoes compared with minimalist shoes
(P=0.01). However, there was no significant difference for CTO2 (P=0.61) and running performance (P=0.52) comparing the
TPU (710±60 s) and the minimalist (718±63 s) shoe models. These novel findings demonstrate that shoes with enhanced
mechanical energy return (i.e. TPU) produced a lower energy cost of running at low (i.e., 12 km/h) but not at high speeds
(i.e., average speed sustained during the 3-km running time-trial, B15 km/h), ultimately resulting in similar running performance
compared to the minimalist shoe.
Key words: Running shoes; Oxygen uptake; Oxygen cost; Running time-trial; Endurance performance
Introduction
Running endurance performance has been traditionally
associated with several physiological variables, including
running economy (RE) (1–5). Individuals with superior RE,
defined as the steady-state oxygen uptake at submaximal
running speeds (4), are able to sustain higher exercise
intensities and/or maintain the same exercise intensity for a
longer period of time compared to their counterparts with
poorer RE (6). In order to acutely improve RE, previous
studies have suggested that footwear characteristics could
play a significant role on the energy cost of running. For
example, minimalist shoes (i.e., with reduced shoe mass
and heel drop) results in significant improvements in RE
(1–4%) compared to conventional shoes (i.e., with ethylene
vinyl acetate midsole) (7–9). This enhanced RE with
minimalist shoes has been associated with a greater
mechanical action of the longitudinal and transverse
arches of the foot, which are capable of restoring/
returning approximately 17% of the mechanical energy
temporarily stored at each step taken (10), thus acutely
improving the RE at submaximal running.
Other studies have demonstrated that midsole char-
acteristics can also enhance RE (11,12). For instance,
Wunsch et al. (13) showed that a leaf spring-structured
midsole acutely improved RE (B1%), probably by changes
in spatio-temporal variables. More recently, a new midsole
material composed by thermoplastic polyurethane (TPU)
has been used to enhance energy return during running
(14). The TPU appears to reduce oxygen cost of running by
increasing the returned mechanical energy from the shoe
midsole material. In fact, Sinclair et al. (15) have shown that
Correspondence: R. Bertuzzi: <[email protected]>
Received September 29, 2020 | Accepted December 24, 2020
Braz J Med Biol Res | doi: 10.1590/1414-431X202010693
Brazilian Journal of Medical and Biological Research (2021) 54(5): e10693, http://dx.doi.org/10.1590/1414-431X202010693
ISSN 1414-431X
Research Article
1/6
running with a footwear with a midsole composed by
TPU exhibits a better RE (4.1%) compared to a con-
ventional running shoe. Given that RE contributes to the
success in endurance running events, they concluded
that footwear with a TPU midsole could lead to better
running performance, probably due to the beneficial
effects on RE.
Despite the remarkable findings from previous studies
showing that both TPU and minimalist shoes can reduce
the oxygen cost of running compared with conventional
shoes (7,15), it is still unknown whether there is a superior
ability between these two models to translate the greater
RE into better running performance. This occurs mainly
because the current evidence is limited to analyzing the
changes in RE (15–17), without necessarily examining
whether improved RE translated into better running
performance. This is particularly relevant because previous
findings have indicated that improvement in RE does not
necessarily result in a better running performance (18).
From the practical perspective, this information might be
helpful to sports physiologists to better select sport shoes
for competition and training. Therefore, the present study
aimed to: i) compare the effects of TPU and minimalist
shoes on the oxygen cost of running; and ii) analyze the
effect of a possible reduced oxygen cost of running
mediated by these shoes on running performance.
Materials and Methods
Participants
The sample size required was estimated using the
G*power software (version 3.1.9.2, Germany), with data
from a previous investigation that analyzed the effect of
different midsole characteristics (TPU vs conventional) on
RE (15). A sample size of five participants was estimated
to achieve statistically significance in RE, for an expected
effect size of 1.92 and power of 0.8 with an alpha level
of 0.05. In order to improve statistical power, eleven
recreational male runners volunteered to participate in this
study. Participants were engaged in local competitions
and their best performances in 10-km race times ranged
from 35 to 45 min. All participants performed only low-
intensity continuous aerobic training (50–70% maximal
oxygen uptake, O2max) and were instructed to maintain a
similar aerobic training during the experimental period.
The exclusion criteria were: i) exhibited a forefoot contact
running technique; ii) use of dietary supplement; iii)
neuromuscular disorders; and iv) cardiovascular dysfunc-
tions. The participants received a verbal explanation about
the possible benefits, risks, and discomforts associated
with the study and signed a written informed consent
before participating in the study. The study was conducted
according to the Declaration of Helsinki and approved by
the Research Ethics Committee of the School of Physical
Education and Sport of the University of São Paulo
(protocol number 37502714.8.0000.5391).
Experimental design
Participants visited the laboratory on four separate
occasions at least 48 h apart and within a 4-week period.
On the first visit, after the anthropometric measurements,
the participants performed an incremental test to exhaus-
tion in order to determine their O2max wearing their
own running shoes. During second and third visits, the
participants performed, in a counterbalanced design, two
3-km running time-trials (3-km TT) on an outdoor 400-m
track wearing either the TPU or minimalist shoes. On the
fourth visit, the oxygen uptake was measured, in a
counterbalanced design, during constant-speed running
tests performed at 12 km/h (i.e., RE) and at individual
average speed sustained during the 3-km TT (i.e., CTO2),
wearing either the TPU or minimalist shoes. Between
each experimental condition (speed vs shoes), the partici-
pant rested for 10 min. Prior to the experimental laboratory
and field tests, the participants were submitted to a 3-min
familiarization run wearing either the TPU or minimalist
shoes, as previously described (8). The tests were
performed during the preparatory training period, at the
same time of the day, and at least 2 h after the last meal.
The participants were instructed to record their diet 24 h
before the first experimental session and to repeat it prior
to the subsequent experimental sessions. They were
also asked to refrain from any exhaustive or unusual
exercise during the experimental period.
Anthropometric measurements
An experienced investigator performed the anthro-
pometric measurements according to the procedures
described by Norton and Olds (19). Participants were
weighed to the nearest 0.1 kg using an electronic scale
(Filizola, model ID 1500, Brazil). Height was measured
to the nearest 0.1 cm using a stadiometer. Skinfold
thickness was measured to the nearest 0.2 mm at six
body sites (triceps brachial, suprailiac, abdominal, chest,
subscapular, and anterior thigh) using a Harpenden caliper
(West Sussex, UK). The mean of three values was used for
further analysis. Body density and body fat were estimated
by the equations from Jackson et al. (20) and Brozek et al.
(21), respectively.
Incremental maximal test
The incremental maximal test was performed on a
motor-driven treadmill (model TK35, CEFISE, Brazil).
All participants were requested to wear their favorite
running shoes for this test. After a 3-min warm-up at 8 km/h,
the speed was increased by 1 km/h every minute until
participants were unable to maintain the required running
speed. The subjects received strong verbal encourage-
ment to continue as long as possible. Oxygen uptake
( .VO2), carbon dioxide production, and ventilation were
measured breath-by-breath using an automatic metabolic
cart (Cortex, Metalyzer 3B, Germany) and subsequently
averaged over 30-s intervals throughout the test. Before
Braz J Med Biol Res | doi: 10.1590/1414-431X202010693
Footwear, running economy, and endurance performance
2/6
each test, the metabolic cart was calibrated using a 3-L
syringe and a standard gas of known O2 (12%) and CO2
(15%) concentrations. The O2max was determined as
the average of the oxygen uptake during the last 30 s of
the test.
Constant-speed tests
The constant-speed tests were performed using the
same motor-driven treadmill and .VO2 procedures adopted
during the maximal incremental test. The subjects per-
formed a standardized warm-up, consisting of a 5-min run
at 8 km/h followed by a 5-min light stretch. The treadmill
speeds were adjusted after warm-up and the subjects ran
for 6 min wearing either the TPU or minimalist shoes.
Given that previous findings have shown that an athlete
who is energetically economical at a given speed will not
necessarily be economical at other speeds (22), the
participants performed constant-speed running tests at
two distinct speeds wearing either the TPE or minimalist
shoes. RE was determined at 12 km/h similar to a
previous study (15) and the CTO2 was determined at
individual average speed sustained during the 3-km TT
(TPU=15.6±0.8 km/h, minimalist=15.4±.6 km/h). The
oxygen uptake associated with the RE and CTO2 was
measured by averaging the last 30 s from each running
constant-speed
bout.
Recovery
time
between
these
constant-speed running tests was 10 min.
Running performance
The 3-km race is the official track running event that
has been used by previous studies that analyzed the
determinants of running performance (23,24). The par-
ticipants used either the TPU or minimalist shoes during
the 3-km TT, in separate sessions (i.e., second and third
visits). Running performance was measured as the total
time elapsed during the 3-km TT. The time to cover the
3-km TT was registered at each lap (i.e., 400 m) on an
outdoor 400-m track with a manual stopwatch (model HS-
1000, Casio, Japan), while the rating of perceived exertion
(RPE) was reported by participants at each lap using the
Borg 15-point scale (25). Copies of this scale were
laminated, reduced to 10 by 5 cm and fixed to the
participant’s wrist on the dominant arm. Subjects were
instructed to finish the race as quickly as possible, as in a
competitive
event.
Before
the
test,
the
participants
warmed up for 10 min at 8 km/h. They were instructed
to maintain regular water consumption 24 h before the test
and water was provided ad libitum during the entire event.
Verbal encouragement was provided during the entire
event, but runners were not advised of their lap splits.
Ambient temperature and humidity were provided by the
Institute of Astronomy, Geophysics and Atmospheric
Sciences of the University of São Paulo, Brazil. The
mean±SD values for temperature and humidity were
24±2°C and 60±8%, respectively.
Experimental footwear
The experimental footwear used in the current study
consisted of minimalist (Nike Free Run 2, average shoe
mass: 275 g, heel drop: 4 mm) and TPU (Adidas Energy
BoostTM, average shoe mass: 320 g, heel drop: 10 mm)
running shoes. The minimalist model was characterized
by a ultraflexible sole, lightweight, and no motion control
or stability features, as previously described (26). The
midsoles of the TPU model were composed of 80 and
20% of TPU and ethylene vinyl acetate, respectively.
Shoe sizes ranged from 8–10 (UK system). Both shoes
were wrapped with black tape to blind the participants
regarding the footwear used in each experimental
session. Participants had not used either of these shoes
and, therefore, TPU and minimalist shoes were novel to
all participants.
Statistical analysis
A normal data distribution was confirmed by the
Shapiro-Wilk test. Data are reported as means±SD. RE,
CTO2, and running performance were compared between
shoes using paired t-tests. Two-way ANOVA (shoes vs
distance) was used to compare running speed distribution
and RPE responses throughout the 3-km TT. Effect size
(ES) was quantified using standardized mean differences
and defined as trivial (o0.20), small (0.20–0.49), moderate
(0.50–0.79), and large (X0.80). A significance criterion of
Po0.05 was adopted for all analyses. All statistical
analyses were performed using the Statistica 8 software
package (StataSoft Inc., USA).
Results
Table 1 presents the main characteristics of the
participants. The RE and CTO2 are shown in Figure 1,
while running overall performance, running speed dis-
tribution, and RPE changes throughout the 3-km TT for
each shoe condition are shown in Figure 2. TPU footwear
( .VO2=42.5±2.6 mLkg–1min–1) resulted in better RE
(B2.4%) compared to minimalist footwear ( .VO2=43.6±
2.1 mLkg–1min–1) (P=0.01, ES=0.42) (Figure 1A). In
contrast, the CTO2 was not significantly different (P=0.61,
ES=0.18) between TPU ( .VO2=53.1±3.7 mLkg–1min–1)
Table 1. Characteristics of the runners that participated in the
study.
Age (years)
33.1±7.2
Running experience (years)
4.1±2.5
Training volume (km/week)
44.7±14.3
Height (m)
1.74±0.05
Body mass (kg)
70.1±9.9
Maximal oxygen uptake (mLkg-1min-1)
52.1±4.9
Data are reported as means±SD.
Braz J Med Biol Res | doi: 10.1590/1414-431X202010693
Footwear, running economy, and endurance performance
3/6
and minimalist ( .VO2=52.4±3.8 mLkg–1min–1) shoes
(Figure 1B). For both experimental conditions, running
speed distribution showed a classical U-shaped pacing
profile (Figure 2B) with the first and last laps faster than
other laps, while RPE showed a linear profile (Figure 2C).
However, there was a main effect only for distance
(Po0.01), without main effects for shoes (P=0.67) and
interaction (P=0.75) for running speed distribution. Also,
there was a main effect for distance (Po0.01), but not for
shoes (P=0.62) and interaction (P=0.38) for RPE. In
addition, there was no statistical difference for overall
running performance between footwear models (TPU
shoe=701±62 s, minimalist shoe=709±61 s, P=0.52,
ES=0.18) (Figure 2A).
Discussion
Based on the assumption that oxygen cost of running
is one of the best predictors of endurance performance
(27,28) and that RE is acutely improved wearing both
minimalist and TPU shoes (29,30), the present study
aimed to compare the effects of these distinct models on
RE and running performance. The main results of the
current study revealed that: i) TPU shoes resulted in better
RE (B2.4%), but similar CTO2 compared to the minimalist
shoes, and ii) there was no significant difference for
running parameters (i.e., overall performance, running
speed distribution, and RPE responses) between the
TPU and minimalist shoes. These findings suggested that
TPU shoes reduced the oxygen cost of running at low
(i.e., 12 km/h) but not at high (i.e., individual average
speed sustained during the 3-km TT) running speeds,
resulting in a similar endurance performance compared
with minimalist shoes.
Endurance running has become a very popular phys-
ical activity with millions of recreational runners starting
the activity in the past few years (5). This increased
popularity brought attention to the development of different
training methods and technologies focused on acute
improvement of endurance performance, such as a wide
Figure 1. Oxygen uptake during constant-speed tests. A,
Running economy at 12 km/h. B, Oxygen cost of running at
individual average speed sustained during the 3-km running time-
trial. TPU: midsole material composed by thermoplastic polyure-
thane. Data are reported as means±SD. **Po0.05 (t-test).
Figure 2. A, Overall performance; B, running speed distribution;
C, rating of perceived exertion (RPE). TPU: midsole material
composed by thermoplastic polyurethane. Data are reported as
means±SD. #Po0.05 for main effects for distance (t-test and
ANOVA).
Braz J Med Biol Res | doi: 10.1590/1414-431X202010693
Footwear, running economy, and endurance performance
4/6
range of running shoe models commercially available
(16,31). Considering previous studies suggesting that the
low mass of minimalist models is the main characteristic
that affects the energetic cost of running (14,15), in the
present study the footwear mass was normalized by
adding lead tape to the minimalist shoes in order to reduce
the possible influence on RE, CTO2, and running perfor-
mance. Our results showed that the footwear with TPU
midsole material resulted in better RE (2.4%) compared to
minimalist shoes (Figure 1). In comparison with previous
findings, the changes in RE with TPU shoes was slightly
below those reported by Sinclair et al. (14) who compared
the TPU and minimalist shoes (B5%), but similar to
previous results by Worobets et al. (32) and Sinclair
et al. (15), wherein the TPU shoes were able to improve
approximately 1–4% of RE compared with conventional
shoes. These findings are in accordance with the energy
return advantage attributed to the TPU material, which
suggested that RE could be significantly improved (15).
The mechanisms by which the TPU material could
improve RE are not fully understood, but it has been
suggested that the TPU material within the midsole
would reduce the force needed to push the ground
during the propulsion phase, resulting in lower meta-
bolic stress in active skeletal muscles of lower limbs
(33). Together, these findings expand the notion that
TPU could improve RE by enhanced mechanical energy
return (14–16,32), even when compared with minimalist
footwear.
Even though a growing amount of evidence has shown
substantial gains in RE with different footwear models
(7–10), there is a lack of information in the literature
concerning their acute effects on endurance performance.
In the present study, we analyzed the effects of TPU and
minimalist shoes on a 3-km TT performance. Our results
revealed that running parameters (i.e., overall perfor-
mance, running speed distribution, and RPE responses)
were not different between TPU and minimalist conditions
(Figure 2), despite better RE shown for TPU shoes
(Figure 1). The reasons for the similar running parameters
despite a better RE with the TPU are not clear, but it could
be related to the oxygen cost of running at the average
speed at which the 3-km TT was performed. The CTO2
was not significantly different between the models,
indicating that the TPU was not able to maintain the
reduced oxygen cost during high running intensity
compared to the minimalist model. This finding is novel
and relevant because previous findings have demon-
strated that improvements in energetic cost of running
are more effective to endurance performance if observed
at intensities similar to the speed performed in the
actual race rather than fixed submaximal constant-speed
tests (34), such as the speed chosen for the RE test
(i.e., 12 km/h). Therefore, the similar running perfor-
mance observed between running shoes could suggest
an inability of the TPU material to maintain reduced
energetic cost at running speeds similar to those
adopted by athletes during 3-km TT.
In order to address a final conclusion, some limitations
of the present study must be highlighted. First, the running
performance was determined as the total time to cover a
3-km course, which is relatively short compared to other
running events (e.g., 5-, 10-, and 21-km). Thus, given that
longer running events are performed at lower relative
intensities (closer to 12 km/h), it would be important to
compare the effects of the footwear in longer distance
running events. Second, we tested only one model of
shoes designed to enhance energy return, which was
composed by B80% of TPU in its midsole. Perhaps
shoes with different percentages of TPU in the midsole
(50–100% of TPU) could exacerbate the responses in
oxygen cost found in the current study.
In conclusion, the findings of the current study revealed
that footwear with TPU midsole material increases RE at
low running speed (12 km/h) compared with minimalist
shoes. However, the better RE was not evident at the
average speed sustained during 3-km TT (B15 km/h),
ultimately resulting in a similar running performance
compared to minimalist shoes. Therefore, it could be
suggested that improved RE observed with the shoe
material designed to enhance energy return could be
more relevant than the minimalist nature of models for
longer distance running events (X5 km).
Acknowledgments
The authors declare no conflict of interest. All costs
were supported by a grant from National Council for
Scientific and Technological Development (CNPq, #446337/
2014-5).
References
1.
Rusko H, Havu M, Karvinen E. Aerobic performance capacity
in athletes. Eur J Appl Physiol Occup Physiol 1978; 38: 151–
159, doi: 10.1007/BF00421531.
2.
Tanaka K, Matsuura Y, Kumagai S, Matsuzaka A, Hirakoba
K, Asano K. Relationships of anaerobic threshold and onset
of blood lactate accumulation with endurance performance.
Eur J Appl Physiol Occup Physiol 1983; 52: 51–56, doi:
10.1007/BF00429025.
3.
Daniels J, Daniels N. Running economy of elite male
and
elite
female
runners.
Med
Sci
Sports
Exerc
1992; 24: 483–489, doi: 10.1249/00005768-199204000-
00015.
4.
Saunders PU, Pyne DB, Telford RD, Hawley JA. Factors
affecting running economy in trained distance runners.
Sports Med 2004; 34: 465–485, doi: 10.2165/00007256-
200434070-00005.
Braz J Med Biol Res | doi: 10.1590/1414-431X202010693
Footwear, running economy, and endurance performance
5/6
5.
Bertuzzi R, Lima-Silva AE, Pires FO, Damasceno MV,
Bueno S, Pasqua LA et al. Pacing strategy determinants
during a 10-km running time trial: contributions of perceived
effort, physiological, and muscular parameters. J Strength
Cond Res 2014; 28: 1688–1696, doi: 10.1519/JSC.0000000
000000314.
6.
Hoogkamer W, Kipp S, Spiering BA, Kram R, Altered
running economy directly translates to altered distance-
running performance. Med Sci Sports Exerc 2016; 48:
2175–2180, doi: 10.1249/MSS.0000000000001012.
7.
Squadrone R, Gallozzi C. Biomechanical and physiological
comparison of barefoot and two shod conditions in experi-
enced barefoot runners. J Sports Med Phys Fitness 2009; 49:
6–13.
8.
Hanson NJ, Berg K, Deka P, Meendering J. Oxygen cost of
running barefoot vs. running shod. Int J Sports Med 2011;
32: 401–406, doi: 10.1055/s-0030-1265203.
9.
Franz JR, Wierzbinski CM, Kram R. Metabolic cost of
running barefoot versus shod: is lighter better? Med Sci
Sports Exerc 2012; 44: 1519–1525, doi: 10.1249/MSS.0b01
3e3182514a88.
10.
Perl DP, Daoud AI, Lieberman DE. Effects of footwear
and strike type on running economy. Med Sci Sports Exerc
2012; 44: 1335–1343, doi: 10.1249/MSS.0b013e318247
989e.
11.
Bosco C, Rusko H. The effect of prolonged skeletal muscle
stretch–shortening cycle on recoil of elastic energy and on
energy expenditure. Acta Physiol Scand 1983; 119: 219–
224, doi: 10.1111/j.1748-1716.1983.tb07331.x.
12.
Frederick EC. Physiological and ergonomics factors in
running shoe design. Appl Ergon 1984; 15: 281–287, doi:
10.1016/0003-6870(84)90199-6.
13.
Wunsch T, Kroll J, Stoggl T, Schwameder H. Effects
of a structured midsole on spatio-temporal variables
and running economy in overground running. Eur J Sport
Sci 2017; 17: 303–309, doi: 10.1080/17461391.2016.
1253776.
14.
Sinclair J, Mcgrath R, Brook O, Taylo PJ, Dillon S. Influence
of footwear designed to boost energy return on running
economy in comparison to a conventional running shoe.
J Sports Sci 2016; 34: 1094–1098, doi: 10.1080/02640414.
2015.1088961.
15.
Sinclair J, Taylo PJ, Edmunson C, Brooks D. Influence of
footwear kinetic, kinematic and electromyographical para-
meters on the energy requirements of steady state running.
Mov Sports Sci 2015; 80: 39–49.
16.
Worobets J, Wannop JW, Tomaras EK, Stefanyshyn D.
Softer and more resilient running shoe cushioning properties
enhance running economy. Footwear Sci 2014; 6: 147–153,
doi: 10.1080/19424280.2014.918184.
17.
Sinclair J, Shore H, Dillon S. The effect of minimalist,
maximalist and energy return footwear of equal mass on
running economy and substrate utilisation. Compar Exerc
Physiol 2016; 12: 49–54, doi: 10.3920/CEP150029.
18.
Silva R, Damasceno M, Cruz R, Silva-Cavalcante MD, Lima-
Silva AE, Bishop DJ, et al. Effects of a 4-week high-intensity
interval training on pacing during 5-km running trial. Braz J
Med Biol Res 2017; 50: e6335, doi: 10.1590/1414-431x
20176335.
19.
Norton K, Olds T. Anthropometrica. Sidney: Southwood
Press; 1996.
20.
Jackson ASS, Pollock ML. Practical assessment of body
composition. Phys Sportsmed 1985; 13: 76–90, doi: 10.1080/
00913847.1985.11708790.
21.
Brozek J,
Kihlberg JK, Taylor
HL,
Keys A, Skinfold
distributions in middle-aged american men: a contribution
to norms of leanness-fatness. Ann N Y Acad Sci 1963; 110,
492–502, doi: 10.1111/j.1749-6632.1963.tb15776.x.
22.
Kyrolainen H, Kivela R, Koskinen S, McBride J, Andersen
JL, Takala T. et al. Interrelationships between muscle
structure, muscle strength, and running economy. Med Sci
Sports Exerc 2003; 35: 45–49, doi: 10.1097/00005768-
200301000-00008.
23.
Damasceno MV, Duarte M, Pasqua LA, Lima-Silva AE,
MacIntosh BR, Bertuzzi R. Static streching alters neuro-
muscular function and pacing strategy, but not performance
during a 3-Km running time-trial. Plos One 2014; 9: e99238,
doi: 10.1371/journal.pone.0099238.
24.
Duffield
R,
Dawson
B,
Goodman
C.
Energy
system
contribution to 1500- and 3000-metre track running. J Sports
Sci 2005; 10: 993–1002, doi: 10.1080/02640410400021963.
25.
Borg G. Perceived exertion as an indicator of somatic stress.
Scand J Rehabil Med 1970; 2: 92–98.
26.
Bonacci J, Saunders PU, Hicks A, Rantalainen T, Vicenzino
BGT, Spratford W. Running in a minimalist and lightweight
shoe is not the same as running barefoot: a biomechanical
study. Br J Sports Med 2013; 47: 387–392, doi: 10.1136/
bjsports-2012-091837.
27.
Svedenhag J, Sjödin B. Physiological characteristics of elite
male runners in and off-season. Can J Appl Sport Sci 1985;
10: 127–133.
28.
Di Prampero PE, Capelli C, Pagliaro P, Antonutto G, Girardis
M, Zamparo P, et al. Energetics of best performances in
middle-distance running. J Appl Physiol 1985; 74: 2318–
2324, doi: 10.1152/jappl.1993.74.5.2318.
29.
Fuller JT, Bellenger CR, Thewlis D, Tsiros MD, Buckley JD.
The effect of footwear on running performance and running
economy in distance runners. Sports Med 2015; 45: 411–
422, doi: 10.1007/s40279-014-0283-6.
30.
Cheung RT, Ngai SP. Effects of footwear on running
economy in distance runners: a meta-analytical review.
J Sci Med Sport 2016; 19: 260–266, doi: 10.1016/j.jsams.
2015.03.002.
31.
Spurrs RW, Murphy AJ, Watsford ML. The effect of plyometric
training on distance running performance. Eur J Appl Physiol
2003; 89: 1–7, doi: 10.1007/s00421-002-0741-y.
32.
Worobets J, Tomaras EK, Wannop JW, Stefanyshyn DJ.
Running shoe properties can influence oxygen consumption.
Footwear Sci 2013; 5: 575–576, doi: 10.1080/19424280.
2013.799566.
33.
Kyröläinen H, Pullinen T, Candau R, Avela J. Effects of
marathon running on running economy and kinematics. Eur
J Appl Physiol 2000; 82: 297–304, doi: 10.1007/s00421
0000219.
34.
Conley DL, Krahenbuhl GS. Running economy and distance
running performance of highly trained athletes. Med Sci
Sports Exerc 1980; 12: 357–360, doi: 10.1249/00005768-
198025000-00010.
Braz J Med Biol Res | doi: 10.1590/1414-431X202010693
Footwear, running economy, and endurance performance
6/6
| Footwear designed to enhance energy return improves running economy compared to a minimalist footwear: does it matter for running performance? | 03-15-2021 | Dinato, R C,Cruz, R,Azevedo, R A,Hasegawa, J S,Silva, R G,Ribeiro, A P,Lima-Silva, A E,Bertuzzi, R | eng |
PMC4473093 | SYSTEMATIC REVIEW
Incidence of Running-Related Injuries Per 1000 h of running
in Different Types of Runners: A Systematic Review
and Meta-Analysis
Solvej Videbæk1,2 • Andreas Moeballe Bueno3 •
Rasmus Oestergaard Nielsen3 • Sten Rasmussen1,2
Published online: 8 May 2015
The Author(s) 2015. This article is published with open access at Springerlink.com
Abstract
Background
No systematic review has identified the in-
cidence of running-related injuries per 1000 h of running in
different types of runners.
Objective
The purpose of the present review was to
systematically search the literature for the incidence of
running-related injuries per 1000 h of running in different
types of runners, and to include the data in meta-analyses.
Data Sources
A search of the PubMed, Scopus, SPOR-
TDiscus, PEDro and Web of Science databases was conducted.
Study Selection
Titles, abstracts, and full-text articles
were screened by two blinded reviewers to identify
prospective cohort studies and randomized controlled trials
reporting the incidence of running-related injuries in
novice runners, recreational runners, ultra-marathon run-
ners, and track and field athletes.
Study Appraisal and Synthesis Methods
Data were ex-
tracted from all studies and comprised for further analysis.
An adapted scale was applied to assess the risk of bias.
Results
After screening 815 abstracts, 13 original articles
were included in the main analysis. Running-related injuries
per 1000 h of running ranged from a minimum of 2.5 in a
study of long-distance track and field athletes to a maximum
of 33.0 in a study of novice runners. The meta-analyses
revealed a weighted injury incidence of 17.8 (95 % confi-
dence interval [CI] 16.7–19.1) in novice runners and 7.7
(95 % CI 6.9–8.7) in recreational runners.
Limitations
Heterogeneity in definitions of injury, defini-
tion of type of runner, and outcome measures in the included
full-text articles challenged comparison across studies.
Conclusion
Novice runners seem to face a significantly
greater risk of injury per 1000 h of running than recre-
ational runners.
Key Points
‘Injuries per 1000 h of running’ is an important and
useful measure of association that enables
comparison of the risk of injury across studies.
Novice runners are at significantly higher risk of
injury 17.8 (95 % CI 16.7–19.1) than recreational
runners, who sustained 7.7 (95 % CI 6.9–8.7)
running-related injuries per 1000 h of running.
More studies on ultra-marathon runners and track
and field athletes are needed in order to calculate
weighted estimates.
1 Introduction
Running is one of the most popular and accessible sport
activities worldwide [1, 2]. It can be performed with
minimal equipment, and by a broad variety of people in
Electronic supplementary material
The online version of this
article (doi:10.1007/s40279-015-0333-8) contains supplementary
material, which is available to authorized users.
& Solvej Videbæk
[email protected]
1
Department of Orthopaedic Surgery Research Unit, Science
and Innovation Center, Aalborg University Hospital, Aarhus
University, 18–22 Hobrovej, 9000 Aarhus, Denmark
2
Department of Clinical Medicine, Aalborg University,
Aalborg, Denmark
3
Section of Sport Science, Department of Public Health,
Faculty of Health Science, Aarhus University, Room 438,
Dalgas Avenue 4, 8000 Aarhus, Denmark
123
Sports Med (2015) 45:1017–1026
DOI 10.1007/s40279-015-0333-8
almost every part of the world. In the US, more than
40,000,000 people run [2], and in Denmark and The
Netherlands approximately 25 and 12.5 % of the popula-
tion, respectively, run on a regular basis [3, 4].
Running-related injuries affect many runners. Unfortu-
nately, the exact number of injuries is hard to identify
because various studies have provided results on the
prevalence and incidence of running-related injuries using
different measures of association [5, 6]. To name a few,
injuries have been reported as the number of injuries per
1000 km [7, 8]; proportion of injuries in a population [9];
number of injured runners per 100 runners [10]; and
number of injured runners per 1000 h of running [11, 12].
The inconsistent use of such measures in the literature
makes comparison of injury data difficult across studies.
Injuries per 1000 h of running was highlighted by
Jakobsen et al. [12] as an important measure of association.
They stated that the risk of injury must be related to the
time spent running, in order to make the results from dif-
ferent studies comparable. This is supported in a review
from 2012 by Lopes et al. [13], who emphasize that stan-
dardization of the number of injuries per hour of exposure
is highly needed in running-related injury research.
In a review from 1992, van Mechelen [10] compared the
incidence rates of running-related injuries across a few
studies presenting such results. The results revealed an
injury incidence of 2.5–12.5 injuries per 1000 h of running.
Since then, many studies have reported information on
running-related injuries in different types of runners per
1000 h of running—for instance, novice runners, recre-
ational runners, ultra-marathon runners, and track and field
athletes. However, no review has been published which
systematically searched the literature for studies with in-
formation on the incidence of running-related injuries in
different types of runners per 1000 h of running.
The purpose of the present review was to systematically
review the literature for the incidence of running-related
injuries in novice runners, recreational runners, ultra-
marathon runners, and track and field athletes per 1000 h of
running. A secondary objective was to compare the injury
rates across different types of runners per 1000 h of run-
ning and include the data in meta-analyses.
2 Methods
2.1 Search Strategy
Five databases (PubMed, Scopus, SPORTDiscus, PEDro
and Web of Science) were searched electronically, without
restriction on date of publication, to identify studies that
included data regarding running-related injury incidences
per 1000 h of running. The search was performed in
collaboration with a certified librarian at Aarhus University
Library, Denmark. Full details of the electronic search
strategy for PubMed are provided in the electronic sup-
plementary material (ESM) Appendix S1. Additional
studies and trials were identified by checking references of
included full-text articles and published reviews within the
running injury thematic. Full-text articles, which were not
included after searching the databases, were included
afterwards if they, to the authors’ knowledge, had infor-
mation about injuries per 1000 h of running.
2.2 Study Selection
The screening of eligible studies was performed by two
reviewers (SV and AMB), in two steps. In step 1, all ab-
stracts were evaluated according to pre-specified inclusion
and exclusion criteria. Inclusion criteria for abstracts con-
sisted of the following: subjects were children, novice
runners, recreational runners, elite runners, cross-country
runners, orienteers, and/or triathletes; the study was based
on original research (prospective cohort studies and ran-
domized controlled trials); the article was written in Eng-
lish or Danish; and the abstract included data regarding
running-related injuries per 1000 h of running, or indicated
that such data might be available in the full-text article.
Exclusion criteria included the following: subjects were
military or army recruits; studies in which participants
were predominantly exposed to different types of sports
other than running; original study designs consisted of
cross-sectional studies, case–control studies, case series
and case reports; and studies did not include original re-
search, for instance reviews.
All abstracts were evaluated independently by each of
the two reviewers and either included or excluded. In cases
of disagreement between the two blinded reviewers (SV
and AMB), a third reviewer (RON) made the final decision
of selection.
In step 2, the two reviewers (SV and AMB) read all full
texts included in step 1 as well as the full texts of the
additional articles identified in the reference lists. The
following criteria were used to finally include or exclude
full-text articles. Inclusion criteria for full-text articles:
must include findings from which it is possible to extract
data on running-related injuries per 1000 h of running;
articles without data on injuries per 1000 h of running, but
containing data on the incidence of injuries per 1000 km.
Exclusion criteria: studies in which participants were pre-
dominantly exposed to different types of sports other than
running and, consequently, running-related injuries could
not be distinguished from other sport injuries; if injuries
per 1000 h were estimated per leg and not per individual;
and data on injuries per 1000 h of running were missing,
data on number of events and time at risk were unavailable,
1018
S. Videbæk et al.
123
and the corresponding author was unable to provide these
data after being contacted by e-mail.
Each reviewer (SV and AMB) processed the articles
individually and, in cases of disagreement, they followed a
consensus decision-making process. In cases where they
did not reach a consensus, a third reviewer (RON) made the
final judgment.
2.3 Data Collection
The study characteristics of the included full-text articles
were described to gain insight into the homogeneity of the
study populations and definitions of running-related injuries.
The following data were collected: author and year of pub-
lication; study design; type of runners; sample size used in
the analysis; description of the study population; and
definition of the running-related injury (Table 1). Estimates
of the incidence of running-related injuries per 1000 h and
per kilometres were extracted from all studies for further
analysis. Three studies provided estimates of running-related
injuries per 1000 h without 95 % confidence intervals (CIs)
and without presenting the raw data needed to calculate these
[12, 14, 15]. The corresponding authors were contacted and
data were received from two of them [14, 15], which enabled
the inclusion of these results in the meta-analyses.
The study populations of the included studies were
categorized into one of four types of runners: novice run-
ners; recreational runners; ultra-marathon runners; and
track and field athletes. This categorization was made to
enable comparison of results across studies.
Some studies reported the incidence of running-related
injuries per 1000 miles [7, 8, 16] but these results were
converted into running-related injuries per 1000 km using
an online converter [17].
2.4 Risk of Bias Assessment
The tool used for assessing risk of bias of the included
studies was chosen after thorough consideration of the ad-
vantages and disadvantages of the available methods for
evaluating bias. The studies included both prospective co-
hort studies and randomised controlled trials. The main
purpose of this review was to measure the incidence of
running-related injuries per 1000 h of running. The causes
of running-related injuries were not of interest, thus
minimizing the importance of methods of randomization for
the quality of outcome. Quality assessment by one single
tool was therefore possible for both designs. The tool used
to assess the risk of bias of the included studies was a
version of the Newcastle Ottawa Scale, a tool modified by
Saragiotto et al. [18] to evaluate studies undertaking re-
search on runners. The tool contains 11 criteria designed to
assess the risk of bias, and uses a star rating system to
indicate the quality of a study (see ESM Appendix S2 for a
description of each criterion in the original version of the
quality assessment tool modified for runners [18]). Certain
modifications were applied to specify the tool used to assess
the risk of bias on the parameter of concern in our review—
the incidence of running-related injuries. Three of the 11
criteria were excluded. Item 4 was excluded because an
exposed versus non-exposed cohort was irrelevant as long
as the total study population was exposed to running; item 7
was excluded because it was linked to item 4; and item 11
concerned the risk of association and was removed because
these measures relate to research on associations. In item 3
the wording ‘average runners in the community’ was re-
worded to ‘average type of runners researched’, meaning
that the article received a star if the study population were
representative of the type of runner (novice runners,
recreational runners, ultra-marathon runners, or track and
field athletes) described according to item 1. The criteria
adopted to assess risk of bias were (1) description of runners
or type of runner; (2) definition of the running-related in-
jury; (3) representativeness of the exposed cohort; (4)
ascertainment of exposure; (5) demonstration that outcome
of interest was not present at the start of the study; (6)
assessment of outcome; (7) was follow-up long enough for
outcomes to occur; and (8) adequacy of follow-up of co-
horts. The risk of bias assessment was carried out by two
researchers (SV and AMB) in a blinded process and, in
cases of disagreement, they went through a consensus-
making process. Only studies with estimates on injury in-
cidence per 1000 h were quality scored since this outcome
represented the main analysis.
3 Results
A total of 3172 articles were identified through the data-
base searches. Among these articles, 2357 were duplicates,
as determined by the reference program RefWorks. Next,
815 titles and abstracts were evaluated in step one of the
selection process. Of these, 69 full-text articles were in-
cluded and evaluated according to the inclusion and ex-
clusion criteria in step two of the selection process, of
which 58 were excluded. In Fig. 1, the selection process is
visualised in a flowchart. By checking reference lists, one
additional study was included [14]. In addition, the authors
knew of one article that was not included in the search but
in which the relevant information was incorporated [26].
This article was also included. Finally, 13 articles that
presented data on running-related injuries per 1000 h of
running were included—eight prospective cohort studies
and five randomized controlled trials. Overall, ten studies
provided estimates on running-related injuries per 1000 km
and these were used for a subanalysis.
Incidence of Running-Related Injuries
1019
123
Table 1 Description of studies
References,
country of origin
Study design (follow-
up)
Study population
Baseline characteristics
Musculoskeletal injury definition
Novice runners
Bovens et al.
[25], The
Netherlands
Prospective cohort
study (81 weeks)
73 Novice runners with little or no running
experience
Age above 20 years. Only volunteers without
persisting injuries were accepted. (58 men and 15
women)
Any physical complaint developed in relation with
running activities and causing restriction in running
distance, speed, duration or frequency
Bredeweg et al.
[24], The
Netherlands
Randomised
controlled trial
(9 weeks plus
additional 4 weeks
for 211 runners)
362 (171?191) All participants had not been
running on a regular basis in the previous
12 months
Age range 18–65 years. No injury of the lower
extremity within the preceding 3 months
Any musculoskeletal complaint of the lower extremity
or lower back causing restriction of running for at
least a week
Buist et al. [11],
The
Netherlands
Prospective cohort
study (8 weeks)
629 Runners who had signed up for 4-mile
running event. 474 novice runners who
either restarted running or had no running
experience. 155 recreational runners
Age above 18 years
Any musculoskeletal pain of the lower limb or back
causing a restriction of running for at least 1 day
Buist et al. [20],
The
Netherlands
Randomized
controlled trial (8
and 13 weeks)
486 Novice runners who had not been
running on a regular basis in the previous
12 monthsa
Age range 18–65 years. No injury of the lower
extremity within the preceding 3 months
Any self-reported running-related musculoskeletal pain
of the lower extremity or back causing a restriction of
running for at least 1 week (three scheduled
trainings)
Nielsen et al.
[26],
Denmark
Prospective cohort
study (12 months)
930 Novice runners who had not been
running on a regular basis in the previous
12 monthsa
Healthy novice runners age range 18–65 years with
no injury in the lower extremities or back 3 months
preceding baseline investigation. Not participating
in other sports for more than 4 h/week
Any musculoskeletal complaint of the lower extremity
or back causing a restriction of running for at least
1 week
Recreational runners
Jakobsen et al.
[12],
Denmark
Randomised
controlled trial
(12 months)
41 Recreational long-distance runners. Had
all taken part in marathon races and
intended to take part in at least two
marathons during the year of investigation
19 Men and 2 women aged 24–56. No runner had any
symptoms or objective signs of overuse injury at the
start of the investigation
Any injury to the musculoskeletal system that was
sustained during running and prevented training or
competition
Malisoux
et al. [14],
Luxembourg
Prospective cohort
study (22 weeks)
264 Recreational runners. Mean regularity of
runningc in the last 12 months = 9.4–10.8
Healthy participants above 18 years old with any
level of fitness
A physical pain or complaint located at the lower limbs
or lower back region, sustained during or as a result
of running practice and impeding planned running
activity for at least 1 day
Theisen
et al. [15],
Luxembourg
Randomised
controlled trial
(5 months)
247 Recreational runners
Healthy and uninjured leisure-time runners, aged
above 18 years. Participants having more than 6
accumulated months of regular trainingb
Any first-time pain sustained during or as a result of
running practice and impeding normal running
activity for at least 1 day
Van Mechelen
et al. [28],
The
Netherlands
Randomised
controlled trial
(16 weeks)
421 Recreational runners running at least
10 km/week all year-round
Healthy, no current injury, not home from work at
sick leave, not performing sport as a part of their
profession
Any injury that occurred as a result of running and
caused one or more of the following: (1) the subject
had to stop running, (2) the subject could not run on
the next occasion, (3) the subject could not go to
work the next day, (4) the subject needed medical
attention, or (5) the subject suffered from pain or
stiffness during 10 subsequent days while running
1020
S. Videbæk et al.
123
Table 1 continued
References,
country of origin
Study design (follow-
up)
Study population
Baseline characteristics
Musculoskeletal injury definition
Wen et al. [23],
USA
Prospective cohort
study (32 weeks)
MH group:108 recreational runners
previously running a mean of 24.94 km/
weekb. However 8.3 % of these were
novice runners with no running experience
Members of a running group with the purpose to
prepare its members to run a marathon
Answering yes to having had ‘‘injury or pain’’ to an
anatomic part; answering yes to having had to stop
training, slow pace, stop intervals, or otherwise
having had to modify training; and a ‘‘gradual,’’
versus ‘‘immediate’’, onset of the injury or a self-
reported diagnosis that is generally considered an
overuse injury
Ultra marathon runners
Krabak et al.
[21], USA
Prospective cohort
study (7 days)
396 Experienced runners who have
completed marathon or ultraendurance
events
Age range 18–64 years
A disability sustained by a study participant during the
race, resulting in a medical encounter by the medical
staff
Track and field athletes
Bennell et al.
[22],
Australia
Prospective cohort
study (12 months)
95 Competitive track and field athletes
(throwers and walkers excluded)
Age range 17–26 years. Training at least three times a
week, when uninjured
Any musculoskeletal pain or injury that resulted from
athletic training and caused alteration of normal
training mode, duration, intensity or frequency for
1 week or more
Lysholm et al.
[19], Sweden
Prospective cohort
study (12 months)
60 Track and field athletes. Sprinters,
middle-distance runners and
longdistance/marathon runners running in
club and competing
Previous experience of training (7 h per week or
more) varied between 1 and 32 years
Any injuries that markedly hampered training or
competition for at least 1 week
MH mileage-hours
a 10km total in all training sessions in the previous 12 months
b Miles were converted to km [17]
c Regular training (at least once a week)
Incidence of Running-Related Injuries
1021
123
The year of publication for the included studies ranged
from 1987 to 2014, and the studies represented populations
in
Australia,
Denmark,
Luxembourg,
Sweden,
The
Netherlands, and the USA. The follow-up periods ranged
from 7 days to 81 weeks. Eight studies used a time-loss
definition of injury; one study defined an injury as a need
for medical attention; and the remaining four studies used a
mixture of time loss, physical pain, and the need for
medical attention in the definition of injury.
Across studies, the primary purpose was to compare the
incidence of running-related injuries per 1000 h of running.
Five studies reported this estimate in novice runners; five
studies in recreational runners; one study in ultra-marathon
runners; and two studies in track and field athletes. The
estimates ranged from 2.5 [19] to 33.0 [20] running-related
injuries per 1000 h of running. Two meta-analyses were
performed on the estimates of novice runners and recre-
ational runners, respectively. As one article [12] did not
provide data to calculate 95 % CIs, estimates from nine
studies were included in these quantitative analyses
(Fig. 2). The weighted estimates revealed novice runners
faced a significantly greater injury rate of 17.8 (95 % CI
16.7–19.1) than recreational runners, who sustained 7.7
(95 % CI 6.9–8.7) running-related injuries per 1000 h of
running.
Ten studies provided estimates of running-related in-
juries per 1000 km of running, and these results were
pooled in a subanalysis (Table 2). The weighted estimate
revealed an injury incidence of 1.07 (95 % CI 1.01–1.13)
per 1000 km of running.
The risk of bias was assessed for each of the 13 studies
presenting an estimate of the incidence of running-related
Fig. 1 Flowchart visualizing
the selection process of studies
in the systematic review
1022
S. Videbæk et al.
123
injuries per 1000 h of running (Table 3). The criteria most
frequently awarded with a star were description of runners
or type of runners (13/13) and definition of running-related
injury (13/13). The criteria with the least stars awarded
comprised ascertainment of exposure (6/13) and assess-
ment of outcome (8/13). The average stars awarded to the
articles assessed for risk of bias was 6 out of a total of 8
stars, with a maximum of 8 and a minimum of 3.
4 Discussion
The present review is the first to systematically review the
literature on the incidence rate of running-related injuries
in different types of runners. The weighted estimate of 17.8
(95 % CI 16.7–19.1) running-related injuries per 1000 h of
running in novice runners was significantly greater than the
incidence rate of 7.7 (95 % CI 6.9–8.7) in recreational
runners. One study reported the incidence of running-
Fig. 2 Meta-analysis performed on the estimates of running-related
injuries per 1000 h in novice runners and recreational runners. aData
on standard error or 95 % confidence limits were not reported and the
study was therefore not included in the meta-analysis. bData on
standard error or 95 % confidence limits were not reported and
therefore no meta-analysis was performed on track and field athletes.
CI confidence intervals
Table 2 Running-related injuries per 1000 km of running
References
Runners
(n)
Injuries
(n)
Estimate
(RRI per
1000 km)
95 % CI
Bennell et al. [22]
95
130
0.58
0.5, 0.7
Bovens et al. [25]
73
174
0.86
0.7, 1.0
Fields et al. [7]
40
17
0.18
0.1, 0.3
Gerlach et al. [8]
86
47
0.22
0.2, 0.3
Jakobsen et al. [12]
41
50
0.62
0.4, 0.9
Krabak et al. [21]
396
217
2.28
2.0, 2.6
Nielsen et al. [27]
58
13
2.85
1.7, 4.9
Nielsen et al. [26]
930
294
1.64
1.5, 1.8
Van Mechelen et al.
[28]
421
49
0.44
0.3, 0.6
Wen et al. [23]
108
49
0.76
0.6, 0.9
Weighted estimate
2248
1040
1.07
1.01,
1.13
RRI running-related injuries, km kilometres, CI confidence interval
Incidence of Running-Related Injuries
1023
123
related injuries in ultra-marathon runners as 7.2 per 1000 h
[21]. In track and field athletes, two studies reported the
incidences of running-related injuries from 2.5 to 26.3 per
1000 h [19, 22]. In the latter, track and field athletes were
subdivided into sprinters, middle-distance runners, and
long-distance runners, which may be relevant as the re-
ported running-related injury incidence per 1000 h was
greater in sprinters and middle-distance runners than in
long-distance runners [19].
In Fig. 2, a summary of the results in different types of
runners is presented. The healthy participant effect may
play a role when grouping novice versus recreational run-
ners [23]. In novice runners, the five studies are heteroge-
neous since the estimates reported by three of the studies
[11, 20, 24] range from 30.1 to 33 and are significantly
higher than those reported by the two remaining studies [25,
26]. A possible explanation for the discrepancy is the fol-
low-up time in the respective studies. The non-injured
runners accumulate relatively more exposure time in studies
with a long follow-up, while the injured runners are
censored. This will mathematically explain the overall de-
crease in running-related injuries per 1000 h of running in
studies with longer follow-up amongst novice runners. The
two studies with the lowest incidence of running-related
injuries per 1000 h of running had 81 and 52 weeks of
follow-up, while the three studies with the greatest injury
incidence had follow-up periods of 8–13 weeks (Table 1).
The link between a relatively short follow-up time and a
high incidence rate of running-related injuries versus long
follow-up time and a lower incidence rate of running-re-
lated injuries indicates the possibility that runners classified
as novice runners at the beginning of a study may rea-
sonably be classified as recreational runners as time passes.
If novice runners exceed 8–13 weeks without injury, they
may well have adapted to running and face a lower injury
risk after this period, even though they may spend more
time running. Novice runners exceeding 8–13 weeks’ fol-
low-up may then be considered as recreational runners
instead. Based on this, it may be appropriate to identify a
cut-off distinguishing a novice runner from a recreational
runner.
In contrast, the injury incidences are homogeneous in
recreational runners and the weighted estimate is unaf-
fected by bias.
The strengths of the present review are mainly the sys-
tematic search of the literature and the use of meta-analyses
to compare the injury incidences. The searches were per-
formed thoroughly in five databases, in cooperation with a
certified librarian. Moreover, all reference lists of the in-
cluded full-text articles were checked for additional studies
and, to the authors’ knowledge, one article [26] was also
able to be included for analysis, although it was not indexed
in any of the five databases searched. Evaluation of the
quality of all articles presenting estimates of running-re-
lated injuries per 1000 h was accomplished and meta-ana-
lyses on these data were conducted. Thus, the present
systematic review and meta-analyses represent rigorous
evaluations and provide estimates of running-related injury
incidences in novice runners, recreational runners, ultra-
marathon runners, and track and field athletes.
The present study has a number of limitations, including
differences in definitions of injury, definition of type of
runner, and outcome measures used. First, definition of
injury varies considerably across studies. Eight studies
used time-loss definitions, but even within this definition
there is a lack of consensus of the amount of time needed to
classify time loss from running as a running-related injury.
One study did not define the amount of time [12], some
studies used 1 day in their definition [11, 14, 15], while
other studies used 1 week [19, 20, 22, 24]. The only study
[21] solely defining injury as the need for medical attention
was reporting on ultra-marathon runners, and as these data
were collected in real time while the runners participated in
Table 3 Risk of bias assessment
Criteria for assessing risk of bias
1
2
3
4
5
6
7
8
Novice runners
Bovens et al. [25]
*
*
*
0
*
*
*
*
Bredeweg et al. [24] RCT
*
*
*
0
*
0
0
0
Buist et al. [11]
*
*
*
0
0
0
0
0
Buist et al. [20] RCT
*
*
*
0
0
0
0
*
Nielsen et al. [26]
*
*
*
*
*
*
*
*
Recreational runners
Jakobsen et al. [12] RCT
*
*
*
0
*
*
*
0
Malisoux et al. [14]
*
*
*
*
*
0
*
*
Theisen et al. [15] RCT
*
*
0
*
*
*
*
*
Van Mechelen et al. [28] RCT
*
*
*
0
*
*
*
0
Wen et al. [23]
*
*
*
0
0
0
*
*
Ultra-marathoners
Krabak et al. [21]
*
*
*
*
0
*
0
*
Track and field athletes
Bennell et al. [22]
*
*
*
*
*
*
*
*
Lysholm et al. [19]
*
*
*
*
0
*
*
0
Only studies providing estimates of the incidence of running-related
injuries per 1000 h were assessed for risk of bias. The criteria adopted
to assess risk of bias were: (1) description of runners or type of
runner; (2) definition of the running-related injury; (3) representa-
tiveness of the exposed cohort; (4) ascertainment of exposure; (5)
demonstration that outcome of interest was not present at start of
study; (6) assessment of outcome; (7) was follow-up long enough for
outcomes to occur?; (8) adequacy of follow-up of cohorts
RCT randomised controlled trial
* A study was awarded a star for every criterion it fulfilled. The more
stars the higher quality
1024
S. Videbæk et al.
123
the ultra-marathon, this method was reasonable. No studies
exclusively defined a running-related injury as physical
pain alone, but, in four studies, physical pain was incor-
porated as part of the injury definition [16, 22, 25, 28].
Second, runners from the included studies were classified
into four groups according to the type of runner, enabling
relevant intergroup comparison. No exact definition of each
category was made, but the baseline characteristics leading
to grouping in one of the four types of runners are listed in
Table 1. Third, the method of gathering data on exposure
time may be questionable. In many studies, runners were
asked to self-report their training exposure in web-based
running diaries. This approach may lead to training hours
or distance being estimated wrongly, possibly because of
recall bias and time spent self-reporting [27]. The quality
assessment tool accounted for this, and awarded no star
when exposure was registered by written self-report (item
5). However, it is questionable whether the risk of bias was,
in reality, higher in the study by Bovens et al. [25], which
received no star in item 5 because running exposure was
collected in diaries, than in the study by Benell et al. [22],
in which a star was awarded for a retrospective personal
interview completed by one of the researchers at the end of
the 12 months of follow-up. Lack of agreement in the way
exposure time was calculated was another challenge. In
some studies [11, 14, 15, 19, 23–25, 28], exposure time was
calculated from the time a participant started the running
programme until the time they reported a running-related
injury (injured runners) or until the end of the programme
(non-injured runners). This way of calculating exposure
time was ideal due to that fact that the same runner could
only contribute exposure time as long as he had not been
injured. Thus, the risk of registering the same injury twice,
if re-occurring, was avoided. Additionally, an injured
person could not add exposure time after the injury oc-
curred, and the number of injuries would be the same as the
number of injured runners. Other studies did not mention
whether study participants were censored if an injury oc-
curred [12, 20, 26]. Further, some studies specified the
premise that the same runner was included and was con-
tributing exposure time, if running was resumed after an
injury occurrence [21, 22]. Due to the varying ways of
calculating exposure time in the included studies, the most
appropriate comparison of the incidence of running-related
injuries across all included studies was to use the total
number of registered injuries instead of the total number of
injured runners. This approach made it possible for one
runner to figure twice or more in the pooled count of in-
juries. However, it would have been preferable if all studies
had used the ideal method of calculating exposure time
since this would have meant that one single runner could
not accumulate exposure time after a first-time injury and
have a recurrent injury counted twice.
Of the 13 studies providing estimates on running-related
injuries per 1000 h of running, not all provided raw data on
exact exposure time or 95 % CIs of the reported estimates.
Corresponding authors from the respective articles [12, 14,
15, 26] were contacted, and data were received from
Malisoux et al. [14], Theisen et al. [15] and Nielsen et al.
[26]. Moreover, the estimate of 30.1 running-related in-
juries per 1000 h used in the meta-analysis relating to
novice runners derives from the complete study population
of runners in the prospective study of Buist et al. [11].
Overall, 155 of these 629 runners were described as run-
ners already participating in running at baseline, running a
mean of 1.2 h per week. Unfortunately, we were unable to
obtain data that allowed us to calculate estimates for each
of the groups of runners separately. Consequently, we de-
cided to include the estimate of 30.1 running-related in-
juries per 1000 h in the category of novice runners;
therefore, the true incidence of running-related injuries in
novice runners might be even higher.
The present study constitutes a thorough and fully up-
dated literature review presenting data regarding the inci-
dence rates of running-related injuries, and outlining
relevant key issues, which limit the comparison of studies
in running-related injury research. The included meta-
analyses form new estimates showing variations in the
incidence rates of running-related injuries among different
types of runners, and can be used as a starting point in
future running-related injury research.
5 Conclusions
The reported weighted analysis of running-related injury
incidence per 1000 h of running revealed that novice run-
ners face a significantly greater risk of injury 17.8 (95 % CI
16.7–19.1) than their recreational peers 7.7 (95 % CI
6.9–8.7). Caution is advisable when comparing estimates
on the incidence of running-related injuries across studies
because of differences in the definition of injury. Only a
few studies reported injury incidences of ultra-marathon
runners and track and field athletes, and no weighted es-
timates were calculated.
Acknowledgments
No sources of funding were used to assist in the
preparation of this review. Solvej Videbæk, Andreas Moeballe
Bueno, Rasmus Oestergaard Nielsen and Sten Rasmussen have no
potential conflicts of interest that are directly relevant to the content
of this review.
Open Access
This article is distributed under the terms of the
Creative Commons Attribution 4.0 International License (http://
creativecommons.org/licenses/by/4.0/), which permits unrestricted
use, distribution, and reproduction in any medium, provided you give
appropriate credit to the original author(s) and the source, provide a
link to the Creative Commons license, and indicate if changes were
made.
Incidence of Running-Related Injuries
1025
123
References
1. AIMS—Association of International Marathons and Distance
Races. Available at: http://aimsworldrunning.org/index.php. Ac-
cessed 26 Aug 2014.
2. Running USA. 2014 state of the sport—part II: running industry
report. Available at: http://www.runningusa.org/2014-running-
industry-report?returnTo=annual-reports. Accessed 6 Aug 2014.
3. Idrættens analyseinstitut. Motionsløbere i Danmark: portræt af
danske
motionsløbere.
Available
from:
http://www.idan.dk/
vidensbank/downloads/motionsloeb-i-danmark-portraet-af-de-
danske-motionsloebere/b6326ae1-9d2c-4744-b5bf-9ff800a868fd.
Accessed 6 Aug 2014.
4. Buist I, Bredeweg SW, Lemmink KA, et al. The GRONORUN
study: is a graded training program for novice runners effective in
preventing running related injuries? Design of a randomized
controlled trial. BMC Musculoskelet Disord. 2007;8:24.
5. van Gent RN, Siem D, van Middelkoop M, et al. Incidence and
determinants of lower extremity running injuries in long distance
runners:
a
systematic
review.
Br
J
Sports
Med.
2007;41(8):469–80.
6. Johnston CA, Taunton JE, Lloyd-Smith DR, et al. Preventing
running injuries. practical approach for family doctors. Can Fam
Physician. 2003;49:1101–9.
7. Fields KB, Delaney M, Hinkle JS. A prospective study of type A
behavior and running injuries. J Fam Pract. 1990;30(4):425–9.
8. Gerlach KE, Burton HW, Dorn JM, et al. Fat intake and injury in
female runners. J Int Soc Sports Nutr. 2008;5:1.
9. Ryan MB, Valiant GA, McDonald K, et al. The effect of three
different levels of footwear stability on pain outcomes in women
runners:
a
randomised
control
trial.
Br
J
Sports
Med.
2011;45(9):715–21.
10. van Mechelen W. Running injuries: a review of the epi-
demiological literature. Sports Med. 1992;14(5):320–35.
11. Buist I, Bredeweg SW, Bessem B, et al. Incidence and risk fac-
tors of running-related injuries during preparation for a 4-mile
recreational
running
event.
Br
J
Sports
Med.
2010;44(8):598–604.
12. Jakobsen BW, Kroner K, Schmidt SA, et al. Prevention of in-
juries in long-distance runners. Knee Surg Sports Traumatol
Arthrosc. 1994;2(4):245–9.
13. Lopes AD, Hespanhol Junior LC, Yeung SS, et al. What are the
main running-related musculoskeletal injuries? Sports Med.
2012;42(10):891–905.
14. Malisoux L, Ramesh J, Mann R, et al. Can parallel use of dif-
ferent running shoes decrease running-related injury risk? Scand J
Med Sci Sports. 2015;25(1):110–5.
15. Theisen D, Malisoux L, Genin J, et al. Influence of midsole
hardness of standard cushioned shoes on running-related injury
risk. Br J Sports Med. 2014;48(5):1–6.
16. Wen DY, Puffer JC, Schmalzried TP. Lower extremity alignment
and risk of overuse injuries in runners. Med Sci Sports Exerc.
1997;29(10):1291–8.
17. Mile converter. Available at: https://www.google.dk/search?q=
mile?converter&ie=utf-8&oe=utf-8&rls=org.mozilla:da:official
&client=firefox-a&channel=sb&gws_rd=cr&ei=w-oFVKGiLsbg
yQPom4LYDg. Accessed 2 Sep 2014.
18. Saragiotto BT, Yamato TP, Hespanhol Junior LC, et al. What are
the main risk factors for running-related injuries? Sports Med.
2014;44(8):1153–63.
19. Lysholm J, Wiklander J. Injuries in runners. Am J Sports Med.
1987;15(2):168–71.
20. Buist I, Bredeweg SW, van Mechelen W, et al. No effect of a
graded training program on the number of running-related in-
juries in novice runners. Am J Sports Med. 2008;36(1):35–41.
21. Krabak BJ, Waite B, Schiff MA. Study of injury and illness rates
in multiday ultramarathon runners. Med Sci Sports Exerc.
2011;43(12):2314–20.
22. Bennell K, Malcolm S, Thomas S, et al. The incidence and dis-
tribution of stress fractures in competitive track and field athletes:
a
twelve-month
prospective
study.
Am
J
Sports
Med.
1996;24(2):211–7.
23. Wen DY, Puffer JC, Schmalzried TP. Injuries in runners: a
prospective
study
of
alignment.
Clin
J
Sport
Med.
1998;8(3):187–94.
24. Bredeweg SW, Zijlstra S, Bessem B, et al. The effectiveness of a
preconditioning programme on preventing running-related in-
juries in novice runners: a randomised controlled trial. Br J Sports
Med. 2012;46(12):865–70.
25. Bovens AMP, Janssen GME, Vermeer HGW, et al. Occurrence of
running injuries in adults following a supervised training pro-
gram. Int J Sports Med. 1989;10(Suppl 3):S186–90.
26. Nielsen RO, Buist I, Parner ET, et al. Predictors of running-
related injuries among 930 novice runners: a 1-year prospective
follow-up study. Orthop J Sports Med. 2013;1(1):1–7.
27. Nielsen RO, Cederholm P, Buist I, et al. Can GPS be used to
detect deleterious progression in training volume among runners?
J Strength Cond Res. 2013;27(6):1471–8.
28. Van Mechelen W, Hlobil H, Kemper HCG, et al. Prevention of
running injuries by warm-up, cool-down, and stretching exer-
cises. Am J Sports Med. 1993;21(5):711–9.
1026
S. Videbæk et al.
123
| Incidence of Running-Related Injuries Per 1000 h of running in Different Types of Runners: A Systematic Review and Meta-Analysis. | [] | Videbæk, Solvej,Bueno, Andreas Moeballe,Nielsen, Rasmus Oestergaard,Rasmussen, Sten | eng |
PMC8461211 | Physiological Reports. 2021;9:e15037.
| 1 of 14
https://doi.org/10.14814/phy2.15037
wileyonlinelibrary.com/journal/phy2
Received: 19 May 2021 | Revised: 20 August 2021 | Accepted: 26 August 2021
DOI: 10.14814/phy2.15037
O R I G I N A L A R T I C L E
Comparison of constant load exercise intensity for
verification of maximal oxygen uptake following a graded
exercise test in older adults
Ian R. Villanueva1 | John C. Campbell1 | Serena M. Medina1 | Theresa
M. Jorgensen1 | Shannon L. Wilson1 | Siddhartha S. Angadi2 | Glenn A. Gaesser1 |
Jared M. Dickinson3
This is an open access article under the terms of the Creat ive Commo ns Attri bution License, which permits use, distribution and reproduction in any medium, provided
the original work is properly cited.
© 2021 The Authors. Physiological Reports published by Wiley Periodicals LLC on behalf of The Physiological Society and the American Physiological Society
1Arizona State University, Phoenix,
Arizona, USA
2Department of Kinesiology, University
of Virginia, Charlottesville, Virginia,
USA
3Department of Health Sciences,
Central Washington University,
Ellensburg, Washington, USA
Correspondence
Jared M. Dickinson, Department of
Health Sciences, Central Washington
University, 400 E University Way,
Ellensburg, WA 98926, USA.
Email: [email protected]
Funding information
No funding information provided.
Abstract
Maximal oxygen uptake (VO2max) declines with advancing age and is a predic-
tor of morbidity and mortality risk. The purpose here was to assess the utility
of constant load tests performed either above or below peak work rate obtained
from a graded exercise test for verification of VO2max in older adults. Twenty-
two healthy older adults (9M, 13F, 67 ± 6 years, BMI: 26.3 ± 5.1 kg·m−2) par-
ticipated in the study. Participants were asked to complete two experimental
trials in a randomized, counterbalanced cross- over design. Both trials (cycle er-
gometer) consisted of (1) an identical graded exercise test (ramp) and (2) a con-
stant load test at either 85% (CL85; n = 22) or 110% (CL110; n = 20) of the peak
work rate achieved during the associated ramp (performed 10- min post ramp).
No significant differences were observed for peak VO2 (L·min−1) between CL85
(1.86 ± 0.72; p = 0.679) or CL110 (1.79 ± 0.73; p = 0.200) and the associated ramp
(Ramp85, 1.85 ± 0.73; Ramp110, 1.85 ± 0.57). Using the study participant's mean
coefficient of variation in peak VO2 between the two identical ramp tests (2.9%)
to compare individual differences between constant load tests and the associated
ramp revealed 19/22 (86%) of participants achieved a peak VO2 during CL85 that
was similar or higher versus the ramp, while only 13/20 (65%) of participants
achieved a peak VO2 during CL110 that was similar or higher versus the ramp.
These data indicate that if a verification of VO2max is warranted when testing
older adults, a constant load effort at 85% of ramp peak power may be more likely
to verify VO2max as compared to an effort at 110% of ramp peak power.
K E Y W O R D S
aerobic power, aging, exercise physiology, verification, VO2max, VO2peak
2 of 14 |
VILLANUEVA et al.
1 | INTRODUCTION
Advancing age is associated with a variety of physiologi-
cal and biological changes that can contribute to impaired
physical function. Of particular interest is the steady de-
cline in the maximal rate of oxygen uptake (VO2max) that
is well documented during advancing age (Betik & Hepple,
2008; Gries et al., 1985; Kaminsky et al., 2015). Not only is
a reduced VO2max in older adults associated with func-
tional limitations, such as difficulty with walking, climb-
ing stairs, and performing daily activities (Kaminsky et al.,
2013; Paterson et al., 1999, 2004; Paterson & Warburton,
2010), but a low VO2max is a powerful independent pre-
dictor of cardiovascular disease and all- cause mortality
(Imboden et al., 2018; Kokkinos et al., 2010; Myers et al.,
2002; Ross et al., 2016). Moreover, the addition of VO2max
to other traditional risk factors improves risk stratifica-
tion, and inclusion of VO2max to classify morbidity and
mortality risk may be particularly powerful for those on
the lower end of the VO2max spectrum, such as older
adults (Ross et al., 2016). Consequently, developing effec-
tive exercise testing strategies that can be used to verify a
maximal exercise effort, and thus VO2max, in older adults
could have important implications for accurate assess-
ment of morbidity and mortality risk in this population.
Traditionally, VO2max is often assessed through the use
of a graded exercise test, employing either a steady ramp
or an incremental step test until volitional exhaustion. In
theory, a VO2max is achieved when there is no increase
in VO2 with a concomitant increase in power or speed
(Day et al., 2003; Hill & Lupton, 1923; Taylor et al., 1955),
which is often referred to as a VO2 plateau (Taylor et al.,
1955). While sampling rate/interval can influence the oc-
currence of a plateau in VO2 (Astorino, 2009), a plateau is
not always observed, and in fact has been found to only
occur in 17% of VO2max assessments (Day et al., 2003).
The absence of an observed VO2 plateau has produced
queries as to the validatity of these tests for accurately
assessing VO2max (Day et al., 2003; Howley et al., 1995;
Midgley & Carroll, 2009; Poole et al., 2008). Consequently,
the development of secondary criteria that are predicated
on expected values for respiratory exchange ratio, heart
rate (HR), and blood lactate, for example, have been used
to validate a maximal effort (Howley et al., 1995; Midgley
et al., 2007, 2009; Wagner et al., 2020). However, the use
of these secondary criteria is problematic (Poole & Jones,
2017) as these criteria can often be achieved at a “submax-
imal” effort (Poole et al., 2008).
More recently, the use of a secondary constant load test
that is performed following a graded exercise test has been
implemented as a strategy to verify a maximal effort and
VO2max (Costa et al., 2021; Midgley & Carroll, 2009; Poole
et al., 2008). While the use of a constant load test to verify
VO2max has gained consideration, the intensity at which
these constant load tests have been performed is vari-
able. For instance, these constant load bouts have been
performed at work rates below (Day et al., 2003; Murias
et al., 2018; Rossiter et al., 2006; Sedgeman et al., 2013),
equal to (Sawyer et al., 2015), or above (Astorino et al.,
2009; Barker et al., 2011; Hawkins et al., 2007; Iannetta
et al., 2020; Kuffel et al., 2005; Leicht et al., 2013; Midgley
et al., 2006; Murias et al., 2018; Nolan et al., 2014; Poole
et al., 2008; Rossiter et al., 2006; Scharhag- Rosenberger
et al., 2011; Sedgeman et al., 2013; Weatherwax et al.,
2016) those achieved during the preceding graded exer-
cise test [most previous studies employ constant load tests
between 85% and 115% of peak work rate (Astorino et al.,
2009; Barker et al., 2011; Dalleck et al., 2012; Hawkins
et al., 2007; Iannetta et al., 2020; Kuffel et al., 2005; Leicht
et al., 2013; Midgley & Carroll, 2009; Midgley et al., 2006;
Murias et al., 2018; Niemela et al., 1980; Nolan et al., 2014;
Poole et al., 2008; Rossiter et al., 2006; Sawyer et al., 2015;
Scharhag- Rosenberger et al., 2011; Sedgeman et al., 2013;
Weatherwax et al., 2016)]. Furthermore, there is lack of
agreement on the work rate at which the constant load
tests should be performed to best verify a maximal effort
(Breda et al., 1985; Iannetta et al., 2020; Poole & Jones,
2017). Specifically, some propose that if a VO2max is to be
verified, the constant load effort needs to be conducted at
a work rate higher than that achieved during the graded
exercise test (Poole & Jones, 2017). On the other hand,
recent evidence indicates that a work rate below that
achieved during the graded exercise test is more likely to
verify a maximal effort (Iannetta et al., 2020). In partic-
ular, the use of a “submaximal” constant load work rate
to verify VO2max may be more reliable versus the “supra-
maximal” work rate when coupled with graded exercise
test protocols that are shorter in duration (e.g., steeper
ramp protocols) (Iannetta et al., 2020). Consequently, the
use of a constant load work rate below the peak work rate
achieved during the graded exercise test may be a more
reliable strategy to verify VO2max in individuals with a
lower VO2max, such as older adults, who may experience
shorter graded exercise tests.
Therefore, the primary purpose of this study was to em-
ploy a cross- over design to assess the utility of a constant
load test performed at a work rate below (85%) and a work
rate above (110%) the peak work rate achieved during a
graded exercise test (ramp) for validating a maximal ef-
fort and verifying VO2max in healthy older adults. While
comparison of constant load intensities above and below
the peak achieved during a ramp test has been previously
reported (Murias et al., 2018), to our knowledge no study
has employed a randomized, counterbalanced cross- over
design. We hypothesized that in healthy older adults, the
constant load test below the peak work rate of the ramp
| 3 of 14
VILLANUEVA et al.
test would be more likely to verify a maximal effort and
VO2max, which would be demonstrated by a greater num-
ber of individuals achieving a peak VO2 value during the
constant load effort below peak work rate that is similar or
higher to the ramp test as compared to the constant load
effort above peak work rate. In addition, the randomized
cross- over design of the study included the performance
of two identical ramp tests by each participant. Therefore,
a secondary purpose of this study was to evaluate a second
identical ramp test as a strategy to verify VO2max in older
adults.
2 | MATERIALS AND METHODS
2.1 | Participants
Twenty- four healthy older adults volunteered to partici-
pate in this study. All participants were between the ages
of 60– 80 years and were recruited by advertisement, locally
posted flyers, and word of mouth. Participants completed a
brief online pre- screening questionnaire to assess general
health characteristics which was reviewed by a member
of the research team. Following the pre- screening ques-
tionnaire, qualified participants were invited to the labo-
ratory for a formal informed consent process. Additional
screening included a medical history, the Physical Activity
Readiness Questionnaire for Everyone (PARQ+), and
assessment of resting blood pressure. Participants were
excluded if they had uncontrolled hypertension, or any
self- reported heart, liver, kidney, blood, or respiratory dis-
ease, peripheral vascular disease, diabetes or endocrine
disease, active cancer or use of tobacco, self- reported acute
or chronic illness, medical/orthopedic conditions preclud-
ing exercise, or if they were currently training for an en-
durance event (i.e., marathon, triathlon). All participants
provided written informed consent prior to participation.
Participant characteristics for those that participated in the
study are presented in Table 1. This study was approved by
the University Institutional Review Board (in compliance
with the Declaration of Helsinki, as revised in 1983).
2.2 | Study design and procedures
Participants were studied during two separate experimen-
tal trials. The experimental trials were separated on av-
erage by 9 days (range, 6– 14 days) and were performed
in a randomized, counterbalanced cross- over design at a
similar time of day (e.g., morning vs. afternoon). Each ex-
perimental trial consisted of a graded exercise ramp test
and a constant load test that was performed after 10 min
of active rest (pedaling at a work rate no higher than the
warm- up) following completion the ramp test. The ramp
tests were identical for each experimental trial, however,
the visits differed in the work rate at which the subsequent
constant load test was performed.
During each experimental trial participants reported to
the laboratory for testing at least 3 h postprandial and after
abstaining from caffeine, alcohol, supplements, and exer-
cise for at least 24 h. The participant's height and weight
were measured on a calibrated stadiometer and resting
blood pressure measures were obtained during each visit
(Dinamap® PRO 100 Vital Signs Monitor; GE Healthcare).
Participants were equipped with a mouthpiece connected
to a standard nonrebreathing valve (Hans Rudolph) for
continuous measurement of ventilation and respiratory
gas exchange data using a TrueOne 2400 metabolic cart
(Parvomedics). A standard calibration was performed
before each test per manufacturer recommendations.
Participants were also equipped with a chest worn HR mon-
itor (Polar, Inc.) to continuously monitor HR. After 2 min
of rest, participants performed a standardized warm- up in
which the participants pedaled at a cadence of their choice,
between 50 and 90 revolutions per minute (RPM), on a sta-
tionary cycle ergometer (Ergoline Viasprint 150) at 50 W
for males and 40 W for females for 5 min. The chosen RPM
was maintained for the remainder of the testing.
Ramp test
During both experimental trials, participants performed
an identical ramp test on a cycle ergometer. Immediately
following the warm- up phase (described above), the work
rate on the cycle ergometer was increased in a ramp fash-
ion corresponding to 20 W·min−1 for males (1 W every 3 s)
and 15 W·min−1 for females (1 W every 4 s) until volitional
exhaustion. Ratings of perceived exertion (RPE) were as-
sessed every 60 s throughout the duration of the ramp test.
The test was terminated at volitional exhaustion or if the
TABLE 1 Participant characteristics
Men
(n = 9)
Women
(n = 13)
Total
(n = 22)
Age, year
69 ± 6
65 ± 6
67 ± 6
Height, cm
172 ± 9
161 ± 5
165 ± 9
Weight, kg
77 ± 18
69 ± 16
72 ± 17
BMI, kg·m−2
26.0 ± 4.1
26.6 ± 5.8
26.3 ± 5.1
Body fat, %
28.1 ± 6.0
37.8 ± 10.5
34.0 ± 10.0
Lean body mass, kg
53 ± 12
39 ± 3
44 ± 10
Data are presented as mean ± SD.
Abbreviation: BMI, body mass index; Body fat % is whole body derived from
dual energy x- ray absorptiometry.
4 of 14 |
VILLANUEVA et al.
participant was unable to maintain his/her RPM despite
verbal encouragement.
Constant load test
During each experimental trial a constant load test was
completed following the ramp test, which occurred after
10 min of light active recovery (pedaling at a work rate
no higher than the warm- up) on the stationary ergometer.
During active recovery, the participants were provided a
break from the breathing valve, which was reconnected to
the participant at least 3 min prior to the start of the con-
stant load test. The constant load test consisted of cycling
at a work rate equivalent to either 85% (CL85) or 110%
(CL110) of the peak work rate reached during the preced-
ing ramp test. Specifically, in a randomized, counterbal-
anced cross- over design, participants were randomized
to perform either CL85 or CL110 during the first visit,
whereas during the second visit the participant completed
the constant load test at the other work rate. Participants
were instructed to increase cadence as the resistance on
the cycle ergometer increased from that during active re-
covery to the prescribed intensity. Both constant load tests
were performed at a constant work rate until volitional
exhaustion. RPE was assessed at the end of the constant
load test. The test was terminated at volitional exhaustion
(i.e., the participant requesting to stop) or if the partici-
pant was unable to maintain his/her RPM despite verbal
encouragement.
Assessment of body composition
During the second visit, participants underwent a dual-
energy x- ray absorptiometry (DEXA) whole- body scan
(Lunar iDXA, GE Healthcare). The DEXA was per-
formed prior to any testing and after voiding the bladder.
Participants laid down on the DEXA for 15 min prior to
the DEXA to avoid any influence of fluid shifts. A trained
and certified radiologist administered the DEXA scan.
2.3 | Assessment of physiological
outcomes
All ventilation and gas exchange data were assessed using
10- s average measurements, with O2 and CO2 concentra-
tion of expired air derived from samples obtained from a
mixing chamber. Peak VO2 values for the ramp tests and
constant load tests were taken as the highest three con-
secutive 10- s measurements, which were averaged to yield
data collected over a 30- s timeframe. Peak RER values
were taken as an average of the three 10- s measurements
at the same time point as peak VO2. Peak HR for the ramp
and constant load tests were taken as the highest recorded
HR. Peak power during the ramp was identified as the
highest work rate achieved prior to a drop in cadence or
volitional exhaustion. Individual data were calculated to
determine the percent change of physiological outcomes
between the constant load test and the associated ramp,
as well as between the ramp during the first visit (Ramp1)
compared to the ramp during the second visit (Ramp2).
The mean coefficient of variation (CV) for Ramp1 and
Ramp2 was used to identify if a similar (within CV) or
a higher or lower value (outside CV) for a physiological
variable occurred between the constant load test and the
associated ramp test and between Ramp1 and Ramp2.
2.4 | Statistical analysis
All data were tested for normality through skewness and
kurtosis analyses and visual inspection of the normality
plots using SPSS v.24 (IBM). A one- way, repeated meas-
ures, analysis of variance (ANOVA) was used to assess
differences between the ramp tests and the constant load
tests for all outcomes. Pairwise comparisons were per-
formed following the ANOVA using a least significant
difference (LSD) post hoc analyses adjusted for the fol-
lowing two comparisons: constant load test at 85% of peak
work rate (CL85) versus the associated ramp (Ramp85);
and constant load test at 110% of peak work rate (CL110)
versus the associated ramp (Ramp110). Outcome vari-
ables obtained from the first (Ramp1) and second ramp
test (Ramp2) were compared using a dependent t- test
for equivalence. Pearson's correlations were used to de-
termine the relationships between variables for the con-
stant load test versus ramp and for Ramp1 versus Ramp2.
Bland– Altman plots and CVs were used to compare the
agreement for all variables between the constant load
test and associated ramp test and between Ramp1 and
Ramp2. Intraclass correlation coefficients (ICCs) were
used to examine the reliability of peak VO2 and peak HR
between the constant load test and associated ramp test
and between Ramp1 and Ramp2. All comparisons includ-
ing a constant load test were made to the ramp performed
during the same experimental trial. Pearson's correlations
were also used to examine the following relationships
within each experimental trial: (1) Difference in peak
VO2 (L·min−1) between CL85 and Ramp85 and time to ex-
haustion for CL85, (2) Difference in peak VO2 (L·min−1)
between CL85 and Ramp85 and time to exhaustion of
Ramp85, (3) Difference in peak VO2 (L·min−1) between
CL110 and Ramp110 and time to exhaustion of CL110,
and (4) Difference in peak VO2 (L·min−1) between CL110
| 5 of 14
VILLANUEVA et al.
and Ramp110 and time to exhaustion during Ramp110.
All data were analyzed using SPSS Software (SPSS v24)
and significance was set a priori at p ≤ 0.05. All data are
presented as means ± SD.
3 | RESULTS
Of the 24 participants who enrolled in the study, two
participants were excluded during the screening process
(one for high blood pressure, one for underlying medical
disease). Two additional participants dropped out of the
study after completing the first visit due to circumstances
unrelated to the study. Both of these participants only
completed the CL85 exercise trial, and these participants
were included in the analysis for ramp versus CL85 (e.g.,
CL85, n = 22; CL110, n = 20). Only participants who com-
pleted both trials (n = 20; 67 ± 6 year, 8 males and 12 fe-
males) were included in the comparisons between Ramp1
and Ramp2. Participant characteristics are presented in
Table 1. In addition, the peak VO2 (mLO2·kg−1·min−1)
and peak HR of these participants are expressed relative
to age- based reference standards (Kaminsky et al., 2015)
in Table 2.
3.1 | Ramp versus constant load test
(group data)
Peak VO2 (L·min−1) did not differ (p = 0.679) be-
tween Ramp85 (1.85 ± 0.73 L·min−1) and CL85
(1.86 ± 0.72 L·min−1) (CV = 2.07 ± 2.14%) (Table 3,
Figures 1a and 2a). Similarly, peak VO2 was not sig-
nificantly different (p = 0.200) between Ramp110
(1.85 ± 0.57 L·min−1) and CL110 (1.79 ± 0.73 L·min−1)
(CV =3.64% ± 4.47%) (Table 3, Figures 1b and 2b).
Intraclass correlations also showed agreement in peak
VO2 (L·min−1) between the ramp and constant load test
for both CL85 (ICC = 0.997) and CL110 (ICC = 0.979)
(Table 3). Time to exhaustion during the ramp and con-
stant load tests were examined to determine whether
time to exhaustion of the various tests influenced differ-
ences in peak VO2 between the constant load test and
associated ramp. Time to exhaustion for Ramp85 was not
statistically correlated with the difference in peak VO2
between CL85 and Ramp85 (r = 0.17; p = 0.458) (Figure
3a). However, a longer Ramp110 time to exhaustion was
negatively associated with the difference in peak VO2 be-
tween CL110 and Ramp110 (r = 0.48; p = 0.031) (Figure
3b), indicating that a longer time to exhaustion during
the ramp test was associated with a greater likelihood of
attaining a lower peak VO2 during CL110 compared to
the ramp. Time to exhaustion for the respective constant
load test protocols was not statistically correlated with
the difference in peak VO2 between CL85 and the associ-
ated ramp (r = 0.33; p = 0.134) (Figure 3c) or between
CL110 and the associated ramp (r = 0.20; p = 0.393)
(Figure 3d).
Peak HR did not differ (p = 0.243) between Ramp85
(150 ± 17 bpm) and CL85 (153 ± 17 bpm) (Table 3,
Figures 1c and 2c). Similarly, peak HR did not differ
(p = 0.085) between Ramp110 (149 ± 16 bpm) and CL110
(146 ± 16 bpm) (Table 3, Figures 1d and 2d). Intraclass
correlations showed agreement in peak HR between
ramp and constant load test for both CL85 (ICC = 0.950)
and CL110 (ICC = 0.906). Peak RER was significantly
different (p < 0.01) between Ramp85 (1.17 ± 0.09) and
CL85 (1.07 ± 0.08) (Table 3). Similarly, peak RER was
significantly different (p < 0.01) between Ramp110
(1.16 ± 0.08) and CL110 (1.03 ± 1.0) (Table 3). Peak RPE
did not differ (p = 0.602) between Ramp85 (18.5 ± 1.3)
and CL85 (18.3 ± 1.7). Similarly, peak RPE did not differ
(p = 0.629) between Ramp110 (18.7 ± 1.0) and CL110
(18.6 ± 1.1).
Study participants
Reference
Percentile
Males
Peak VO2
(mlO2·kg−1·min−1)
29.8 ± 9.6 (18.5– 49.9)
29.4 ± 7.9
~50th
Peak HR (bpm)
159 ± 17 (135– 186)
158 ± 17
N/A
Females
Peak VO2
(mLO2·kg−1·min−1)
24.2 ± 10.5 (14.1– 47.9)
20.7 ± 5.0
~75th
Peak HR (bpm)
147 ± 17 (120– 175)
157 ± 17
N/A
Study Participant data (9M, 13F, 67 ± 6 years) are presented as mean ± SD (range) from the first visit
ramp test. Reference and percentile data are derived from FRIEND for age 60– 69 years (Kaminsky et al.,
2015).
Abbreviation: bpm, beats per minute.
TABLE 2 Study participants relative
peak VO2 (mLO2·kg−1·min−1) and heart
rate (HR) in comparison to reference
standards derived from FRIEND
(Kaminsky et al., 2015)
6 of 14 |
VILLANUEVA et al.
3.2 | Ramp1 versus Ramp2 (group data)
Peak VO2 during Ramp1 (1.82 ± 0.72 L·min−1) was
not significantly different (p = 0.100) from Ramp2
(1.86 ± 0.81 L·min−1) (CV = 2.90 ± 1.89%) (Table 3). Peak
VO2 was also strongly correlated (R2 = 0.987) (p < 0.01) and
was in high agreement (ICC = 0.994) between Ramp1 and
Ramp2 (Figure 4). Peak HR did not differ (p = 0.115) be-
tween Ramp1 (150 ± 17 bpm) and Ramp2 (149 ± 15 bpm)
(CV = 2.30 ± 2.06%) (Table 3) and values were strongly
correlated (R2 = 0.876) (p < 0.01) and in high agreement
(ICC = 0.936) between Ramp1 and Ramp2. RER did not
differ (p = 0.348) between Ramp1 (1.16 ± 0.09) and Ramp2
(1.16 ± 0.08) (CV = 3.20 ± 2.05%) (Table 3) and values
were correlated (R2 = 0.529) (p < 0.01) (Table 3) and in
agreement (ICC = 0.727). Peak power output (W) did not
differ between Ramp1 (156 ± 53) and Ramp2 (158 ± 53)
(CV = 5.3 ± 5.40%) (Table 3) and values were strongly
correlated (R2 = 0.905) (p < 0.01) and in high agreement
(ICC = 0.951) between Ramp1 and Ramp2. RPE did not
differ (p = 0.481) between Ramp1 (18.5 ± 1.1) and Ramp2
(18.6 ± 1.3) and values were correlated (R2 = 0.480)
(p < 0.01).
3.3 | Individual data
We calculated the mean individual participant CV (%)
between Ramp1 and Ramp2 to examine individual dif-
ferences in physiological variables between the ramps
(Ramp1 vs. Ramp2) and between the constant load tests
and their associated ramp tests. Using this participant-
based CV- derived cut point from Ramp1 and Ramp2 (see
Table 3), 68% of participants (15 of 22) achieved a peak
VO2 during CL85 that was similar (within 2.9%, CV be-
tween Ramp1 and Ramp2 for peak VO2) to the associated
TABLE 3 Physiological group and individual responses to the ramp and constant load tests
Peak VO2
(L·min−1)
Peak HR
(bpm)
VE
(L·min−1)
Peak RER
Power
(W)
Time to
exhaustion (s)
CV (%)a
2.9
2.3
6.3
3.2
5.3
8.0
Ramp 1 versus Ramp 2
Ramp1
1.82 ± 0.72
150 ± 17
76.91 ± 31.69
1.16 ± 0.09
156 ± 53
402 ± 151
Ramp2
1.86 ± 0.81
149 ± 15
79.31 ± 31.84
1.16 ± 0.08
158 ± 53
408 ± 160
Ind. Similarb
(9/20)
(9/19)
(9/20)
(7/20)
(9/20)
(9/20)
Ind. Higherb
(8/20)
(3/19)
(6/20)
(5/20)
(7/20)
(6/20)
Ind. Lowerb
(3/20)
(7/19)
(5/20)
(8/20)
(4/20)
(5/20)
Ramp versus constant load test at 85%
Ramp
1.85 ± 0.73
150 ± 17
77.85 ± 30.01
1.17 ± 0.09
158 ± 52
401 ± 142
CL85
1.86 ± 0.72
153 ± 17
80.15 ± 30.20
1.07 ± 0.08*
133 ± 45
185 ± 88
Ind. Similarb
(15/22)
(11/21)
(10/22)
(1/22)
Ind. Higherb
(4/22)
(7/21)
(8/22)
(2/22)
Ind. Lowerb
(3/22)
(3/21)
(4/22)
(19/22)
Ramp versus constant load test at 110%
Ramp
1.85 ± 0.57
149 ± 16
78.28 ± 33.63
1.16 ± 0.08
156 ± 54
410 ± 162
CL110
1.79 ± 0.73
146 ± 16
75.83 ± 34.83
1.03 ± 0.10*
170 ± 60
79 ± 62
Ind. Similarb
(8/20)
(7/19)
(9/20)
(2/20)
Ind. Higherb
(5/20)
(3/19)
(6/20)
(0/20)
Ind. Lowerb
(7/20)
(9/19)
(5/20)
(18/20)
Data are presented as mean ± SD.
aMean individual participant coefficient of variation (CV) from Ramp1 to Ramp2, presented as percent (%).
bNumber of participants with values within the CV (for Ramp1 to Ramp2) between tests (similar), a value that is identified as higher (>CV for Ramp1 to
Ramp2) compared to the Ramp (or compared to Ramp 1 for Ramp2 vs. Ramp1) (higher), a value that is identified as lower (>CV for Ramp1 to Ramp2)
compared to the Ramp (or compared to Ramp 1 for Ramp2 vs. Ramp1) (lower). HR, heart rate; RER, respiratory exchange ratio; VE, ventilation; Ind. Similar,
represents the number of participants that achieved a similar value (within CV) during the constant load test versus the associated ramp or for Ramp2 versus
Ramp1; Ind. Higher, represents the number of participants that achieved a higher value (outside CV) during the constant load (CL) test versus the associated
ramp or for Ramp2 versus Ramp1; Ind. Lower, represents the number of participants that achieved a lower value (outside CV) during the constant load test
versus the associated ramp or for Ramp2 versus Ramp1.
*p < 0.05 Ramp.
| 7 of 14
VILLANUEVA et al.
FIGURE 1 Bland– Altman plots for peak oxygen uptake (VO2, L·min−1) and heart rate (HR). Presented are (a) peak VO2 obtained
during the constant load test performed at 85% of ramp peak work rate (CL85) and the associated ramp test (Ramp85), (b) peak VO2
obtained during the constant load test performed at 110% of ramp peak work rate (CL110) and the associated ramp test (Ramp110), (c)
peak HR obtained during CL85 and Ramp85, and (d) peak HR obtained during CL110 and Ramp110. Y- axis = constant load test − ramp;
x- axis = mean of ramp and constant load test; dotted lines = mean ± 1.96 × SD; dark solid lines = 0 on the y- axis; light solid lines = mean
of constant load test − ramp. Filled squares (■) represent male participants and open diamonds (♢) represent female participants. Ramp85
versus CL85, n = 22; Ramp110 versus CL110, n = 20
(a)
(c)
(d)
(b)
FIGURE 2 Peak oxygen uptake (VO2, L·min−1) and heart rate (HR) achieved during the ramp (x- axis) and constant load (y- axis) test
for each participant. The dotted lines represent the line of identity (y = x). Presented are (a) peak VO2 obtained during the constant load
test at 85% of ramp peak work rate (CL85) versus the associated ramp (Ramp85), (b) peak VO2 obtained during the constant load test at
110% of ramp peak work rate (CL110) versus the associated ramp (Ramp110), (c) peak HR obtained during CL85 versus Ramp85, and (d)
peak HR obtained during CL110 versus Ramp110. Filled squares (■) represent male participants and open diamonds (♢) represent female
participants. Ramp85 versus CL85, n = 22; Ramp110 versus CL110, n = 20
(a)
(c)
(d)
(b)
8 of 14 |
VILLANUEVA et al.
ramp peak VO2. Furthermore, 18% of participants (4 of 22)
achieved a peak VO2 during CL85 that was >2.9% higher
than that achieved during Ramp85, while 14% of partici-
pants (3 of 22) achieved a peak VO2 during CL85 that was
>2.9% lower than that achieved during Ramp85 (Table 3).
In contrast, 40% of participants (8 of 20) achieved a peak
VO2 during CL110 that was similar to the associated ramp,
25% of participants (5 of 20) achieved a peak VO2 during
CL110 that was >2.9% higher than Ramp110 (Table 3),
and 35% of participants (7 of 20) achieved a peak VO2 dur-
ing CL110 that was >2.9% lower than Ramp110. Similar
results were observed between CL85 and CL110 for peak
HR and ventilation (Table 3).
When comparing Ramp2 to Ramp1, 45% of participants
(9 of 20) achieved a peak VO2 during Ramp2 that was
similar to Ramp1, 40% of participants (8 of 20) achieved
FIGURE 3 Correlations between time to exhaustion (x- axis) and differences in peak oxygen uptake (VO2, L·min−1) achieved during the
constant load and ramp tests. Presented are (a) time to exhaustion during the associated ramp (Ramp85) compared to the difference in peak
VO2 obtained during the constant load test at 85% of ramp peak power (CL85) and Ramp85, (b) time to exhaustion during the associated
ramp (Ramp110) compared to the difference in peak VO2 obtained during the constant load test at 110% of ramp peak power (CL110) and
Ramp110, (c) time to exhaustion during CL85 compared to the difference in peak VO2 obtained during CL85 and Ramp85, and (d) time to
exhaustion during CL110 compared to difference in peak VO2 obtained during CL110 and Ramp110. *p < 0.05. Filled squares (■) represent
male participants and open diamonds (♢) represent female participants. Ramp85 versus CL85, n = 22; Ramp110 versus CL110, n = 20
(a)
(b)
(c)
(d)
FIGURE 4 Comparison of peak VO2 values (L·min−1) achieved during the first ramp test (Ramp1) and the second ramp test (Ramp2).
Presented are (A) Bland– Altman plot for peak VO2 obtained during Ramp1 and Ramp2 [Y- axis = Ramp2 − Ramp1; x- axis = mean of Ramp1
and Ramp2; dotted lines = mean ± 1.96 × SD; dark solid lines = 0 on the y- axis; light solid lines = mean of Ramp1 − Ramp2] and (b) the
relationship between peak VO2 obtained during Ramp1 and Ramp2 [the line represents the line of identity (y = x)]. Filled squares (■)
represent male participants and open diamonds (♢) represent female participants (n = 20)
(a)
(b)
| 9 of 14
VILLANUEVA et al.
a peak VO2 during Ramp2 that was identified as higher
(>2.9% difference) compared to Ramp1, and 15% (3 of 20)
achieved a peak VO2 during Ramp2 that was identified as
lower (>2.9% difference) compared to Ramp1 (Table 3).
Results for peak HR, VE, and RER are also presented in
Table 3.
We recognize the lack of consensus on methodologi-
cal/statistical approaches for confirming VO2max during
a constant load (verification) test (or any secondary test).
Therefore, Table 4 provides additional information on
individual differences/similarities between tests using
±2 × typical error of the two ramp tests (McCarthy
et al., 2021) and a HR of ±2 bpm (Midgley et al., 2006) or
±4 bpm (Midgley et al., 2009) from the peak HR from the
ramp tests. In all instances (study CV, ±2 × typical error,
HR ±2 or ±4 bpm), when compared to CL110, CL85 had a
greater percentage of individuals with a constant load test
that was considered similar to or higher than the ramp.
4 | DISCUSSION
To our knowledge this is the first study to employ a rand-
omized, counterbalanced cross- over design to evaluate the
utility of constant load tests performed above and below
ramp- derived peak work rate to serve as a strategy to verify
a maximal effort and VO2max in healthy older adults. The
primary finding from this investigation is that in healthy
older adults, a constant load test performed at a work
rate slightly below (85%) peak work rate achieved during
a graded exercise test was more likely to verify VO2max
as compared to a constant load test performed at a work
rate above (110%) that achieved during a graded exercise
test. In addition, our data also indicate that while a second
identical ramp test could produce a slightly higher peak
VO2 in a greater number of individuals as compared to the
constant load test at 85% peak work rate, both strategies
yield reasonably similar outcomes for verifying VO2max.
Relative to younger adults, little attention has been
given to the efficacy of a constant load test for verifying a
maximal effort and VO2max in older adults (Dalleck et al.,
2012; Murias et al., 2018). In this study, we examined to
what extent a constant load test performed above (110%)
or below (85%) ramp peak work rate could be used to ver-
ify VO2max in healthy older adults. We specifically chose
these work rates as they represent the range in intensity
used in previous studies that used a constant load “verifi-
cation” test (Astorino et al., 2009; Barker et al., 2011; Costa
et al., 2021; Dalleck et al., 2012; Day et al., 2003; Kuffel
et al., 2005; Midgley & Carroll, 2009; Murias et al., 2018;
Niemela et al., 1980; Poole et al., 2008; Rossiter et al., 2006;
Sawyer et al., 2015; Sedgeman et al., 2013). Consistent with
many previous studies, we did not identify “group” differ-
ences for peak VO2 achieved between the ramp test and
the corresponding constant load test, regardless of inten-
sity. However, examination of the individual participant
Study CV
(±2.9%)
2 × TE
(±0.156 L·min−1)
Heart rate
(±2 bpm)
Heart rate
(±4 bpm)
Ramp 1 versus Ramp 2
Ind. Similara
(9/20)
(17/20)
(8/19)
(11/19)
Ind. Highera
(8/20)
(3/20)
(2/19)
(1/19)
Ind. Lowera
(3/20)
(0/20)
(9/19)
(7/19)
Ramp versus constant load test at 85%
Ind. Similara
(15/22)
(21/22)
(10/21)
(12/21)
Ind. Highera
(4/22)
(1/22)
(7/21)
(7/21)
Ind. Lowera
(3/22)
(0/22)
(4/21)
(2/21)
Ramp versus constant load test at 110%
Ind. Similara
(8/20)
(15/20)
(8/19)
(12/19)
Ind. Highera
(5/20)
(1/20)
(6/19)
(4/19)
Ind. Lowera
(7/20)
(4/20)
(5/19)
(3/19)
aThe criteria for a similar, higher, or lower value were that the value had to be within or outside (±) the
study coefficient of variation (CV), 2 × typical error (TE) (McCarthy et al., 2021), or a heart rate within
2 beats per minute (bpm) (Midgley et al., 2006) or 4 bpm (Midgley et al., 2009) of the peak heart rate
achieved during the ramp. Ind. Similar, represents the number of participants that achieved a similar
value (within cut points) during the constant load test versus the associated ramp or for Ramp2 versus
Ramp1; Ind. Higher, represents the number of participants that achieved a higher value (outside cut
point) during the constant load test versus the associated ramp or for Ramp2 versus Ramp1; Ind. Lower,
represents the number of participants that achieved a lower value (outside cut point) during the constant
load test versus the associated ramp or for Ramp2 versus Ramp1.
TABLE 4 Comparison of various
individual data “cut points” used in the
literature to determine verification of
VO2max
10 of 14 |
VILLANUEVA et al.
data revealed a greater likelihood for the CL85 test to vali-
date a maximal effort and VO2max as compared to CL110.
Specifically, only 3 of the 22 participants (~14%) achieved
a peak VO2 during the CL85 that was lower (outside the
CV of the two ramp tests) than the value achieved during
the ramp test. These data indicate that ~86% of the par-
ticipants (19 of 22) achieved a peak VO2 during the CL85
that was either similar (15 of 22 participants, within the
CV of the two ramp tests) or higher (4 of 22 participants,
>CV of the two ramp tests) than that achieved during the
associated ramp test.
In contrast, 7 of 20 participants (~35%) achieved a peak
VO2 during the CL110 test that was lower (>CV of the two
ramp tests) than the value achieved during the ramp test,
and thus only ~65% achieved a value that was similar (8 of
20 participants) or higher (5 of 20 participants) than the
associated ramp test. While we acknowledge previously
proposed rationale that the constant load “verification”
test should, theoretically, be conducted at a work rate
higher than that achieved during the ramp test (e.g., su-
pramaximal) (Poole & Jones, 2017), the present results in-
dicate that a constant load test performed at a work rate of
110% of ramp peak power may be too high for some older
adults as a method to verify a maximal effort and VO2max.
Moreover, the greater agreement in VO2peak between
the ramp test and CL85 as compared to the ramp test and
CL110 is also evident through examination of the limits of
agreement and bias presented in the Bland– Altman plots
(Figure 1a and 1b), as well as when employing other cut
points used in the literature (see Table 4). Collectively, our
findings further support (Iannetta et al., 2020) the use of a
work rate slightly below peak ramp work rate, as opposed
to above, when a constant load test to verity a maximal
effort and VO2max in healthy older adults is warranted.
Moreover, these results also further support the use of in-
dividual data for assessment of VO2max and comparison
of constant load “verification” test intensities (Noakes,
2008).
As expected, the CL110 test elicited a shorter exercise
duration (mean ~79 s [range, 30– 330 s]) compared to
CL85 (mean ~185 s [range, 50– 457 s]). Previous research
in older adults that used a constant load test at 105% of
ramp peak work rate reported mean durations of ~102 s
(Murias et al., 2018) and ~150 s (Dalleck et al., 2012). The
shorter duration observed during CL110 in this study
may be due to the 5% difference in constant load test
work rate in participants of approximately the same age
(Dalleck et al., 2012; Murias et al., 2018). It is also im-
portant to note that the greater likelihood of lower peak
VO2 values during CL110 could be the result of a reduced
contribution of the slow component of VO2 (Gaesser &
Poole, 1996). Specifically, it has been reported that an ex-
ercise duration of >3 min is necessary to observe changes
in VO2 kinetics that are due to the VO2 slow component
(Gaesser & Poole, 1996). However, we did not observe
any significant correlations between exercise time of
the constant load test and agreement between peak VO2
achieved during the ramp and corresponding constant
load test (Figure 3). Interestingly, we did observe that a
longer time to exhaustion during Ramp110 (thus, higher
peak power) was more likely to result in a lower peak
VO2 during CL110. This finding would appear to agree
with previous work suggesting that a peak VO2 achieved
during a ramp protocol that resulted in a higher peak
power was less likely to be validated with a constant load
effort above the ramp peak power (Iannetta et al., 2020).
To that end, with the exception of one participant who
had a history of cycling (highest VO2max), participants
were relatively unaccustomed to cycling exercise. Thus,
the lower likelihood of verifying VO2max in these older
adults when using a constant load test above ramp peak
work rate may be due to an inability to tolerate the physio-
logical demands of such high work rates for a sufficiently
long enough time to elicit VO2max. This may also explain
why nearly 50% (9 of 19) of the participants achieved a
peak HR during CL110 that was lower (outside the CV of
the two ramp tests) than that achieved during the associ-
ated ramp.
In this study, participants completed two identi-
cal ramp assessments approximately 1 week apart
(mean = 9 days). We chose this time frame to provide
adequate recovery time from the previous test. The mean
CV observed for peak VO2 between the two ramp tests
is consistent with ranges identified in previous reports
(Fielding et al., 1997; Foster et al., 1986; Skinner et al.,
1999), and as discussed above, we utilized the mean
participant CV (%) from the two identical ramp tests to
identify individual differences in physiological variables
between ramp and constant load tests. The design of the
study also allowed us to examine to what extent a sec-
ond ramp test could be used to assess/verify VO2max in
older adults. Consistent with previous reports (Foster
et al., 1986), we did not observe any significant differ-
ences in any physiological variable between the first visit
(Ramp1) and the second visit (Ramp2). In addition, using
the CV- derived cut point, the number of participants that
achieved a similar or higher peak VO2 during Ramp2
compared to Ramp1 (17/20 participants) was similar to
that observed when comparing CL85 to the ramp (19/22
participants). However, when compared to the ramp ver-
sus constant load test comparisons, more participants
achieved a higher peak VO2 during Ramp2 compared to
Ramp1 (40%; 8/20 participants). Importantly, these dis-
crepancies in peak VO2 achieved during the ramp in the
first and second experimental trial did not impact the
comparison between the associated ramp and constant
| 11 of 14
VILLANUEVA et al.
load tests. Not only was the study counter- balanced,
but among participants who completed both trials and
achieved a peak VO2 during a constant load test that was
different compared to the associated ramp test, there was
a similar number of participants who achieved a different
(higher or lower) peak VO2 during the constant load test
during the first (higher value, n = 5; lower value, n = 4)
and during the second experimental trial (higher value,
n = 4; lower value, n = 6). Collectively, these data indi-
cate that some individuals may not be accustomed to the
maximal intensity of exercise, the mode of exercise, or
perhaps the breathing apparatus (Poole & Jones, 2012).
Moreover, the results of this study indicate that a famil-
iarization trial or second ramp could also increase the
accuracy of VO2max assessments in some older adults,
perhaps for a slightly greater number of individuals as
compared to the use of a constant load test.
Peak HR was not different during the ramp test and
either constant load test intensity. This finding contrasts
with the results of a previous study with older adults that
found a significantly higher peak HR during a ramp test as
compared to a supramaximal verification test (105%) and
submaximal (85%) verification test (Murias et al., 2018), al-
though the magnitude of difference in that study (Murias
et al., 2018) was extremely small (1– 2 bpm). Moreover,
similar to VO2max discussed above, individual data in-
dicate that a greater number of participants achieved a
similar or higher peak HR during CL85 versus the ramp
as compared to CL110 versus the ramp (86% vs. 53%). In
addition, the individual data and visual inspection of the
Bland– Altman plots suggest a greater likelihood for par-
ticipants to achieve a lower peak HR during CL110 ver-
sus the ramp as compared to CL85. Together with the VO2
data, these peak HR data further support the incorpora-
tion of a constant load test performed slightly below peak
ramp work rate for verification of maximal values in older
adults.
We recognize that previous studies have utilized
rest periods as short as 3 min and as long as a full
week between ramp and constant load verification
tests (Astorino et al., 2009; Barker et al., 2011; Dalleck
et al., 2012; Day et al., 2003; Hawkins et al., 2007; Kuffel
et al., 2005; Leicht et al., 2013; Midgley & Carroll, 2009;
Midgley et al., 2006; Murias et al., 2018; Niemela et al.,
1980; Nolan et al., 2014; Poole et al., 2008; Rossiter et al.,
2006; Sawyer et al., 2015; Scharhag- Rosenberger et al.,
2011; Sedgeman et al., 2013; Weatherwax et al., 2016),
and thus we cannot extend our findings to situations
that may utilize different rest periods between tests.
However, we specifically employed a 10- min active rest
period between the end of the ramp test and the initia-
tion of the constant load test as this timeframe is likely
to be more practical for future research and clinical
practice as participants would not be required to come
back for testing at a later time or date. In addition, it
is possible that our findings may have been influenced
by the duration of the ramp test (Iannetta et al., 2020).
Similarly, some reports indicate that a valid VO2max is
achieved with a ramp test of at least 8 min (Buchfuhrer
et al., 1983), although this notion has been challenged
(Midgley et al., 2008). Finally, we acknowledge that
the necessity of verification tests has been questioned
(Murias et al., 2018; Wagner et al., 2021), perhaps on the
basis that a high percentage of verification tests yield
peak VO2 values that are considered similar to the ramp
tests. Indeed, if the graded exercise test was a maximal
effort, then in theory the ramp and constant load tests
should yield similar values. In addition, it is important
to note that previous studies (see (Costa et al., 2021)), as
well as data from the current investigation, demonstrate
that not all ramp tests will yield maximal VO2 values (or
values that are similar between the ramp and secondary
verification test). Importantly, those ramp efforts that
do and do not produce maximal values could not be
identified without employing a secondary test to verify
the results. Future investigators and/or clinicians will
need to determine, for their specific use, the necessity
to obtain an accurate measurement of VO2max and to
what extent a value requires “verification” using a sin-
gle visit or multiple visit approach.
In conclusion, these findings have implications for
the evaluation of VO2max of older adults in both a re-
search and clinical setting. In particular, given the over-
whelming data to suggest VO2max/cardiorespiratory
fitness is perhaps the most powerful predictor of cardio-
vascular disease risk (Kokkinos et al., 2010; Myers et al.,
2002; Ross et al., 2016), identifying strategies to obtain
an accurate assessment of VO2max in older adults will
serve to better identify individuals at risk for cardiovas-
cular disease as well as those with increased risk of mor-
bidity and mortality. Specifically, our data indicate that
when verification of maximal values is warranted in a
single testing session, a constant load test performed at
85% of ramp peak power is more likely to verify a max-
imal effort and VO2max in older adults as compared to
a constant load test at 110% ramp peak power. On the
other hand, in situations where multiple participant vis-
its are feasible, performing an additional ramp test may
also serve to verify VO2max, and could potentially lead
to higher values in a slightly greater number of partici-
pants. However, the logistics and associated participant
burden of recovery times between tests in a single ses-
sion and/or multiple visits must be considered in the ap-
plication of constant load testing to verify VO2max in the
real- world settings (especially clinical environments and
clinical populations).
12 of 14 |
VILLANUEVA et al.
ACKNOWLEDGMENTS
The authors thank the participants for their time. I.R.V.,
S.S.A., G.A.G., and J.M.D. designed research; I.R.V.,
J.C.C., S.M.M., T.M.J., S.L.W., S.S.A., G.A.G., and J.M.D.
conducted research; I.R.V. and J.M.D. analyzed data and
performed statistical analysis; I.R.V. and J.M.D. wrote
the manuscript and have primary responsibility for final
content; all authors approved the final version of the
manuscript.
DISCLOSURES
The authors have no conflict of interest to declare.
ORCID
Jared M. Dickinson
https://orcid.
org/0000-0003-3142-938X
REFERENCES
Astorino, T. A. (2009). Alterations in VOmax and the VO plateau
with manipulation of sampling interval. Clinical Physiology
and Functional Imaging, 29, 60– 67.
Astorino, T. A., White, A. C., & Dalleck, L. C. (2009). Supramaximal
testing to confirm attainment of VO2max in sedentary men and
women. International Journal of Sports Medicine, 30, 279– 284.
Barker, A. R., Williams, C. A., Jones, A. M., & Armstrong, N. (2011).
Establishing maximal oxygen uptake in young people during a
ramp cycle test to exhaustion. British Journal of Sports Medicine,
45, 498– 503. https://doi.org/10.1136/bjsm.2009.063180
Betik, A. C., & Hepple, R. T. (2008). Determinants of VO2max de-
cline with aging: An integrated perspective. Applied Physiology,
Nutrition and Metabolism, 33, 130– 140.
Buchfuhrer, M. J., Hansen, J. E., Robinson, T. E., Sue, D. Y.,
Wasserman, K., & Whipp, B. J. (1983). Optimizing the ex-
ercise protocol for cardiopulmonary assessment. Journal of
Applied Physiology, 55, 1558– 1564. https://doi.org/10.1152/
jappl.1983.55.5.1558
Costa, V. A. B., Midgley, A. W., Carroll, S., Astorino, T. A., de
Paula, T., Farinatti, P., & Cunha, F. A. (2021). Is a verifica-
tion phase useful for confirming maximal oxygen uptake in
apparently healthy adults? A systematic review and meta-
analysis. PLoS One, 16, e0247057. https://doi.org/10.1371/
journ al.pone.0247057
Dalleck, L. C., Astorino, T. A., Erickson, R. M., McCarthy, C. M.,
Beadell, A. A., & Botten, B. H. (2012). Suitability of verification
testing to confirm attainment of VO2max in middle- aged and
older adults. Research in Sports Medicine, 20, 118– 128.
Day, J. R., Rossiter, H. B., Coats, E. M., Skasick, A., & Whipp, B.
J. (2003). The maximally attainable VO2 during exercise in
humans: The peak vs. maximum issue. Journal of Applied
Physiology, 95, 1901– 1907.
Fielding, R. A., Frontera, W. R., Hughes, V. A., Fisher, E. C., & Evans,
W. J. (1997). The reproducibility of the Bruce protocol exercise
test for the determination of aerobic capacity in older women.
Medicine and Science in Sports and Exercise, 29, 1109– 1113.
https://doi.org/10.1097/00005 768- 19970 8000- 00018
Foster, V. L., Hume, G. J., Dickinson, A. L., Chatfield, S. J., & Byrnes,
W. C. (1986). The reproducibility of VO2max, ventilatory, and
lactate thresholds in elderly women. Medicine and Science in
Sports and Exercise, 18, 425– 430. https://doi.org/10.1249/00005
768- 19860 8000- 00011
Gaesser, G. A., & Poole, D. C. (1996). The slow component of ox-
ygen uptake kinetics in humans. Exercise and Sport Sciences
Reviews, 24, 35– 71. https://doi.org/10.1249/00003 677- 19960
0240- 00004
Gries, K. J., Raue, U., Perkins, R. K., Lavin, K. M., Overstreet, B. S.,
D'Acquisto, L. J., Graham, B., Finch, W. H., Kaminsky, L. A.,
Trappe, T. A., & Trappe, S. (1985). Cardiovascular and skele-
tal muscle health with lifelong exercise. Journal of Applied
Physiology, 125(5), 1636– 1645. https://doi.org/10.1152/jappl
physi ol.00174.2018
Hawkins, M. N., Raven, P. B., Snell, P. G., Stray- Gundersen, J., &
Levine, B. D. (2007). Maximal oxygen uptake as a parametric
measure of cardiorespiratory capacity. Medicine and Science in
Sports and Exercise, 39, 103– 107.
Hill, A. V., & Lupton, H. (1923). Muscular exercise, lactic acid, and
the supply and utilisation of oxygen. QJM, 16, 135– 171.
Howley, E. T., Bassett, D. R. Jr, & Welch, H. G. (1995). Criteria for
maximal oxygen uptake: Review and commentary. Medicine
and Science in Sports and Exercise, 27, 1292– 1301.
Iannetta, D., de Almeida, A. R., Ingram, C. P., Keir, D. A., & Murias,
J. M. (2020). Evaluating the suitability of supra- POpeak ver-
ification trials after ramp- incremental exercise to confirm
the attainment of maximum O2 uptake. American Journal of
Physiology: Regulatory, Integrative and Comparative Physiology,
319, R315– R322.
Imboden, M. T., Harber, M. P., Whaley, M. H., Finch, W. H., Bishop,
D. L., & Kaminsky, L. A. (2018). Cardiorespiratory fitness and
mortality in healthy men and women. Journal of the American
College of Cardiology, 72, 2283– 2292. https://doi.org/10.1016/j.
jacc.2018.08.2166
Kaminsky, L. A., Arena, R., Beckie, T. M., Brubaker, P. H., Church,
T. S., Forman, D. E., Franklin, B. A., Gulati, M., Lavie, C. J.,
Myers, J., Patel, M. J., Pina, I. L., Weintraub, W. S., Williams,
M. A., Council Clinical C, & Council Nutr Phys Activity M.
(2013). The importance of cardiorespiratory fitness in the
United States: The need for a National Registry a policy state-
ment from the American Heart Association. Circulation, 127,
652– 662.
Kaminsky, L. A., Arena, R., & Myers, J. (2015). Reference standards
for cardiorespiratory fitness measured with cardiopulmonary
exercise testing: Data from the fitness registry and the impor-
tance of exercise national database. Mayo Clinic Proceedings,
90, 1515– 1523.
Kokkinos, P., Myers, J., Faselis, C., Panagiotakos, D. B., Doumas, M.,
Pittaras, A., Manolis, A., Kokkinos, J. P., Karasik, P., Greenberg,
M., Papademetriou, V., & Fletcher, R. (2010). Exercise capac-
ity and mortality in older men: A 20- year follow- up study.
Circulation, 122, 790– 797.
Kuffel, E. E., Foster, C., & Zabrowski, J. (2005). VO2max during
Successive maximal efforts. Medicine and Science in Sports and
Exercise, 37, S98– S99. https://doi.org/10.1249/00005 768- 20050
5001- 00520
Leicht, C. A., Tolfrey, K., Lenton, J. P., Bishop, N. C., & Goosey-
Tolfrey, V. L. (2013). The verification phase and reliability of
physiological parameters in peak testing of elite wheelchair
athletes. European Journal of Applied Physiology, 113, 337– 345.
https://doi.org/10.1007/s0042 1- 012- 2441- 6
| 13 of 14
VILLANUEVA et al.
McCarthy, S. F., Leung, J. M. P., & Hazell, T. J. (2021). Is a verifica-
tion phase needed to determine VO2max across fitness levels?
European Journal of Applied Physiology, 3, 861– 870.
Midgley, A. W., Bentley, D. J., Luttikholt, H., McNaughton, L. R., &
Millet, G. P. (2008). Challenging a dogma of exercise physiol-
ogy: Does an incremental exercise test for valid VO2max deter-
mination really need to last between 8 and 12 minutes? Sports
Medicine, 38, 441– 447. https://doi.org/10.2165/00007 256- 20083
8060- 00001
Midgley, A. W., & Carroll, S. (2009). Emergence of the verification
phase procedure for confirming ‘true’ VO2max. Scandinavian
Journal of Medicine & Science in Sports, 19, 313– 322.
Midgley, A. W., Carroll, S., Marchant, D., McNaughton, L. R., &
Siegler, J. (2009). Evaluation of true maximal oxygen up-
take based on a novel set of standardized criteria. Applied
Physiology, Nutrition and Metabolism, 34, 115– 123. https://doi.
org/10.1139/H08- 146
Midgley, A. W., McNaughton, L. R., & Carroll, S. (2006).
Verification phase as a useful tool in the determination of
the maximal oxygen uptake of distance runners. Applied
Physiology Nutrition and Metabolism- Physiologie Appliquee
Nutrition
Et
Metabolisme,
31,
541– 548.
https://doi.
org/10.1139/h06- 023
Midgley, A. W., McNaughton, L. R., Polman, R., & Marchant, D.
(2007). Criteria for determination of maximal oxygen uptake: A
brief critique and recommendations for future research. Sports
Medicine, 37, 1019– 1028.
Murias, J. M., Pogliaghi, S., & Paterson, D. H. (2018). Measurement
of a true (V)OverdotO(2max) during a Ramp Incremental
Test Is Not Confirmed by a Verification Phase. Frontiers in
Physiology, 9, 143.
Myers, J., Prakash, M., Froelicher, V., Do, D., Partington, S., & Atwood,
J. E. (2002). Exercise capacity and mortality among men re-
ferred for exercise testing. New England Journal of Medicine,
346, 793– 801. https://doi.org/10.1056/NEJMo a011858
Niemela, K., Palatsi, I., Linnaluoto, M., & Takkunen, J. (1980).
Criteria for maximum oxygen- uptake in progressive bicy-
cle tests. European Journal of Applied Physiology, 44, 51– 59.
https://doi.org/10.1007/BF004 21763
Noakes, T. D. (2008). Maximal oxygen uptake as a parametric mea-
sure of cardiorespiratory capacity: Comment. Medicine and
Science in Sports and Exercise, 40, 585; author reply 586.
Nolan, P. B., Beaven, M. L., & Dalleck, L. (2014). Comparison of in-
tensities and rest periods for VO2max verification testing proce-
dures. International Journal of Sports Medicine, 35, 1024– 1029.
https://doi.org/10.1055/s- 0034- 1367065
Paterson, D. H., Cunningham, D. A., Koval, J. J., & St Croix, C.
M. (1999). Aerobic fitness in a population of independently
living men and women aged 55– 86 years. Medicine and
Science in Sports and Exercise, 31, 1813– 1820. https://doi.
org/10.1097/00005 768- 19991 2000- 00018
Paterson, D. H., Govindasamy, D., Vidmar, M., Cunningham,
D. A., & Koval, J. J. (2004). Longitudinal study of determi-
nants of dependence in an elderly population. Journal of
the American Geriatrics Society, 52, 1632– 1638. https://doi.
org/10.1111/j.1532- 5415.2004.52454.x
Paterson, D. H., & Warburton, D. E. (2010). Physical activity and
functional limitations in older adults: A systematic review
related to Canada's Physical Activity Guidelines. International
Journal of Behavioral Nutrition and Physical Activity, 7, 38.
Poole, D. C., & Jones, A. M. (2012). Oxygen uptake kinetics.
Comprehensive Physiology, 2, 933– 996.
Poole, D. C., & Jones, A. M. (2017). Measurement of the maximum
oxygen uptake (V) over dotO(2max): (V) over dotO(2peak)
is no longer acceptable. Journal of Applied Physiology, 122,
997– 1002.
Poole, D. C., Wilkerson, D. P., & Jones, A. M. (2008). Validity of
criteria for establishing maximal O2 uptake during ramp
exercise tests. European Journal of Applied Physiology,
102, 403– 410. https://doi.org/10.1007/s0042 1- 007- 0596- 3
Ross, R., Blair, S. N., Arena, R., Church, T. S., Despres, J. P.,
Franklin, B. A., Haskell, W. L., Kaminsky, L. A., Levine, B.
D., Lavie, C. J., Myers, J., Niebauer, J., Sallis, R., Sawada, S.
S., Sui, X., Wisloff, U., American Heart Association Physical
Activity Committee of the Council on L, Cardiometabolic
H, Council on Clinical C, Council on E, Prevention, Council
on C, Stroke N, Council on Functional G, Translational B, &
Stroke C. (2016). Importance of assessing cardiorespiratory
fitness in clinical practice: A case for fitness as a clinical
vital sign: A scientific statement from the American Heart
Association. Circulation, 134, e653– e699.
Rossiter, H. B., Kowalchuk, J. M., & Whipp, B. J. (2006). A test to
establish maximum O2 uptake despite no plateau in the O2 up-
take response to ramp incremental exercise. Journal of Applied
Physiology, 100, 764– 770.
Sawyer, B. J., Tucker, W. J., Bhammar, D. M., & Gaesser, G. A. (2015).
Using a verification test for determination of (V)over dotO(2)
max in sedentary adults with obesity. Journal of Strength and
Conditioning Research, 29, 3432– 3438.
Scharhag- Rosenberger, F., Carlsohn, A., Cassel, M., Mayer, F., &
Scharhag, J. (2011). How to test maximal oxygen uptake: A
study on timing and testing procedure of a supramaximal ver-
ification test. Applied Physiology Nutrition and Metabolism, 36,
153– 160.
Sedgeman, D., Dalleck, L., Clark, I. E., Jamnick, N., & Pettitt, R.
W. (2013). Analysis of square- wave bouts to verify VO2max.
International Journal of Sports Medicine, 34, 1058– 1062. https://
doi.org/10.1055/s- 0033- 1341436
Skinner, J. S., Wilmore, K. M., Jaskolska, A., Jaskolski, A.,
Daw, E. W., Rice, T., Gagnon, J., Leon, A. S., Wilmore,
J. H., Rao, D. C., & Bouchard, C. (1999). Reproducibility
of maximal exercise test data in the HERITAGE fam-
ily study. Medicine and Science in Sports and Exercise,
31, 1623– 1628. https://doi.org/10.1097/00005 768- 19991
1000- 00020
Taylor, H. L., Buskirk, E., & Henschel, A. (1955). Maximal oxygen in-
take as an objective measure of cardio- respiratory performance.
Journal of Applied Physiology, 8, 73– 80.
van Breda, E., Schoffelen, P. F. M., & Plasqui, G. (2017). Clinical
Vo2peak is "part of the deal". Journal of Applied Physiology,
122, 1370.
Wagner, J., Niemeyer, M., Infanger, D., Hinrichs, T., Guerra, C.,
Klenk, C., Konigstein, K., Cajochen, C., Schmidt- Trucksass, A.,
& Knaier, R. (2021). Verification- phase tests show low reliabil-
ity and add little value in determining VO2max in young trained
adults. PLoS One, 16, e0245306.
14 of 14 |
VILLANUEVA et al.
Wagner, J., Niemeyer, M., Infanger, D., Hinrichs, T., Streese, L.,
Hanssen, H., Myers, J., Schmidt- TrucksAss, A., & Knaier,
R. (2020). New data- based cutoffs for maximal exercise cri-
teria across the lifespan. Medicine and Science in Sports and
Exercise, 52, 1915– 1923. https://doi.org/10.1249/MSS.00000
00000 002344
Weatherwax, R. M., Richardson, T. B., Beltz, N. M., Nolan, P. B.,
& Dalleck, L. (2016). Verification testing to confirm VO2max
in altitude- residing, endurance- trained runners. International
Journal of Sports Medicine, 37, 525– 530. https://doi.
org/10.1055/s- 0035- 1569346
How to cite this article: Villanueva, I. R.,
Campbell, J. C., Medina, S. M., Jorgensen, T. M.,
Wilson, S. L., Angadi, S. S., Gaesser, G. A., &
Dickinson, J. M. (2021). Comparison of constant-
load exercise intensity for verification of maximal
oxygen uptake following a graded exercise test in
older adults. Physiological Reports, 9, e15037.
https://doi.org/10.14814/ phy2.15037
| Comparison of constant load exercise intensity for verification of maximal oxygen uptake following a graded exercise test in older adults. | [] | Villanueva, Ian R,Campbell, John C,Medina, Serena M,Jorgensen, Theresa M,Wilson, Shannon L,Angadi, Siddhartha S,Gaesser, Glenn A,Dickinson, Jared M | eng |
PMC9102981 | Citation: King, K.M.; McKay, T.;
Thrasher, B.J.; Wintergerst, K.A.
Maximal Oxygen Uptake, VO2 Max,
Testing Effect on Blood Glucose Level
in Adolescents with Type 1 Diabetes
Mellitus. Int. J. Environ. Res. Public
Health 2022, 19, 5543. https://
doi.org/10.3390/ijerph19095543
Academic Editors: Juan
Pablo Rey-López and Paul
B. Tchounwou
Received: 30 March 2022
Accepted: 30 April 2022
Published: 3 May 2022
Publisher’s Note: MDPI stays neutral
with regard to jurisdictional claims in
published maps and institutional affil-
iations.
Copyright:
© 2022 by the authors.
Licensee MDPI, Basel, Switzerland.
This article is an open access article
distributed
under
the
terms
and
conditions of the Creative Commons
Attribution (CC BY) license (https://
creativecommons.org/licenses/by/
4.0/).
International Journal of
Environmental Research
and Public Health
Article
Maximal Oxygen Uptake, VO2 Max, Testing Effect on Blood
Glucose Level in Adolescents with Type 1 Diabetes Mellitus
Kristi M. King 1,*
, Timothy McKay 2, Bradly J. Thrasher 2,3 and Kupper A. Wintergerst 2,3
1
Department of Health and Sport Sciences, University of Louisville, Louisville, KY 40292, USA
2
Norton Children’s Hospital, Louisville, KY 40202, USA; [email protected] (T.M.);
[email protected] (B.J.T.); [email protected] (K.A.W.)
3
Wendy Novak Diabetes Center, Pediatric Endocrinology, School of Medicine, University of Louisville,
Louisville, KY 40202, USA
*
Correspondence: [email protected]
Abstract: Assessing maximal oxygen uptake (VO2 max) is generally considered safe when performed
properly for most adolescents; however, for adolescents with type 1 diabetes mellitus (T1DM),
monitoring glucose levels before and after exercise is critical to maintaining euglycemic ranges.
Limited guidance exists for glucose level recommendations for the pediatric population; there-
fore, the purpose of this retrospective clinical chart review study was to determine the effects of
VO2 max testing on blood glucose levels for adolescents with T1DM. A total of 22 adolescents
(mean age = 15.6 ± 1.8 years; male = 13, 59.1%) with a diagnosis of T1DM participated in a Bruce
protocol for VO2 max from January 2019 through February 2020. A statistically significant reduction
in glucose levels between pretest (<30 min, mean = 191.1 mg/dL ± 61.2) and post-test VO2 max
(<5 min, mean = 166.7 mg/dL ± 57.9); t(21) = 2.3, p < 0.05) was detected. The results from this current
study can help guide health and fitness professionals in formulating glycemic management strategies
in preparatory activities prior to exercise testing and during exercise testing.
Keywords: maximal oxygen uptake; VO2 max; blood glucose; type 1 diabetes mellitus (T1DM);
adolescents; exercise testing; pediatric; clinical exercise
1. Introduction
One of the tenets of the sports medicine field is to advance and integrate scientific
research to provide educational and practical applications of exercise science and sports
medicine. For individuals engaging in physical activity at any level, whether it is recre-
ational physical activity or competitive sports, there is clear, scientifically based guidance
regarding exercise testing and prescription for health and fitness professionals to employ
with healthy individuals as well as those living with chronic illnesses [1]. One component
of health-related physical fitness is cardiorespiratory fitness (CRF), the body’s ability to
perform large-muscle, dynamic, moderate-to-vigorous-intensity exercise for prolonged
periods of time. Assessing the maximal oxygen uptake (VO2 max) the body is able to use
during exercise is an established exercise test for determining CRF and is more predictive
of long-term survival than is any traditional risk factor or other measured physiologic
parameter [2]. VO2 max testing provides a measurement of the relative amount of oxygen
consumption per an amount of work. For example, an improved VO2 may allow one to
run longer at the same speed or faster with the same relative effort [3].
The graded exercise test used to elicit VO2 max is aggressive in nature to achieve
a maximal response from the participant. Under stress conditions, the hypothalamus
controls many hormone secretions to adjust glucose metabolism and energy production.
Glucose secretion and uptake are under the control of nervous and hormonal factors
such as catecholamines, cortisol, glucagon, growth hormone, and insulin, and all have
an immediate impact [4]. Even though exercise testing is generally considered safe when
Int. J. Environ. Res. Public Health 2022, 19, 5543. https://doi.org/10.3390/ijerph19095543
https://www.mdpi.com/journal/ijerph
Int. J. Environ. Res. Public Health 2022, 19, 5543
2 of 7
performed properly for most individuals, maximal- or vigorous-intensity exercise testing
does pose some risk [5–8]. Specifically, for individuals with type 1 diabetes mellitus
(T1DM), the risk of hyperglycemia in the initial portion of exercise testing and the risk
for hypoglycemia following completion of testing both present themselves. Monitoring
glucose levels before and after physical activity is fundamental to maintaining glucose
levels in euglycemic ranges during and after exercise [9].
Unfortunately, understanding of safety parameters and the effect of CRF testing on
adolescent populations is limited and in need of research [10]. Glucose level recommen-
dations have yet to be established for adolescents diagnosed with T1DM who participate
in VO2 max exercise testing. Although a decrease in glucose may be expected throughout
and immediately after exercise testing, a minimal pretest glucose setpoint has not been
established to reduce the risk of hypoglycemia. Therefore, the purpose of this study was to
examine the impact of VO2 max testing on blood glucose levels for adolescents with T1DM.
2. Materials and Methods
2.1. Study Design and Setting
This cross-sectional, non-interventional, retrospective chart review study was con-
ducted at a nationally certified pediatric diabetes care and academic medical center located
in the Southeast region of the United States. At this center, pediatric endocrinologists,
registered nurses, registered dieticians, certified diabetes educators, and clinical exercise
physiologists treat pediatric patients diagnosed with T1DM up to 26 years of age. The
study was approved by the University Institutional Review board. Retrospective clinical
chart reviews were conducted of clinical pediatric sports medicine and physical activity
program patients with a diagnosis of T1DM who participated in a Bruce protocol for VO2
max from January 2019 through February 2020.
2.2. Participant Characteristics
The baseline characteristics of study participants are displayed in Table 1.
Table 1. Characteristics of the Participants, N = 22.
Characteristics
Mean ± SD or n, %
Range
(Minimum–Maximum)
Age, years
15.6 ± 1.8
13–20
Duration of T1DM diagnosis, years
7.1 ± 4.9
>1–16
Height, centimeters
170.9 ± 8.3
157.3–187.4
Weight, kilograms
67.1 ± 13.4
44.6–107.3
BMI percentile, n = 21
67th percentile ± 17.7
26–99
HbA1c level, n = 21
8.9 ± 1.8
6.1–14.9
Gender
Male, n = 13, 59.1%
Female, n = 9, 40.9%
-
Ethnicity
Non-Hispanic, n = 20, 90.1%
-
Race
White, n = 17, 77.3%
Black or African American, n = 3, 13.6%
Unknown or Not Reported, n = 2, 9.1%
-
Treatment Plan
CGM only, n = 2, 9.1%
Insulin pump only, n = 4, 18.2%
Insulin pump integrated with CGM, n = 9, 40.9%
MDI, n = 7, 31.8%
Note. Data are presented as mean ± standard deviation (SD) or number of participants (n), percent (%);
BMI, body mass index percentile; HbA1c, hemoglobin A1c; CGM, continuous glucose monitor; MDI, multi-
ple daily injections.
Int. J. Environ. Res. Public Health 2022, 19, 5543
3 of 7
2.3. Measures
Socio-demographic, anthropometric, diabetes monitoring and treatment plans, and
hemoglobin A1c (HbA1c) levels data were retrieved from patients’ medical records, main-
tained in the clinical database, as part of standard practice at each patient’s appointment in
the diabetes clinic. The socio-demographic and anthropometric characteristics utilized were
participant’s age, date and duration of T1DM diagnosis, ethnicity, race, gender, insurance
type, and body mass index (BMI) percentile. Diabetes monitoring was assessed (includ-
ing whether the participant used a continuous glucose monitor (CGM)), and the type of
treatment plan was recorded. The hemoglobin A1c (HbA1c) level was obtained from that
day’s clinical lab measures at check-in. HbA1c, the most prevalent and accessible measure
in determining glucose control, was used as an indicator of the average blood glucose
levels over the past 3 months. Adolescents managing T1DM should strive for HbA1c levels
less than 7%, as an elevated HbA1c level is known to increase the risk for diabetes-related
complications [11]. Blood glucose levels and VO2 max data were obtained from sports
medicine records collected by clinical exercise physiologists in the sports medicine clinic.
2.4. Preparatory Activities Prior to Exercise Testing
Upon registration for an exercise testing appointment, participants were instructed to
not eat a heavy meal two hours prior to testing, to maintain their insulin regimen as they
would on a regular day, and to dress in exercise attire (e.g., shorts, t-shirt, athletic shoes).
Upon arrival to the sports medicine clinic, participants’ blood glucose value was screened
by a clinical exercise physiologist. If the blood glucose was >250 mg/dL the clinical exercise
physiologist obtained urinary ketones with the next void. If urinary ketones were moderate
or large, the participant was excluded from participating in VO2 max testing at that time. If
the blood glucose was >300 g/dL and the participant had no or small ketones, the clinical
exercise physiologist instructed the participant to give a conservative insulin correction of
50% their calculated correction dose.
Approximately 30 min prior to VO2 max testing, a pretest blood glucose sample was
taken using a point-of-care glucometer and lancet device. Upon determination that blood
glucose levels were in the safe range for physical activity, the clinical exercise physiologist
conducted the VO2 max test. Upon completion of the VO2 max test, a post-test glucose
level check was conducted within 5 min using the same glucometer for all participants.
2.5. Exercise Testing Procedures
A Bruce protocol [12] for VO2 max tests, a valid and reliable measure for assessing
cardiorespiratory fitness, was performed on a Woodway ELG treadmill while the partic-
ipants were connected to the Parvo Medics metabolic gas exchange analyzer by way of
respiratory mask. Participants walked on a treadmill in 3 min bouts, starting at 1.7 mph
(45.6 m·min−1) and 10% grade. At each stage, the speed was increased by either 0.8
or 0.9 mph (21.4 or 24.1 m·min−1) and the grade was increased by 2%. This test lasted
approximately 10–20 min.
If following the graded exercise test the participant’s blood glucose was found to be
<70 mg/dL on the glucometer, the participant was treated for hypoglycemia with 15 g
of rapid-acting carbohydrate. Blood glucose was then rechecked at 15 min. This process
was repeated until their blood glucose was >70 mg/dL. Blood glucose and VO2 max data
stored within REDCap on a secure server in the sports medicine program data were linked
to the clinical database by the researchers utilizing the patients’ medical record numbers.
All clinical data were retrieved from that same-day appointment for each participant. Once
data were collected and merged, the full dataset was de-identified for analysis.
2.6. Data Analysis
All statistical analyses were conducted using IBM SPSS 27.0 (IBM Corp., Armonk,
NY, USA). Descriptive statistics and frequencies for socio-demographic, anthropometric,
diabetes monitoring and treatment plans, HbA1c levels, and pre- (<30 min) and post-
Int. J. Environ. Res. Public Health 2022, 19, 5543
4 of 7
test (<5 min) blood glucose levels were calculated. Shapiro–Wilk’s test (p < 0.05) [13,14],
histograms, Norman Q–Q plots, and box plots were employed to test the normality of
the distribution of the data. A paired-samples t-test was employed to detect differences
in blood glucose levels from pretest to post-test. p-Values of <0.05 were considered as
statistically significant.
3. Results
Retrospective VO2 max data were analyzed from a total of 22 adolescents (N = 22;
mean age = 15.6 ± 1.8 years; male = 13, 59.1%) (see Table 1). Most of the participants
identified as non-Hispanic (n = 20, 90.9%), and over three-quarters identified as White
(n = 17, 77.3%). Continuous glucose monitors were worn by 13 of the 22 participants
(59.1%). Their average HbA1c prior to participating in the VO2 max test was 8.9% ± 1.8.
The average BMI, based on age and sex, was in the 67th percentile ± 17.7. The average VO2
max peak was 43.4 mL/kg/min ± 6.4 (See Table 2).
Table 2. VO2 max testing measurements, N = 22.
Characteristics
Mean ± SD
Range
(Minimum–Maximum)
VO2 max, mL/kg/min, n = 21
43.4 ± 6.4
29.3–50.5
Peak HR bpm, n = 19
192.3 ± 21.3
119–212
Glucose, mg/dL pretest, n = 22
191.1 ± 61.1
96–296
Glucose, mg/dL post-test, n = 22
166.7 ± 57.9
83–297
Note. Data are presented as mean ± standard deviation (SD); Peak HR bpm, heart rate beats per minute.
Pre- and post-test blood glucose measurements were obtained from 22 participants.
The results of a Shapiro–Wilk’s test indicated that the pre- and post-glucose data were
normally distributed, and a visual inspection of their histograms, Norman Q–Q plots,
and box plots showed that the glucose scores were normally distributed at pretest with
a skewness of 0.107 (SE = 0.49) and a kurtosis of −0.868 (SE = 0.95) and at post-test
with a skewness of 0.657 (SE = 0.49) and a kurtosis of −0.015 (SE = 0.95). A paired-
samples t-test was employed to detect a statistically significant reduction in glucose levels
between pretest (<30 min, mean = 191.1 mg/dL ± 61.2) and post-test VO2 max (<5 min,
mean = 166.7 mg/dL ± 57.9); t(21) = 2.3, p < 0.05).
4. Discussion
It is well established that significant changes in blood glucose concentration during
physical activity can lead to hypoglycemia or hyperglycemia and, if not prevented or
treated quickly and properly, can lead to a medical emergency [1,9,15–27]. This current
study sought to examine if there was a significant drop in blood glucose levels after VO2
max testing, yet it is unique in that it specialized in a pediatric population of adolescents.
Results from a recent retrospective study with adults with T1DM (mean age = 32 years,
SD ± 13; range 18–65 years) who participated in VO2 max exercise testing using a cycle
ergometer did not demonstrate statistically significant glucose levels from pretest to post-
test [28], which aligned with similar results from other studies [29,30]. The conflicting
results from the present study may be attributed to differences in the age of participants
(and in body composition and hormones) and possibly the modality used during testing.
Given that individuals with T1DM are recommended to participate in daily moderate-
to-vigorous-intensity physical activity [31], and general guidance for glucose targets as
well as nutritional and insulin dose adjustments to protect against exercise-related glucose
excursions are available [9,20,21,26,27], fear of activity-related hypoglycemia has been
regularly cited as a barrier to physical activity [32,33]. Health care providers wishing
to prescribe even modest increases in intensity levels of daily activity, such as walking
and/or jogging, or sport participation that may include moderate-to-vigorous-intensity
Int. J. Environ. Res. Public Health 2022, 19, 5543
5 of 7
activity to their patients with T1DM may consider VO2 max testing as a first step in
establishing safety precautions and working toward the adoption and maintenance of
an active lifestyle. For example, participation in sports is touted as a beneficial means
for adolescents to accumulate physical activity [34,35]. However, caution must be taken
if prescribing sport only without the engagement in additional physical activity. This is
because many adolescents who participate in a single sport often do not meet sufficient
physical activity recommendations. A recent study involving 153 children and adolescents
diagnosed with T1DM demonstrated this fact [36]. Although almost two-thirds of the
participants reported playing one or more sports in the previous year, they were only
physically active for at least one hour or more on an average 3.5 days per week, with less
than 8% of the children and adolescents in the study meeting the recommended duration
of one hour and frequency of seven days per week of physical activity.
The results from this current study may help guide health and fitness professionals in
formulating glycemic management strategies in preparatory activities prior to exercise test-
ing and during exercise testing. A pre-exercise glucose level of 90–250 mg/dL is suggested
in order to prevent symptoms of hypoglycemia and to minimize hyperglycemia [9,11,26].
Considering that the adolescents in this current study experienced a 24.4 mg/dL drop in
glucose levels from pretest to post-test, the implications of these results have clinical and
practical importance. These results can and should be used to help inform patients and
practitioners in clinical care decision making and the formulation of glycemic management
strategies. Similar research findings suggest that patients and clinical care teams under-
stand the glycemic changes that occur during progressive exercise so that nutritional and
medicinal preparatory routines are safely established [28]. To ensure safe exercise perfor-
mance ahead of exercise testing, practitioners, physiologists, and patients should be aware
of the interindividual responses to VO2 max testing and treat each case accordingly. With
the assistance of a clinical exercise physiologist, physicians can incorporate individualized
recommendations for increasing physical activity and/or exercise prescriptions into their
clinical practices [37]. Physicians and medical care teams can prescribe physical activity
and sport participation when designing treatment plans and refer their patients to qualified
health and fitness professionals such as athletic trainers, strength and conditioning coaches,
and physical educators who coach or train adolescent athletes diagnosed with T1DM.
These protective measures that are grounded in scientific evidence [9,11,38] suggest that
adolescent patients diagnosed with T1DM can complete maximal exercise testing without
fear of inducing hypoglycemia if the necessary safety precautions as described in this study
are taken.
Limitations and Future Research
Various limitations have been identified in this study. Although all participants
were instructed to avoid eating greater than 60 g of carbohydrates prior to exercising,
unless hypoglycemic, the study was not controlled for nutrition. Future studies should
analyze dietary practices leading into exercise testing. In addition, participants were also
instructed to administer insulin per their routine standard of care to create a “real-world”
testing situation for this study. Future studies could benefit from more restrictive insulin
use parameters.
The data in this study were derived from a retrospective chart review of clinical pa-
tients who participated in a pre–post VO2 max test at a newly established (2018) clinic
serving only pediatric patients diagnosed with T1DM up to 26 years of age from January
2019 until February 2020. In March 2020, non-emergency clinical operations were sus-
pended due to COVID-19 safety precautions protocol, and sports medicine programming
and study endeavors resumed in August 2021, which further limited our total sample
size. At the time of the study, there were no matched control group data available. Next,
although the clinic houses the only pediatric endocrinology sports medicine program in
the state, the homogeneity of the participants in race and ethnicity does not make the
findings generalizable to adolescents in other areas. Future research will benefit from a
Int. J. Environ. Res. Public Health 2022, 19, 5543
6 of 7
more extensive and longitudinal review of the pre- and post-VO2 max testing windows to
further understand what variables influence blood glucose variability
5. Conclusions
The results from this retrospective VO2 max testing study on blood glucose levels in
adolescents with T1DM can add to the scientific literature for sports medicine programs
that provide clinical care to individuals and their families through patient-centered and
community education as well as clinical research. Regardless of sport or physical activity,
care is focused on improving the health, safety, and athletic performance of every child
and young adult with T1DM. Knowing that a significant drop in glucose levels during
VO2 max testing may occur with their adolescent patients, health and fitness professionals
can discuss and implement preventive glycemic management strategies prior to exercise
testing and during exercise testing.
Author Contributions: Conceptualization, K.M.K., T.M., B.J.T. and K.A.W.; methodology, K.M.K.
and T.M.; formal analysis, K.M.K.; investigation, T.M.; data curation, K.M.K.; writing—original draft
preparation, K.M.K.; writing—review and editing, K.M.K., T.M., B.J.T. and K.A.W.; visualization,
K.M.K. All authors have read and agreed to the published version of the manuscript.
Funding: This study was made possible by generous support from the Christensen Family, the
Norton Children’s Hospital Foundation, and the University of Louisville Foundation.
Institutional Review Board Statement: This study was approved by the University’s Institutional
Review Board on 6-18-2020 (Approval # 20.0506).
Informed Consent Statement: This retrospective research study was considered “exempt” by the
Institutional Review Board of University of Louisville.
Conflicts of Interest: The authors declare no conflict of interest. The funders had no role in the design
of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, or
in the decision to publish the results.
References
1.
American College of Sports Medicine (ACSM); Liguori, G.; Feito, Y.; Fontaine, C.; Roy, B. ACSM’s Guidelines for Exercise Testing
and Prescription, 11th ed.; Wolters Kluwer: Philadelphia, PA, USA, 2022.
2.
Leeper, N.J.; Myers, J.; Zhou, M.; Nead, K.T.; Syed, A.; Kojima, Y.; Caceres, R.D.; Cooke, J. Exercise capacity is the strongest
predictor of mortality in patients with peripheral arterial disease. J. Vasc. Surg. 2013, 57, 728–733. [CrossRef] [PubMed]
3.
Johnston, R.E.; Quinn, T.J.; Kertzer, R.; Vroman, N.B. Strength training female distance runners: Impact on running economy.
J. Strength Cond. Res. 1997, 11, 224–229. [CrossRef]
4.
Ciccarelli, L.; Connell, S.R.; Enderle, M.; Mills, D.J.; Vonck, J.; Grininger, M. Structure and conformational variability of the
mycobacterium tuberculosis fatty acid synthase multienzyme complex. Structure 2013, 21, 1251–1257. [CrossRef]
5.
Gibbons, L.; Blair, S.N.; Kohl, H.W.; Cooper, K. The safety of maximal exercise testing. Circulation 1989, 80, 846–852. [CrossRef]
6.
Knight, J.A.; Laubach, C.A.; Butcher, R.J., Jr.; Menapace, F.J. Supervision of clinical exercise testing by exercise physiologists. Am.
J. Cardiol. 1995, 75, 390–391. [CrossRef]
7.
McHenry, P.L. Risks of graded exercise testing. Am. J. Cardiol. 1977, 39, 935–937. [CrossRef]
8.
Stuart, R.J., Jr.; Ellestad, M.H. National survey of exercise stress testing facilities. Chest 1980, 77, 94–97. [CrossRef]
9.
Riddell, M.C.; Gallen, I.W.; Smart, C.E.; Taplin, C.E.; Adolfsson, P.; Lumb, A.N.; Kowalski, A.; Rabasa-Lhoret, R.; McCrimmon,
R.J.; Hume, C.; et al. Exercise management in type 1 diabetes: A consensus statement. Lancet Diabetes Endocrinol. 2017, 5, 377–390.
[CrossRef]
10.
Patterson, C.C.; Karuranga, S.; Salpea, P.; Saeedi, P.; Dahlquist, G.; Soltesz, G.; Ogle, G.D. Worldwide estimates of incidence,
prevalence and mortality of type 1 diabetes in children and adolescents: Results from the International Diabetes Federation
Diabetes Atlas, 9th edition. Diabetes Res. Clin. Pract. 2019, 157, 107842. [CrossRef]
11.
American Diabetes Association. Children and adolescents: Standards of medical care in diabetes—2021. Diabetes Care 2021,
44 (Suppl. 1), S180–S199. [CrossRef]
12.
Bruce, R.A.; Kusumi, F.; Hosmer, D. Maximal oxygen intake and nomographic assessment of functional aerobic impairment in
cardiovascular disease. Am. Heart J. 1973, 85, 546–562. [CrossRef]
13.
Shapiro, S.S.; Wilk, M.B. An Analysis of Variance Test for Normality (Complete Samples). Biometrika 1965, 52, 591–611. [CrossRef]
14.
Razali, N.M.; Wah, Y.B. Power comparisons of Shapiro-Wilk, Kolmogorov-Smirnov, Lilliefors and Anderson-Darling Tests. J. Stat.
Model. Anal. 2011, 2, 21–33.
Int. J. Environ. Res. Public Health 2022, 19, 5543
7 of 7
15.
McCarthy, O.; Deere, R.; Churm, R.; Dunseath, G.J.; Jones, C.; Eckstein, M.L.; Williams, D.M.; Hayes, J.; Pitt, J.; Bain, S.C.;
et al. Extent and prevalence of post-exercise and nocturnal hypoglycemia following peri-exercise bolus insulin adjustments in
individuals with type 1 diabetes. Nutr. Metab. Cardiovasc. Dis. 2021, 31, 227–236. [CrossRef] [PubMed]
16.
Notkin, G.T.; Kristensen, P.L.; Pedersen-Bjergaard, U.; Jensen, A.K.; Molsted, S. Reproducibility of glucose fluctuations induced
by moderate intensity cycling exercise in persons with type 1 diabetes. J. Diabetes Res. 2021, 2021, 6640600. [CrossRef]
17.
Aljawarneh, Y.M.; Wardell, D.W.; Wood, G.L.; Rozmus, C.L. A systematic review of physical activity and exercise on physiological
and biochemical outcomes in children and adolescents with type 1 diabetes. J. Nurs. Scholarsh. 2019, 51, 337–345. [CrossRef]
18.
Riddell, M.C.; Zaharieva, D.P.; Tansey, M.; Tsalikian, E.; Admon, G.; Li, Z.; Kollman, C.; Beck, R.W. Individual glucose responses
to prolonged moderate intensity aerobic exercise in adolescents with type 1 diabetes: The higher they start, the harder they fall.
Pediatr. Diabetes 2019, 20, 99–106. [CrossRef]
19.
Aronson, R.; Brown, R.E.; Li, A.; Riddell, M.C. Optimal Insulin Correction Factor in Post-High-Intensity Exercise Hyperglycemia
in Adults With Type 1 Diabetes: The FIT Study. Diabetes Care 2019, 42, 10–16. [CrossRef]
20.
Jaggers, J.R.; King, K.M.; Watson, S.E.; Wintergerst, K.A. Predicting nocturnal hypoglycemia with measures of physical activity
intensity in adolescent athletes with type 1 diabetes. Diabetes Technol. Ther. 2019, 21, 406–408. [CrossRef]
21.
Jaggers, J.R.; Hynes, K.C.; Wintergerst, K.A. Exercise and Sport Participation for Individuals with Type 1 Diabetes. ACSM’s Health
Fit. J. 2016, 20, 40–44. [CrossRef]
22.
Yardley, J.E.; Hay, J.; Abou-Setta, A.M.; Marks, S.D.; McGavock, J. A systematic review and meta-analysis of exercise interventions
in adults with type 1 diabetes. Diabetes Res. Clin. Pract. 2014, 106, 393–400. [CrossRef]
23.
Colberg, S.R.; Kannane, J.; Diawara, N. Physical Activity, Dietary Patterns, and Glycemic Management in Active Individuals with
Type 1 Diabetes: An Online Survey. Int. J. Environ. Res. Public Health 2021, 18, 9332. [CrossRef] [PubMed]
24.
ACSM. Exercise Is Medicine. Exercise Is Medicine Web Site. Available online: http://www.exerciseismedicine.org/ (accessed on
25 September 2018).
25.
MacMillan, F.; Kirk, A.; Mutrie, N.; Matthews, L.; Robertson, K.; Saunders, D.H. A systematic review of physical activity and
sedentary behavior intervention studies in youth with type 1 diabetes: Study characteristics, intervention design, and efficacy.
Pediatr. Diabetes 2014, 15, 175–189. [CrossRef] [PubMed]
26.
Colberg, S.R.; Sigal, R.J.; Yardley, J.E.; Riddell, M.C.; Dunstan, D.W.; Dempsey, P.C.; Horton, E.S.; Castorino, K.; Tate, D.F. Physical
Activity/Exercise and Diabetes: A Position Statement of the American Diabetes Association. Diabetes Care 2016, 39, 2065–2079.
[CrossRef]
27.
Colberg, S.R. The Athlete’s Guide to Diabetes; Human Kinetics: Champaign, IL, USA, 2020.
28.
McCarthy, O.; Pitt, J.; Wellman, B.; Eckstein, M.L.; Moser, O.; Bain, S.C.; Bracken, R.M. Blood Glucose Responses during
Cardiopulmonary Incremental Exercise Testing in Type 1 Diabetes: A Pooled Analysis. Med. Sci. Sports Exerc. 2021, 53, 1142–1150.
[CrossRef] [PubMed]
29.
Turinese, I.; Marinelli, P.; Bonini, M.; Statuto, G.; Filardi, T.; Paris, A.; Lenzi, A.; Morano, S.; Palange, P.; Rossetti, M. Metabolic
and cardiovascular response to exercise in patients with type 1 diabetes. J. Endocrinol. Investig. 2017, 40, 999–1005. [CrossRef]
[PubMed]
30.
Peltonen, J.E.; Koponen, A.S.; Pullinen, K.; Hägglund, H.; Aho, J.M.; Kyröläinen, H.; Tikkanen, H.O. Alveolar gas exchange and
tissue deoxygenation during exercise in type 1 diabetes patients and healthy controls. Respir. Physiol. Neurobiol. 2012, 181, 267–276.
[CrossRef]
31.
United States Department of Health and Human Services (USDHHS). Physical Activity Guidelines for Americans, 2nd ed.; USDHHS:
Washington, DC, USA, 2018.
32.
Martyn-Nemeth, P.; Quinn, L.; Penckofer, S.; Park, C.; Hofer, V.; Burke, L. Fear of hypoglycemia: Influence on glycemic variability
and self-management behavior in young adults with type 1 diabetes. J. Diabetes Its Complicat. 2017, 31, 735–741. [CrossRef]
33.
Berkovic, M.C.; Bilic-Curcic, I.; Sabolic, L.L.G.; Mrzljak, A.; Cigrovski, V. Fear of hypoglycemia, a game changer during physical
activity in type 1 diabetes mellitus patients. World J. Diabetes 2021, 12, 569–577. [CrossRef]
34.
National Physical Activity Plan Alliance (NPAPA). The 2018 United States Report Card on Physical Activity for Children and Youth;
National Physical Activity Plan Alliance: Washington, DC, USA, 2018.
35.
Mandic, S.; Bengoechea, E.G.; Stevens, E.; de la Barra, S.L.; Skidmore, P. Getting kids active by participating in sport and doing it
more often: Focusing on what matters. Int. J. Behav. Nutr. Phys. Act. 2012, 9, 86. [CrossRef]
36.
King, K.; Jaggers, J.; Della, L.; McKay, T.; Watson, S.; Kozerski, A.; Hartson, K.; Wintergerst, K. Association between physical
activity and sport participation on hemoglobin A1c among children and adolescents with type 1 diabetes. Int. J. Environ. Res.
Public Health 2021, 18, 7490. [CrossRef] [PubMed]
37.
King, K.M.; Jaggers, J.R.; Wintergerst, K. Strategies for partnering with health care settings to increase physical activity promotion.
ACSM’s Health Fit. J. 2019, 23, 40–43. [CrossRef]
38.
Jaggers, J.R.; McKay, T.; King, K.M.; Thrasher, B.J.; Wintergerst, K.A. Integration of consumer-based activity monitors into clinical
practice for children with type 1 diabetes: A feasibility study. Int. J. Environ. Res. Public Health 2021, 18, 10611. [CrossRef]
[PubMed]
| Maximal Oxygen Uptake, VO<sub>2</sub> Max, Testing Effect on Blood Glucose Level in Adolescents with Type 1 Diabetes Mellitus. | 05-03-2022 | King, Kristi M,McKay, Timothy,Thrasher, Bradly J,Wintergerst, Kupper A | eng |
PMC10649254 | Citation: van Rassel, C.R.; Ajayi,
O.O.; Sales, K.M.; Griffiths, J.K.;
Fletcher, J.R.; Edwards, W.B.;
MacInnis, M.J. Is Running Power a
Useful Metric? Quantifying Training
Intensity and Aerobic Fitness Using
Stryd Running Power Near the
Maximal Lactate Steady State. Sensors
2023, 23, 8729. https://doi.org/
10.3390/s23218729
Academic Editors: Manuel E.
Hernandez, Yih-Kuen Jan, Chi-Wen
Lung and Ben-Yi Liau
Received: 27 September 2023
Revised: 20 October 2023
Accepted: 23 October 2023
Published: 26 October 2023
Copyright:
© 2023 by the authors.
Licensee MDPI, Basel, Switzerland.
This article is an open access article
distributed
under
the
terms
and
conditions of the Creative Commons
Attribution (CC BY) license (https://
creativecommons.org/licenses/by/
4.0/).
sensors
Article
Is Running Power a Useful Metric? Quantifying Training
Intensity and Aerobic Fitness Using Stryd Running Power Near
the Maximal Lactate Steady State
Cody R. van Rassel 1
, Oluwatimilehin O. Ajayi 1, Kate M. Sales 1, James K. Griffiths 1, Jared R. Fletcher 2
,
W. Brent Edwards 1 and Martin J. MacInnis 1,*
1
Faculty of Kinesiology, University of Calgary, Calgary, AB T2N 1N4, Canada; [email protected] (C.R.v.R.)
2
Department of Health and Physical Education, Mount Royal University, Calgary, AB T3E 6K6, Canada
*
Correspondence: [email protected]
Abstract: We sought to determine the utility of Stryd, a commercially available inertial measure-
ment unit, to quantify running intensity and aerobic fitness. Fifteen (eight male, seven female)
runners (age = 30.2 [4.3] years;
·
VO2max = 54.5 [6.5] mL·kg−1·min−1) performed moderate- and
heavy-intensity step transitions, an incremental exercise test, and constant-speed running trials to
establish the maximal lactate steady state (MLSS). Stryd running power stability, sensitivity, and
reliability were evaluated near the MLSS. Stryd running power was also compared to running speed,
·
VO2, and metabolic power measures to estimate running mechanical efficiency (EFF) and to deter-
mine the efficacy of using Stryd to delineate exercise intensities, quantify aerobic fitness, and estimate
running economy (RE). Stryd running power was strongly associated with
·
VO2 (R2 = 0.84; p < 0.001)
and running speed at the MLSS (R2 = 0.91; p < 0.001). Stryd running power measures were strongly
correlated with RE at the MLSS when combined with metabolic data (R2 = 0.79; p < 0.001) but not in
isolation from the metabolic data (R2 = 0.08; p = 0.313). Measures of running EFF near the MLSS were
not different across intensities (~21%; p > 0.05). In conclusion, although Stryd could not quantify
RE in isolation, it provided a stable, sensitive, and reliable metric that can estimate aerobic fitness,
delineate exercise intensities, and approximate the metabolic requirements of running near the MLSS.
Keywords: wearable technology; running economy; critical intensity; human performance; inertial
measurement unit; treadmill
1. Introduction
A consensus regarding an approach to evaluate mechanical running power output
(PO) is lacking, resulting in a range of PO values for a given running speed, depending on
the method [1,2]. During level running, the working muscles transfer energy to produce
and absorb the forces needed to move body segments. As a result, there is no dissipative
load external to the body that can be measured to quantify mechanical PO. Instead, running
mechanical PO measurements may be derived from “external” or “internal” work per-
spectives by evaluating the centre of mass (CoM) or the body segments, respectively [1–3].
Such approaches require sophisticated laboratory assessments of joint kinetics and/or kine-
matics based on ground reaction force and motion-capture data. Several methodological
challenges also limit the utility of running mechanical PO to approximate the metabolic
work rate [1,3], and, in contrast to cycling, where there is a strong relationship between
mechanical and metabolic PO [4,5], many factors complicate the relationship between me-
chanical and metabolic PO when running [6–8]. Nevertheless, a wearable running device
that can quantify and monitor training intensity, analogous to a cycling power meter [9,10],
would be useful to guide training and maximize running performance.
Sensors 2023, 23, 8729. https://doi.org/10.3390/s23218729
https://www.mdpi.com/journal/sensors
Sensors 2023, 23, 8729
2 of 19
Several consumer technologies providing a running power metric have been devel-
oped [11,12]. These technologies derive a measurement of mechanical PO using estimates
of ground reaction forces, CoM velocity, and/or vertical displacement from global position-
ing system (GPS) and/or inertial measurement unit (IMU) sensor data [11–13]. Previously,
the Stryd running power device (a portable IMU), has provided the closest relationship
with
·
VO2 when compared to other available commercial devices [11]. Possibly by gen-
erating a running power metric based on estimates of horizontal velocity and vertical
displacement using acceleration data, it is purported that Stryd power can be used as a
proxy for metabolic PO, despite changes in external conditions such as air resistance or
gradient [13]. Thus, Stryd running power can theoretically quantify training intensity
in a manner analogous to cycling mechanical PO and could be superior to conventional
measurement approaches using running speed. Despite evidence of repeatability [11],
reliability [14,15], stability during prolonged running [16], and strong linear correlations
with running speed [17,18], limited research has investigated the Stryd running metric
at stable metabolic work rates relative to exercising thresholds. Thus, to determine the
utility of Stryd power to indicate relative exercise intensity and assess running fitness and
performance, the relationship between Stryd mechanical power and metabolic power needs
to be established using an exercise intensity domain training approach (i.e., evaluating
running power metrics during steady-state exercise relative to the gas exchange threshold
(GET) and maximal metabolic steady state (MMSS)).
Prior to determining whether Stryd running power can monitor training, like cycling
power output, in uncontrolled environments (e.g., variable inclines, wind speeds, and
surfaces), the primary purpose of the present study was to evaluate the Stryd power metric
in a controlled environment (i.e., in situ). Using an exercise intensity domain approach,
we assessed the stability, sensitivity, and reliability of Stryd at stable metabolic work rates
to (i) determine the efficacy of Stryd running power as a training intensity and running
performance metric, (ii) explore the relationship between running power and running
economy (RE), (iii) estimate mechanical efficiency during constant-speed treadmill running,
and (iv) contrast steady-state measurements with measurements derived from incremental
exercise. We hypothesized that Stryd running power would be repeatable across two visits,
stable during a 30-min run, and sensitive to running speeds near the maximal lactate steady
state (MLSS)—a proxy measure of the MMSS. In addition, we hypothesized that Stryd
power would be strongly associated with running speed,
·
VO2, and RE measurements,
thereby providing a tool to guide exercise training and assess running fitness.
2. Materials and Methods
2.1. Participants
Fifteen (8 male; 7 female) recreationally active or trained/developmental runners [19]
(mean [SD]; age = 30.2 [4.3] years; body mass = 68.8 [8.2] kg; height = 173.2 [8.4] cm;
·
VO2max, 54.5 [6.5] mL·kg−1·min−1) were recruited using convenience sampling. Partic-
ipants were included if they were healthy, uninjured, and between 18 and 45 years of
age, with recent 10-km performances of ≤50 min and ≤55 min for males and females,
respectively. Within the 3 months prior to testing, runners reported exercising an average
of 3.5 [1.4] days per week, running an average of 27.7 [17.1] km each week, and having
10-km best performance times of 44.6 [6.5] min. Written informed consent was provided by
the runners to participate in the experimental procedures, which were approved by the
University of Calgary Conjoint Health Research Ethics Board (REB20-0111) and conducted
in accordance with the declaration of Helsinki, except for pre-trial registration. Participants
had the option to cease participation at any time during the experimental procedures. Prior
to test administration, runners completed the physical activity readiness questionnaire
(PAR-Q+) to identify contraindications to exercise testing and to ensure that participants
were free of medical conditions and injuries that could interfere with metabolic and car-
Sensors 2023, 23, 8729
3 of 19
diorespiratory exercise responses. All runners provided their own lightweight running
shoes and wore the same shoes for all testing sessions.
2.2. Experimental Design
Runners visited the laboratory for five to six exercise testing sessions, with a minimum
of 48 h between visits. The exercise sessions included: (1) a “Step-Ramp-Step” (SRS) exercise
test to determine maximal exercising parameters [20]; (2) a series of 3–4 constant-speed
bouts to determine the MLSS; and (3) a repeated trial at the MLSS running speed. Runners
were asked to refrain from smoking, eating, or consuming caffeine within 2 h prior to their
testing sessions. Runners did not engage in strenuous exercise on the same day as the
testing sessions. A manuscript validating the SRS approach to identify the running speed
and Stryd running power associated with the MLSS has been published [20]; however,
despite the overlap in experimental procedures, the results presented herein are distinct.
2.3. Exercise Protocols
2.3.1. Step-Ramp-Step (SRS) Protocol
As described in detail in our previous study [20], runners performed an SRS exercise
protocol during their first testing visit to establish their maximal exercising values and
estimate the running speed associated with the MLSS. This SRS protocol was modified for
treadmill running from a cycle ergometer-based method [21]. Of relevance to the present
study, the SRS protocol involved a moderate-intensity step-transition (MOD; 6 min at
1.9 m·s−1, 6 min at 2.4 m·s−1, and 6 min at 1.9 m·s−1); an incremental treadmill running
test (an initial speed of 1.9 m·s−1, increasing by ~0.2 m·s−1 (i.e., 0.5 mph) per min, until
volitional exhaustion); and a heavy-intensity step transition (HVY; 4 min of treadmill
running at 1.9 m·s−1, followed by 12 min of treadmill running at a speed associated with
the heavy-intensity exercise domain). The incremental treadmill test immediately preceded
the MOD step, but the participants recovered passively for 30 min between the incremental
test and the HVY step. The SRS protocol facilitated the identification of the MLSS in
2–3 constant-speed trials [20].
2.3.2. Constant-Speed Treadmill Running—MLSS Determination
Following the initial SRS testing visit, runners completed the constant-speed exercise
sessions during 4 to 5 separate visits to the lab. These visits consisted of 5 min of treadmill
running at 1.9 m·s−1, followed by treadmill running at the predetermined testing speed.
During all constant-speed testing visits, participants were encouraged to run until volitional
exhaustion, up to a maximum duration of 45 min (excluding warm-up). Data collected
beyond 30 min were not used in this study. All runners performed their initial constant-
speed test at the running speed estimated to be the MLSS by the SRS protocol. Depending on
whether the difference between the 10- and 30-min blood lactate concentrations ([BLa]) was
≤1 mmol·L−1 or >1 mmol·L−1, the subsequent visit was performed at a treadmill speed
that was 5% faster or 5% slower, respectively. The MLSS for each runner was identified
as the highest treadmill speed whereby at least 30 min of exercise was performed and the
difference between the [Bla] at 10 and 30 min was ≤1 mmol·L−1 [22]. All participants
performed constant-speed treadmill running trials at the MLSS, 5% above the MLSS, 5%
below the MLSS, and once more at the MLSS. Data analysis was primarily based on data
collected at the 10- and 30-min (or at task failure if <30 min) time points.
2.4. Equipment and Measurements
2.4.1. Cardiorespiratory Measurements
All exercise sessions were performed on a treadmill (Desmo Pro Evo, Woodway USA
Inc., Waukesha, WI, USA) with an incline set to a 1% gradient [23]. Adjustments to treadmill
speed could be made in 0.1 mph increments (i.e., ~0.04 m·s−1); however, all running speed
data were reported in SI units (i.e., m·s−1). Ventilatory and gas exchange variables were
measured using the Quark CPET metabolic cart (COSMED, Rome, Italy), with a mixing
Sensors 2023, 23, 8729
4 of 19
chamber (COSMED), facemask (7450 Series V2, Hans-Rudolph, Shawnee, KS, USA), 2-way
non-rebreathing valve (Hans-Rudolph), and gas collection hose. The metabolic cart system
was calibrated using a 3 L syringe and gas mixture of known composition (5% CO2, 16% O2,
and N2 for the balance) prior to each testing visit. For the analysis, 10-s average ventilatory
and gas exchange data were used. Heart rates were recorded during all testing sessions
using a Polar H10 chest strap (Polar Electro Oy, Kempele, Finland).
The
·
VO2 associated with a disproportionate increase in the rate of carbon dioxide
production (
·
VCO2) and minute ventilation (
·
VE) relative to the increase in
·
VO2 was used
to identify the GET [24]. The
·
VO2 associated with a second disproportionate increase in
·
VE and a disproportionate increase in
·
VE/
·
VCO2 relative to the increase in
·
VO2 was used
to identify the respiratory compensation point (RCP) [24,25].
·
VO2max was identified as
the highest 30-s average
·
VO2 achieved during the incremental test. All incremental tests
were considered maximal, based on the observation of a
·
VO2 plateau (defined as a change
in
·
VO2 of less than 150 mL·min−1, despite an increased intensity) or any of the following
criteria: maximum HR within 10 bpm of the age-predicted maximal value, a respiratory
exchange ratio (RER) greater than 1.15, or [Bla] ≥ 8 mmol·L−1 upon test completion.
2.4.2. Blood Lactate Measurements
All [Bla] data were collected using capillary blood drawn from a pinprick of the finger
and analyzed for [Bla] using the Biosen C-Line (EKF Diagnostics, Cardiff, Wales; n = 7)
or Lactate Plus (Nova Biomedical, Waltham, MA, USA; n = 8) lactate analyzer. Runners
straddled the treadmill (~60–75 s) during [Bla] measurements at 10 and 30 min (or at task
failure if <30 min).
2.4.3. Perceptual Responses
After familiarization with the scale, the rating of perceived exertion (RPE) was mea-
sured every 5 min during constant-speed running, using the Borg RPE scale (6–20) [26].
2.4.4. Running Power—Stryd
Running power measurements were made using the Stryd Summit Running Pod
(Stryd, Boulder, CO, USA). The Stryd pod, which is a lightweight (8.0 g) and unobtrusive
(4.0 cm in length) wearable sensor (Model v.19, firmware v.2.1.16, software v.4), was affixed
to the runner’s left shoe, approximately equidistant between the runner’s malleoli and the
shoe’s toe. A unique Stryd user profile was created for each runner that included their
respective height and body mass, which was kept constant for all testing sessions. The
iPhone Stryd application (Apple Inc., Cupertino, CA, USA) was used to pair the Stryd
device and collect the Stryd running power data during the testing sessions. Running
power data were sampled at 1 Hz (see Figure 1).
2.5. Data Analysis
2.5.1. Cardiorespiratory, Running Speed, and Stryd Running Power Data
The average
·
VO2 and running power, measured between minutes 4 and 6 of the MOD
step and between minutes 10 and 12 of the HVY step, were calculated from the SRS test.
Maximal aerobic speed (MAS) and maximal aerobic power (MAP) were determined as the
running speed associated with the highest completed 1-min stage during the incremental
test and the average running power during that stage, respectively.
Cardiorespiratory and running power data used for analysis from the constant-speed
MLSS-determination running trials included the 10- and 30-min
·
VO2,
·
VCO2, RER,
·
VE, HR,
and running power measures for running trials 5% below, at, and 5% above MLSS. To
align with the timing of [BLa] measurement (i.e., a short pause in running), mean values
Sensors 2023, 23, 8729
5 of 19
for
·
VO2,
·
VCO2, RER,
·
VE, HR, and Stryd running power were calculated from the 2 min
of data collected prior to the 10-min and the 30-min (or at task failure if <30 min) time
points. Although the MLSS is thought to represent the highest intensity at which energy
provision is supplied exclusively via oxidative metabolism [27], data collected at 5% above
the MLSS were included in the analysis due to the similarly stable
·
VO2 measurements
between the 10- and 30-min values across the three intensities (i.e., differences between
10- and 30-min
·
VO2 measures were ~50 mL·min−1 at intensities of 5% below, at, and 5%
above the MLSS)—with similar findings previously being reported [28]—and to provide a
more comprehensive dataset for the analyses.
Sensors 2023, 23, x FOR PEER REVIEW
5 of 20
Figure 1. Example of the running power signal during constant-speed treadmill running at different
intensities for one participant. Data are shown for the moderate (MOD; 6 min) and heavy (HVY; 12
min) intensity steps, and during 30 min of running at 5% below the maximal lactate steady state
(MLSS), at the MLSS, 5% above the MLSS, and during a repeat trial at the MLSS, preceded by run-
ning power data recorded for 3–4 min at a running speed of 1.9 m·s−1. Running power data were not
collected during the first ~1–2 min of each exercise protocol (i.e., warm-up) to allow for synchroni-
zation with other measurements. Note that the repeat MLSS trial is obscured by the first MLSS trial.
2.5. Data Analysis
2.5.1. Cardiorespiratory, Running Speed, and Stryd Running Power Data
The average V̇ O2 and running power, measured between minutes 4 and 6 of the MOD
step and between minutes 10 and 12 of the HVY step, were calculated from the SRS test.
Maximal aerobic speed (MAS) and maximal aerobic power (MAP) were determined as the
running speed associated with the highest completed 1-min stage during the incremental
test and the average running power during that stage, respectively.
Cardiorespiratory and running power data used for analysis from the constant-speed
MLSS-determination running trials included the 10- and 30-min V̇ O2, V̇ CO2, RER, V̇ E, HR,
and running power measures for running trials 5% below, at, and 5% above MLSS. To
align with the timing of [BLa] measurement (i.e., a short pause in running), mean values
for V̇ O2, V̇ CO2, RER, V̇ E, HR, and Stryd running power were calculated from the 2 min of
data collected prior to the 10-min and the 30-min (or at task failure if <30 min) time points.
Although the MLSS is thought to represent the highest intensity at which energy provi-
sion is supplied exclusively via oxidative metabolism [27], data collected at 5% above the
MLSS were included in the analysis due to the similarly stable V̇ O2 measurements be-
tween the 10- and 30-min values across the three intensities (i.e., differences between 10-
and 30-min V̇ O2 measures were ~50 mL·min−1 at intensities of 5% below, at, and 5% above
the MLSS)—with similar findings previously being reported [28]—and to provide a more
comprehensive dataset for the analyses.
2.5.2. Incremental and Constant-Speed V̇ O2–Power and Speed–Power Gains
A least-squares linear regression was performed to calculate the V̇ O2–power gain
(i.e., the slope of the regression equation) for each participant during both incremental
and constant-speed exercise trials, measured separately. This method allowed for the cal-
l
i
f
V̇ O
i
d
d
d d
i
i
i
hi h
i
ld
-5.0
0.0
5.0
10.0
15.0
20.0
25.0
30.0
0
120
160
200
240
280
320
Time (min)
Stryd Power (W)
MOD
HVY
5% Below MLSS
At MLSS
5% Above MLSS
Repeat at MLSS
Figure 1. Example of the running power signal during constant-speed treadmill running at different
intensities for one participant. Data are shown for the moderate (MOD; 6 min) and heavy (HVY;
12 min) intensity steps, and during 30 min of running at 5% below the maximal lactate steady
state (MLSS), at the MLSS, 5% above the MLSS, and during a repeat trial at the MLSS, preceded by
running power data recorded for 3–4 min at a running speed of 1.9 m·s−1. Running power data
were not collected during the first ~1–2 min of each exercise protocol (i.e., warm-up) to allow for
synchronization with other measurements. Note that the repeat MLSS trial is obscured by the first
MLSS trial.
2.5.2. Incremental and Constant-Speed
·
VO2–Power and Speed–Power Gains
A least-squares linear regression was performed to calculate the
·
VO2–power gain
(i.e., the slope of the regression equation) for each participant during both incremental
and constant-speed exercise trials, measured separately. This method allowed for the
calculation of a
·
VO2–power gain mean and standard deviation in which comparisons could
be made between the incremental and constant-speed running tests and to contrast the
measurements with cycling data [29]. The “incremental
·
VO2–power gain” for each runner
was calculated as the slope ((mL·min−1)·W−1) of a least-squares linear regression line
through the incremental exercise
·
VO2–power response, from the onset of a systemic rise
in
·
VO2 until test termination or the onset of a plateau, if detected. The “constant-speed
·
VO2–power gain” for each runner was calculated as the slope ((mL·min−1)·W−1) of a least-
squares linear regression line for the steady-state
·
VO2 and power data from five constant-
speed intensities: the MOD and HVY steps from the SRS-protocol and the constant-speed
exercise trials 5% below, at, and 5% above the MLSS. Replacing
·
VO2 with running speed,
these same procedures were used to calculate the “incremental speed–power gain” and the
“constant-speed speed–power gain” for each participant.
Sensors 2023, 23, 8729
6 of 19
2.5.3. Metabolic Power, Mechanical Power, and Mechanical Efficiency
Running metabolic power, mechanical power, and mechanical efficiency measure-
ments were calculated during constant-speed running trials at MOD, HVY, 5% below, at,
and 5% above MLSS. Metabolic power was calculated as a gross energy cost per unit of
body mass and distance travelled (kJ·kg−1·km−1) using
·
VO2 and RER [30]. This calcula-
tion of metabolic power was used to represent the energy cost of running (i.e., RE) at each
respective intensity, providing a reference measure of RE in which all subsequent compar-
isons were made. Metabolic power (i.e., StrydMET) was also calculated by expressing the
energy cost—using
·
VO2 and RER [30]—per unit of absolute Stryd power ((J·s−1)·W−1)
and per unit of relative Stryd power ((kJ·s−1)·(W·kg−1)−1). StrydMET was calculated in
isolation from running speed to provide a metric that characterized the metabolic power
requirements per unit of Stryd running power. The units used to describe StrydMET were
not simplified in order to distinguish among related terms and to provide units that clearly
described the energy cost of running per unit of absolute and relative Stryd running power.
Mechanical power (i.e., StrydMECH) was calculated in isolation from the
·
VO2 and RER by
expressing Stryd running power (W) in units of J·s−1 and by converting mechanical power
to an absolute energy cost per unit of the distance travelled (kJ·km−1) and a relative energy
cost per unit of distance (kJ·kg−1·km−1). Mechanical efficiency (EFF) was calculated as the
ratio between StrydMECH (kJ·kg−1·km−1) and metabolic power (kJ·kg−1·km−1), expressed
as a percentage.
2.6. Statistical Analysis
2.6.1. General
Statistical analyses were performed using the Statistical Package for the Social Sciences
(SPSS, version 26, IBM, Armonk, NY, USA). Linear mixed-effects models were performed us-
ing the nlme package (version 3.1-157) in RStudio (version 4.2.0) (R Core Team (2018)). Data
visualization was performed using Prism (version 9.5.1 for macOS; GraphPad Software,
San Diego, CA, USA). Data are presented as mean [standard deviation (SD)]. Statistical sig-
nificance was set at an α level of <0.05. Where appropriate, Bonferroni post hoc tests were
used. Test-retest reliability was measured using two-way mixed effects, absolute agree-
ment, and single-rater intraclass correlation models wherein reliability was interpreted
as poor (ICC < 0.5), moderate (0.5 ≤ ICC < 0.75), good (0.75 ≤ ICC < 0.9), or excellent
(ICC ≥ 0.9) [31].
2.6.2. Stability, Sensitivity, and Reliability
Multiple two-way repeated-measure ANOVAs were used to assess stability (the main
effect of duration) and sensitivity (the main effect of intensity) of Stryd running power and
the physiological and perceptual responses (i.e.,
·
VO2,
·
VCO2, RER,
·
VE, HR, [BLa], and RPE)
at 10 min and 30 min (or task failure, if <30 min) during constant-speed treadmill running
at 5% below, at, and 5% above the MLSS. For the same variables, the two MLSS trials were
compared using paired Student’s t tests, intraclass correlations, and Bland–Altman analyses
(with 95% limits of agreement) to assess reliability at the 30-min timepoint. Stryd running
power stability was further assessed by evaluating the linear association and agreement
between the 10- and 30-min running power data for the first MLSS trial using a Pearson’s
correlation coefficient and Bland–Altman analysis, respectively.
2.6.3. Stryd Running Power—Association with
·
VO2 and Running Speed
Paired Student’s t tests were used to compare the mean
·
VO2–power gains and
·
VO2–speed gains between incremental and constant-speed exercise trials. To determine
the association between Stryd running power and training intensity, linear mixed-effects
models were used to assess the within-individual and between-individual association
Sensors 2023, 23, 8729
7 of 19
between running power and
·
VO2 measurements, and between running power and running
speed during the MOD, HVY, MLSS −5%, MLSS, and MLSS +5% running trials. Models
included fixed-effects models of absolute running power and relative running power while
allowing intercepts as random effects for the participants to account for repeated measure-
ments within individuals [32]. Models were estimated using maximum likelihood, model
selection was assessed using a chi-squared likelihood ratio test, and model fit was assessed
using pseudo-R2 [32]. These analyses were performed for absolute (i.e., W) and relative
measures of power (i.e., W·kg−1). The spread of the participants’ intercepts was compared
using the
·
VO2 and absolute power and the
·
VO2 and relative power relationships and using
the speed and absolute power and speed and relative power relationships, employing the
Pitman–Morgan test for the homogeneity of variance of paired samples.
2.6.4. Stryd Running Power—Running Economy and Efficiency
To determine whether Stryd running power provides an indication of RE during
constant-speed treadmill running trials at MLSS, Pearson’s correlation coefficients were
calculated between metabolic power (kJ·kg−1·km−1) and each of the following variables:
absolute StrydMECH (kJ·km−1), relative StrydMECH (kJ·kg−1·km−1), absolute StrydMET
((J·s−1)·W−1), and relative StrydMET ((kJ·s−1)·(W·kg−1)−1). One-way repeated-measures
ANOVAs were used to assess the main effect of running intensity on metabolic power,
StrydMECH, and StrydMET measurements. Using the 30-min timepoint mean (i.e., 28–30 min)
data from the two MLSS trials, the reliability of metabolic power, StrydMECH, and StrydMET
were assessed by paired Student’s t tests, intraclass correlations, and Bland–Altman analy-
ses (with 95% limits of agreement). One-way repeated-measures ANOVAs were also used
to determine whether running intensity affected Stryd-derived assessments of EFF.
2.6.5. Stryd Running Power—Aerobic Fitness
To determine whether Stryd running power provides an indication of an athlete’s
aerobic fitness during constant-speed treadmill running, Pearson’s correlation coefficients
were calculated for the following pairs of variables:
·
VO2 at MLSS and running power at
MLSS, and
·
VO2 at MLSS and running speed at MLSS.
3. Results
3.1. Participants
Table 1 displays the female and male participant characteristics, incremental exercise
testing results, and MLSS testing results. All incremental tests were maximal, and the
duration of the incremental test portion of the SRS protocol was 12.1 [2.0] min. The
measured
·
VO2 during constant-speed running at MOD and HVY was 91.2 [8.0]% and 92.9
[5.6]% of the
·
VO2 at GET and RCP, respectively. All runners completed at least 30 min of
treadmill running at 5% below the MLSS, at the MLSS, and during the repeat trial at the
MLSS; however, seven runners were unable to complete 30 min of running at 5% above
the MLSS.
3.2. Stability, Sensitivity, and Reliability of Stryd Running Power
The 10- and 30-min running power measurements taken during constant-speed run-
ning trials near the MLSS are presented in Figure 2. While the intensity × duration
interaction and the main effect of duration were not statistically significant for running
power, there was a significant main effect for running intensity, with significant differences
between all pairs of intensities (p < 0.001 for all post hoc comparisons; Table 2).
Sensors 2023, 23, 8729
8 of 19
Table 1. Participant characteristics, maximal exercise results, and maximal lactate steady-state (MLSS)
results.
Sex (n)
Weight
(kg)
Maximal Exercising Measurements
MLSS Measurements
·
VO2max
(L·min−1)
MAS
(m·s−1)
MAP
(W)
·
VO2 at
MLSS
(L·min−1)
Speed at
MLSS
(m·s−1)
Power at
MLSS
(W)
Female (7)
65.5 [2.9]
3.37 [0.26]
4.29 [0.44]
280 [19]
2.96 [0.17]
3.29 [0.34]
224 [18]
Male (8)
71.7 [10.4]
4.10 [0.82]
4.54 [0.44]
330 [58]
3.61 [0.58]
3.43 [0.42]
254 [44]
Total (15)
68.8 [8.2]
3.76 [0.71]
4.42 [0.44]
307 [50]
3.30 [0.54]
3.35 [0.37]
240 [37]
·
VO2max, maximal oxygen uptake; MAS, maximal aerobic speed; MAP, maximal aerobic power. Data are reported
as mean [standard deviation].
Sensors 2023, 23, x FOR PEER REVIEW
9 of 20
Figure 2. Running power data near the maximal lactate steady state (MLSS). Panel (A) shows the
comparison between the 10-min and 30-min mean running power measurements during treadmill
running near the MLSS. Lines representing individual participants, asterisks (*) indicate statistically
significant differences between speeds, and error bars represent one standard deviation. Panel (B)
shows the relationship between 10 min and 30 min of running power from the first run at the MLSS,
and Panel (C) shows the relationship between 30 min of running power from the two separate runs
at the MLSS. Panels (D,E) show Bland–Altman plots corresponding to the data in Panels B and C,
respectively. In Panels (B–E), squares represent individual data, solid lines represent y=0, dashed
lines represent bias, and dotted lines represent 95% limits of agreement. n = 15 for all panels.
5% Below
MLSS
5% Above
Repeat MLSS
0
150
200
250
300
350
Speed
Running Power (W)
10-min
30-min
175
225
275
325
175
225
275
325
10-min Running Power (W)
30-min Running Power (W)
Y = 1.01 * X − 1.57
R2 = 1.00; P < 0.001
150
200
250
300
350
-5.0
-2.5
0.0
2.5
5.0
Mean of 10-min and 30-min
Running Power at MLSS (W)
30-min - 10-min
Running Power at MLSS (%)
LOA: −1.4 to 1.3%
Bias: 0.1%
175
225
275
325
175
225
275
325
First MLSS Trial
Running Power (W)
Repeat MLSS Trial
Running Power (W)
Y = 1.00 * X + 0.32
R2 = 0.99; P < 0.001
B
D
C
E
A
150
200
250
300
350
-5.0
-2.5
0.0
2.5
5.0
Mean of First and Repeat Trial
Running Power at MLSS (W)
First - Repeat MLSS Trial
Running Power (%)
LOA: −3.4 to 2.6%
Bias: -0.4%
Figure 2. Running power data near the maximal lactate steady state (MLSS). Panel (A) shows the compar-
ison between the 10-min and 30-min mean running power measurements during treadmill running near
the MLSS. Lines representing individual participants, asterisks (*) indicate statistically significant differ-
ences between speeds, and error bars represent one standard deviation. Panel (B) shows the relationship
Sensors 2023, 23, 8729
9 of 19
between 10 min and 30 min of running power from the first run at the MLSS, and Panel (C) shows
the relationship between 30 min of running power from the two separate runs at the MLSS. Panels
(D,E) show Bland–Altman plots corresponding to the data in Panels (B) and (C), respectively. In
Panels (B–E), squares represent individual data, solid lines represent y=0, dashed lines represent bias,
and dotted lines represent 95% limits of agreement. n = 15 for all panels.
Table 2.
·
VO2 and running power responses to exercise near the maximal lactate steady state (MLSS).
5% below MLSS
At MLSS
5% above MLSS
ANOVA
(DxI, D, I) b
10 min
30 min
10 min
30 min
10 min
30 min a
·
VO2
(L·min−1)
3.12 [0.55] ¶
3.16 [0.56] *,¶
3.26 [0.54] †
3.30 [0.54] *,†
3.41 [0.55] †,¶
3.46 [0.57] *,†,¶
0.202, 0.001,
<0.001
Running
Power (W)
231 [35] ¶
230 [35] ¶
239 [36] †
240 [37] †
250 [38] †,¶
250 [38] †,¶
0.334, 0.528,
<0.001
DxI, duration by intensity interaction; D, duration; I, intensity;
·
VO2, oxygen uptake. a Or the final 2 min if task
failure was < 30 min. b p-values are provided for these statistical tests. The * denotes a significant difference from
the 10-min timepoint at the same intensity (p < 0.05); the † denotes a significant difference from 5% below the
MLSS (p < 0.05); the ¶ denotes a significant difference from the MLSS (p < 0.05). n = 15 for all variables. Data are
reported as mean [standard deviation].
The 10- and 30-min running power measurements during the repeat constant-speed
running trial at MLSS are reported in Table 3. There was excellent reliability and low bias
between the running power measured at two time points within one run at the MLSS and
across two runs at the MLSS, without differences between repeated trials at the MLSS
(Figure 2; Table 3).
Table 3. Reliability of
·
VO2 and running power responses and metabolic and mechanical power
measurements to exercise at the maximal lactate steady state (MLSS).
At MLSS (Repeat) a
Reliability of Repeated Runs at MLSS (30-min)
10 min
30 min
t Test b
Bias
LOA
ICC
·
VO2 (L·min−1)
3.25 [0.54]
3.26 [0.52]
0.177
0.04
−0.18 to 0.27
0.99 (0.96 to 1.00)
Running Power (W)
240 [37]
241 [37]
0.322
−1
−8 to 6
1.00 (0.99 to 1.00)
Metabolic Power (kJ·kg−1·km−1)
-
4.99 [0.29]
0.249
0.06
−0.30 to 0.42
0.91 (0.74 to 0.97)
StrydMECH (kJ·km−1)
-
71.9 [9.5]
0.324
−0.28
−2.32 to 1.77
1.00 (0.99 to 1.00)
StrydMECH (kJ·kg−1·km−1)
-
1.04 [0.03]
0.331
0
−0.04 to 0.03
0.94 (0.82 to 0.98)
StrydMET((J·s−1)·W−1)
-
4.78 [0.30]
0.153
0.07
−0.29 to 0.44
0.91 (0.73 to 0.97)
StrydMET ((kJ·s−1)·(W·kg−1)−1)
-
0.33 [0.04]
0.121
0.01
−0.02 to 0.03
0.98 (0.93 to 0.99)
LOA, limits of agreement; ICC, intraclass correlation. a See Tables 2 and 4 for data from the first MLSS trial. Note
that metabolic and mechanical power measures are based on the 30-min time point only. b p-values are provided
for these statistical tests. Data are reported as mean [standard deviation].
Table 4. Mean metabolic and mechanical power measures during the moderate-intensity step
(MOD) and during constant-speed running 5% below, at, and 5% above the maximal lactate steady
state (MLSS).
6-min
30-min
ANOVA
(p-Value)
MOD
5% below MLSS
At MLSS
5% above MLSS
Metabolic Power (kJ·kg−1·km−1)
4.31 [0.36] *,†,¶
5.05 [0.35]
5.05 [0.36]
5.07 [0.30]
<0.001
StrydMECH (kJ·km−1)
73.8 [9.8] *,†,¶
72.3 [9.6] *,¶
71.7 [9.5]
71.1 [9.0]
<0.001
StrydMECH (kJ·kg−1·km−1)
1.07 [0.03] *,†,¶
1.05 [0.03] ¶
1.04 [0.03]
1.03 [0.03]
<0.001
StrydMET ((J·s−1)·W−1)
4.02 [0.29] *,†,¶
4.82 [0.40]
4.86 [0.34]
4.91 [0.32]
<0.001
StrydMET ((kJ·s−1)·(W·kg−1)−1)
0.28 [0.04] *,†,¶
0.33 [0.05]
0.33 [0.05]
0.34 [0.05]
<0.001
Data are based on the moderate-intensity step (MOD) from the “Step-Ramp-Step” protocol or the 30-min timepoint
of the indicated trial. * Denotes a significant difference between the denoted intensity compared to MLSS (p < 0.05).
† Denotes a significant difference between the denoted intensity compared to 5% below MLSS (p < 0.05). ¶ Denotes
a significant difference between the denoted intensity compared to 5% above MLSS (p < 0.05). Data are reported
as mean [standard deviation]. n = 15 for all variables.
Sensors 2023, 23, 8729
10 of 19
3.3. Physiological and Perceptual Responses
The duration × intensity interaction was not significant for
·
VO2; however, there was a
main effect of intensity, with significant differences across the three running speeds and
a main effect of duration, demonstrating 30-min values greater than the 10-min values
(p < 0.05 for all post hoc comparisons; Table 2). The
·
VO2 values measured at two time
points within one run at the MLSS had excellent reliability and low bias across two runs at
the MLSS, without differences between repeated trials at the MLSS (Table 3).
Descriptive data and statistical results for
·
VCO2, RER,
·
VE, HR, [BLa], and RPE mea-
sured at two time points (10-min and 30-min) for three speeds near the MLSS are reported
in the Supplementary Materials (Table S1).
3.4. Stryd Running Power—Association with
·
VO2 and Running Speed
The incremental and constant-speed
·
VO2–power gains and speed–power gains are
reported in Table 5. From the constant-speed running trials, the linear mixed-effects models
revealed a strong, positive relationship between absolute running power and
·
VO2 and
between relative power and
·
VO2 (Table 6; Figure 3). There was significant variance between
participant intercepts for both models that differed between models (Table 6; Figure 3),
providing evidence that the relationship between absolute power and
·
VO2 was stronger
and less variable between the participants than the relationship between relative power
and
·
VO2.
Table 5. The
·
VO2–power gain and speed–power gain calculated from the incremental exercise test
and constant-speed running trials.
Variable
Test
p-Value
Incremental
Constant-Speed
Absolute
·
VO2–power gain ((mL·min−1)·W−1)
11.6 [1.5]
19.8 [3.5]
<0.001
Relative
·
VO2–power gain ((mL·min−1)·(W·kg−1) −1)
810.9 [148.9]
1364.2 [298.7]
<0.001
Absolute speed–power gain ((m·s−1)·W−1)
0.015 [0.002]
0.015 [0.002]
0.365
Relative speed–power gain ((m·s−1)·(W·kg−1) −1)
1.05 [0.08]
1.03 [0.07]
0.333
Data are reported as mean [standard deviation]. N = 15 for all variables.
Table 6. The within-individual and between-individual association between running power and
·
VO2
measurements and between running power and running speed during the MOD, HVY, maximal
lactate steady state (MLSS) −5%, MLSS, and MLSS +5% running trials.
Variable
b
[95% CI]
Statistics
SD
[95% CI]
χ2 Statistics
Pitman–Morgan
Test
Absolute running
power and
·
VO2
18.2 [17.1, 19.3]
t(59) = 32.7; p < 0.001;
R2 = 0.97
196.9
[130.7, 296.8]
χ2(4) = −494.8;
p < 0.001
t(13) = −3.08;
p = 0.009
Relative running
power and
·
VO2
1246.3
[1150.2, 1342.4]
t(59) = 25.6; p < 0.001;
R2 = 0.95
414.0
[285.9, 599.6]
χ2(4) = 92.3;
p < 0.001
Absolute running
power and speed
0.015 [0.014, 0.015]
t(59) = 37.9; p < 0.001;
R2 = 0.97
0.414
[0.288, 0.596]
χ2(4) = 125.8;
p < 0.001
t(13) = 15.72;
p = 0.002
Relative running
power and speed
1.01 [0.96, 1.06]
t(59) = 42.7; p < 0.001;
R2 = 0.97
0.063
[0.039, 0.104]
χ2 (4) = 12.7;
p < 0.001
b denotes the calculated slope from the linear mixed-effects model. The units of the slope are (mL·min−1)·W−1
and (mL·min−1)·W−1 for absolute and relative running power and
·
VO2, respectively, and (m·s−1)·W−1 and
(m·s−1)·(W·kg−1)−1 for absolute and relative running power and speed, respectively. SD denotes the standard
deviation, which is presented in units of mL·min−1 and m·s−1 for
·
VO2 and speed variables, respectively.
Sensors 2023, 23, 8729
11 of 19
Variable
p-Value
Incremental
Constant-Speed
Absolute V̇ O2–power gain ((mL·min−1)·W−1)
11.6 [1.5]
19.8 [3.5]
<0.001
Relative V̇ O2–power gain ((mL·min−1)·(W·kg−1) −1)
810.9 [148.9]
1364.2 [298.7]
<0.001
Absolute speed–power gain ((m·s−1)·W−1)
0.015 [0.002]
0.015 [0.002]
0.365
Relative speed–power gain ((m·s−1)·(W·kg−1) −1)
1.05 [0.08]
1.03 [0.07]
0.333
Data are reported as mean [standard deviation]. N = 15 for all variables.
Figure 3. Relationships between absolute and relative running power, running speed, and oxygen
uptake (V̇ O2). Panels (A,B) show the relationships between absolute running power and V̇ O2 and
between absolute running power and running speed for each participant during the moderate
(MOD) and heavy (HVY) intensity steps and constant-speed trials near the maximal lactate steady
state (MLSS). Panels (C,D) show the relationships between relative running power and V̇ O2 and
between relative running power and running speed for each participant at each running intensity,
respectively. Each color represents a single participants set of trials. N = 15 for all panels.
3.5. Stryd Running Power—Association with Running Economy
Based on the constant-speed running trials at MOD and near the MLSS, there were
significant effects of intensity on metabolic power, StrydMECH, and StrydMET measurements
(Table 6). The metabolic power and StrydMET measurements were significantly lower at
MOD compared to measurements 5% below, at, and 5% above the MLSS (p < 0.001 for all
pairwise comparisons; Table 6). In contrast, the StrydMECH measurements were signifi-
cantly higher at MOD compared to the three higher intensities (p < 0.001 for all compari-
sons; Table 6). All variables had excellent reliability for the repeated trials at the MLSS,
without significant differences between trials at the MLSS (Table 3).
150
200
250
300
1000
2000
3000
4000
5000
Running Power (W)
V̇ O2 (mL·min−1)
2.0
2.5
3.0
3.5
4.0
4.5
1000
2000
3000
4000
5000
Running Power (W·kg−1)
V̇ O2 (mL·min−1)
150
200
250
300
2.0
2.5
3.0
3.5
4.0
4.5
Running Power (W)
Speed (m·s−1)
2.0
2.5
3.0
3.5
4.0
4.5
2.0
2.5
3.0
3.5
4.0
4.5
Running Power (W·kg−1)
Speed (m·s−1)
C
D
A
B
Figure 3. Relationships between absolute and relative running power, running speed, and oxygen
uptake (
·
VO2). Panels (A,B) show the relationships between absolute running power and
·
VO2 and
between absolute running power and running speed for each participant during the moderate (MOD)
and heavy (HVY) intensity steps and constant-speed trials near the maximal lactate steady state
(MLSS). Panels (C,D) show the relationships between relative running power and
·
VO2 and between
relative running power and running speed for each participant at each running intensity, respectively.
Each color represents a single participant’s set of trials. N = 15 for all panels.
Results were similar when speed was used in place of
·
VO2; however, the difference in
model intercept variances was in the opposite direction, with a stronger and less variable
relationship between relative power and speed compared to absolute power and speed
(Table 6; Figure 3).
3.5. Stryd Running Power—Association with Running Economy
Based on the constant-speed running trials at MOD and near the MLSS, there were
significant effects of intensity on metabolic power, StrydMECH, and StrydMET measurements
(Table 6). The metabolic power and StrydMET measurements were significantly lower at
MOD compared to measurements 5% below, at, and 5% above the MLSS (p < 0.001 for all
pairwise comparisons; Table 6). In contrast, the StrydMECH measurements were significantly
higher at MOD compared to the three higher intensities (p < 0.001 for all comparisons;
Table 6). All variables had excellent reliability for the repeated trials at the MLSS, without
significant differences between trials at the MLSS (Table 3).
Figure 4 depicts the relationships between metabolic power (kJ·kg−1·km−1) and
absolute StrydMECH (kJ·km−1), relative StrydMECH (kJ·kg−1·km−1), absolute StrydMET
((J·s−1)·W−1), and relative StrydMET ((kJ·s−1)·(W·kg−1)−1) at the MLSS. Metabolic power
(kJ·kg−1·km−1) was not significantly correlated with absolute StrydMECH (kJ·km−1) or
relative StrydMECH (kJ·kg−1·km−1); however, strong positive and moderately positive
correlations were detected between metabolic power (kJ·kg−1·km−1) and absolute StrydMET
(J·s−1)·W−1) and relative StrydMET ((kJ·s−1)·(W·kg−1)−1), respectively (Figure 4). The
results for other intensities were similar.
Sensors 2023, 23, 8729
12 of 19
lute StrydMECH (kJ km ), relative StrydMECH (kJ kg
km ), absolute StrydMET ((J s ) W ),
and relative StrydMET ((kJ·s−1)·(W·kg−1)−1) at the MLSS. Metabolic power (kJ·kg−1·km−1) was
not significantly correlated with absolute StrydMECH (kJ·km−1) or relative StrydMECH
(kJ·kg−1·km−1); however, strong positive and moderately positive correlations were de-
tected between metabolic power (kJ·kg−1·km−1) and absolute StrydMET (J·s−1)·W−1) and rela-
tive StrydMET ((kJ·s−1)·(W·kg−1)−1), respectively (Figure 4). The results for other intensities
were similar.
Figure 4. Relationships between metabolic power and absolute StrydMECH (A), relative StrydMECH (B),
absolute StrydMET (C), and relative StrydMET (D) during constant-speed running trials performed at
the maximal lactate steady state (MLSS). Circles represent individual data. n = 15 for all panels.
4.0
4.5
5.0
5.5
6.0
45.0
60.0
75.0
90.0
Metabolic Power (kJ·kg−1·km−1)
StrydMECH (kJ·km−1)
y = 7.5x + 33.9
R2 = 0.079
p = 0.313
4.0
4.5
5.0
5.5
6.0
0.9
1.0
1.1
1.2
Metabolic Power (kJ·kg−1·km−1)
StrydMECH (kJ·kg−1·km−1)
y = 0.024x + 0.92
R2 = 0.066
p = 0.356
4.0
4.5
5.0
5.5
6.0
4.0
4.5
5.0
5.5
6.0
Metabolic Power (kJ·kg−1·km−1)
StrydMET ((J·s−1)·W−1)
y = 0.83x + 0.64
R2 = 0.791
p < 0.001
4.0
4.5
5.0
5.5
6.0
0.3
0.3
0.4
0.4
0.5
0.5
Metabolic Power (kJ·kg−1·km−1)
StrydMET ((kJ·s−1)·(W·kg−1)−1)
y = 0.084x − 0.091
R2 = 0.373
p = 0.016
A
B
C
D
Figure 4. Relationships between metabolic power and absolute StrydMECH (A), relative StrydMECH
(B), absolute StrydMET (C), and relative StrydMET (D) during constant-speed running trials performed
at the maximal lactate steady state (MLSS). Circles represent individual data. n = 15 for all panels.
3.6. Stryd Running Power—Estimates of Mechanical Running Efficiency
There was a statistically significant main effect of running speed for EFF (p < 0.001;
Figure 5). Pairwise comparisons revealed that EFF was significantly higher at MOD (25.0
[1.8]%) compared to HVY (21.3 [1.2]%), 5% below MLSS (20.9 [1.7]%), MLSS (20.7 [1.4]%),
and 5% above MLSS (20.4 [1.4]%) (p < 0.001 for these pairwise comparisons). No other
significant differences were detected between the EFF measurements.
Sensors 2023, 23, x FOR PEER REVIEW
13 of 20
3.6. Stryd Running Power—Estimates of Mechanical Running Efficiency
There was a statistically significant main effect of running speed for EFF (p < 0.001;
Figure 5). Pairwise comparisons revealed that EFF was significantly higher at MOD (25.0
[1.8]%) compared to HVY (21.3 [1.2]%), 5% below MLSS (20.9 [1.7]%), MLSS (20.7 [1.4]%),
and 5% above MLSS (20.4 [1.4]%) (p < 0.001 for these pairwise comparisons). No other
significant differences were detected between the EFF measurements.
Figure 5. Average running mechanical efficiency (EFF) measurements during the moderate- (MOD)
and heavy-intensity (HVY) steps, and during constant-speed trials near the maximal lactate steady
state (MLSS). The asterisks (*) indicate statistically significant differences between intensities. Error
bars represent one standard deviation. Circles represent individual data. n = 15.
3.7. Stryd Running Power—Association with Aerobic Fitness
Absolute running power at the MLSS was strongly correlated with absolute V̇ O2 at
the MLSS, moderately correlated with relative V̇ O2 and absolute running speed at the
MLSS, and not correlated with relative running speed at the MLSS (Table 1; Figure 6).
Relative running power at the MLSS was not correlated with absolute V̇ O2 at the MLSS,
but it was strongly correlated with relative V̇ O2 and absolute running speed at the MLSS
and moderately correlated with relative running speed at the MLSS (Table 1; Figure 6).
MOD
HVY
Below
MLSS Above
0
10
20
30
Condition
Mechanical Efficiency (%)
*
Figure 5. Average running mechanical efficiency (EFF) measurements during the moderate- (MOD)
and heavy-intensity (HVY) steps, and during constant-speed trials near the maximal lactate steady
state (MLSS). The asterisks (*) indicate statistically significant differences between intensities. Error
bars represent one standard deviation. Circles represent individual data. n = 15.
3.7. Stryd Running Power—Association with Aerobic Fitness
Absolute running power at the MLSS was strongly correlated with absolute
·
VO2 at the
MLSS, moderately correlated with relative
·
VO2 and absolute running speed at the MLSS,
and not correlated with relative running speed at the MLSS (Table 1; Figure 6). Relative
Sensors 2023, 23, 8729
13 of 19
running power at the MLSS was not correlated with absolute
·
VO2 at the MLSS, but it
was strongly correlated with relative
·
VO2 and absolute running speed at the MLSS and
moderately correlated with relative running speed at the MLSS (Table 1; Figure 6).
Sensors 2023, 23, x FOR PEER REVIEW
14 of 20
Figure 6. Relationships between absolute and relative running power, running speed, and oxygen
uptake (V̇ O2) at the maximal lactate steady state (MLSS). Panels (A–D) show the relationship be-
tween V̇ O2 at the MLSS and running power at the MLSS in absolute and relative units. Panels (E–H)
show the relationship between running speed at the MLSS and running power at the MLSS in abso-
lute and relative units. Individual data are plotted, along with the regression lines. n = 15 for all
panels.
4. Discussion
The results from this investigation support the use of Stryd in research and applied
settings. The Stryd running power metric was stable during 30-min constant-speed
2.5
3.0
3.5
4.0
4.5
150
200
250
300
350
V̇ O2 at MLSS (L·min−1)
MLSS Power (W)
y = 61.7x + 35.8
R2 = 0.841
p < 0.001
40
45
50
55
60
150
200
250
300
350
V̇ O2 at MLSS (mL·kg−1·min−1)
MLSS Power (W)
y = 3.9x + 50.9
R2 = 0.280
p = 0.042
2.5
3.0
3.5
4.0
4.5
150
200
250
300
350
Speed at MLSS (m·s−1)
MLSS Power (W)
y = 51.6x + 66.3
R2 = 0.283
p = 0.041
0.03
0.04
0.05
0.06
0.07
150
200
250
300
350
Speed at MLSS (m·s−1·kg−1)
MLSS Power (W)
y = −596x + 269
R2 = 0.018
p = 0.638
2.5
3.0
3.5
4.0
4.5
2.5
3.0
3.5
4.0
4.5
V̇ O2 at MLSS (L·min−1)
MLSS Power (W·kg−1)
y = 0.28x + 2.57
R2 = 0.211
p = 0.085
40
45
50
55
60
2.5
3.0
3.5
4.0
4.5
V̇ O2 at MLSS (mL·kg−1·min−1)
MLSS Power (W·kg−1)
y = 0.052x + 1.00
R2 = 0.604
p < 0.001
2.5
3.0
3.5
4.0
4.5
2.5
3.0
3.5
4.0
4.5
Speed at MLSS (m·s−1)
MLSS Power (W·kg−1)
y = 0.83x + 0.71
R2 = 0.905
p < 0.001
0.03
0.04
0.05
0.06
0.07
2.5
3.0
3.5
4.0
4.5
Speed at MLSS (m·s−1·kg−1)
MLSS Power (W·kg−1)
y = 27.8x + 2.1
R2 = 0.478
p = 0.004
A
B
C
D
E
F
G
H
Figure 6. Relationships between absolute and relative running power, running speed, and oxygen
uptake (
·
VO2) at the maximal lactate steady state (MLSS). Panels (A–D) show the relationship between
·
VO2 at the MLSS and running power at the MLSS in absolute and relative units. Panels (E–H) show
the relationship between running speed at the MLSS and running power at the MLSS in absolute and
relative units. Individual data are plotted, along with the regression lines. n = 15 for all panels.
Sensors 2023, 23, 8729
14 of 19
4. Discussion
The results from this investigation support the use of Stryd in research and applied
settings. The Stryd running power metric was stable during 30-min constant-speed running
trials, repeatable across trials at the MLSS, and sensitive enough to differentiate between
trials performed at running speeds of 5% below, at, and 5% above the MLSS threshold.
Running power was strongly correlated with running speed and
·
VO2 during constant-
speed exercise relative to the GET and MLSS, supporting its use as a training intensity
metric. Furthermore, running power measurements at the MLSS were strongly associated
with both the
·
VO2 and running speed at the MLSS. Although metabolic power was strongly
associated with absolute StrydMET, it appears that Stryd power cannot provide an indication
of RE in isolation from metabolic data, as the associations between metabolic power and
StrydMECH were weak. Despite this finding, the mechanical running efficiency derived
using Stryd (i.e., EFF) remained consistent and proportional at various exercise intensities
near the MLSS threshold.
4.1. Stability, Sensitivity, and Reliability
Mean running power measurements were similar across the 10- and 30-min timepoints
during constant-speed running trials at 5% below, at, and 5% above the MLSS. Along
with a strong correlation, zero bias, and narrow LOA between time points, these findings
indicate that the Stryd signal remained stable during constant-speed treadmill running.
Running power across two runs at the MLSS was also strongly correlated, with a near-
zero bias and narrow LOA, indicating the excellent day-to-day reliability of the metric.
Furthermore, the Stryd power metric was able to distinguish between exercise intensities
near the MLSS. In agreement with our results, previous investigations also reported that
Stryd running power was stable during constant-speed running [33], repeatable [11], and
sensitive between conditions [34]; however, our investigation is the first to evaluate these
running power parameters near the MLSS, an important threshold for training programs
and fitness assessment [35,36]. In support of the Stryd running power metric results, besides
a significantly lower RPE measurement during the second compared to the first MLSS trial
(i.e., 0.8 units on the Borg 6–20 scale), which may indicate increased comfort during testing,
the
·
VO2 (Table 2) and other physiological and perceptual responses to running near the
MLSS were also stable, sensitive, and reliable (Table S1).
4.2. Stryd Running Power and Exercise Intensity
The strong associations observed between running power,
·
VO2, and speed support
the use of Stryd running power to guide exercise training relative to the exercise intensity
domains. Of note, the relationship between Stryd running power and
·
VO2, considered
at the group level, was stronger when running power was expressed in absolute units,
whereas the relationship between Stryd running power and speed, at the group level, was
stronger when running power was expressed in relative units. In practice, our results
suggest that absolute Stryd power may be best used as a metric to approximate the rate of
absolute oxygen consumption, while relative running power may be best used to indicate
running speed—at least during treadmill running. Due to the varying methodological ap-
proaches used to establish
·
VO2–power relationships in previous research [11,17,18,37–39],
it is difficult to make comparisons across studies.
A major strength of the current investigation is that exercise intensity domains were
delineated and the
·
VO2 was subsequently evaluated during appropriate durations of
constant-speed running before examining the relationship between
·
VO2 and running
power. Exercise in the heavy-intensity domain can result in a slow
·
VO2 component that
delays the attainment of a steady
·
VO2 measure by ~10–15 min or longer [40]. Thus, without
Sensors 2023, 23, 8729
15 of 19
appropriately delineating the exercise intensity domain, it is difficult to discern whether
a given absolute work rate or stage duration will produce steady-state exercising condi-
tions. The influence that intensity domain and
·
VO2 kinetic responses have on subsequent
·
VO2–power relationships can be highlighted by the substantial difference between the
incremental (i.e., 11.6 [1.5] (mL·min−1)·W−1) and constant-speed
·
VO2–power gains (i.e.,
19.8 [3.5] (mL·min−1)·W−1). Of interest, this
·
VO2–power gain from incremental tread-
mill running, measured using Stryd running power, is similar to previously observed
·
VO2–power gain measured during 15 W·min−1 incremental cycling protocols (i.e., 11.3
[1.2] (mL·min−1)·W−1) [29].
4.3. Stryd Running Power and Running Fitness
Runners with greater MLSS running powers displayed greater
·
VO2 and running
speeds at the MLSS (Figure 6). As the
·
VO2 and running speed associated with the MLSS
are strong predictors of running performance [41,42], at least in samples with broad aerobic
fitness ranges, it appears that the Stryd running power metric can be used to indicate fitness
in a similar manner to that used for cycling PO from constant-intensity exercise [43,44];
however, in contrast to cycling, where the cycling speed at any given PO is primarily
dictated by surface area and aerodynamics [45], body mass has a more substantial influence
on the relationship between Stryd running power and running speed. Thus, while absolute
Stryd running power may be used to estimate fitness in terms of absolute
·
VO2 at the MLSS,
in order to evaluate fitness from a speed perspective, it is best to interpret Stryd running
power relative to body mass or to only interpret the speed–power relationship relative to the
individual. Previous investigations have also reported strong associations between Stryd
assessments of critical power (CP) and fitness metrics such as the RCP and
·
VO2max [38,46],
providing further support for the utility of Stryd to quantify running fitness.
4.4. Stryd Running Power, Running Economy, and Mechanical Efficiency
Although strong associations were observed between absolute running power and
·
VO2 during constant-speed treadmill running conditions, there was a degree of variabil-
ity between the measured
·
VO2 for a given absolute running power (Figure 3). A large
proportion of this variance may be explained by the range of StrydMET requirements for a
specific metabolic power between runners (Figure 4). Indeed, runners with greater absolute
StrydMET measurements also exhibited greater metabolic power measures during each
constant-speed running intensity test (i.e., MOD and near to the MLSS). Although this
finding may suggest that Stryd can be used as an indication of RE, the strong relationship
between running power and running speed likely explains this finding. Accordingly, when
examining the relationship between metabolic power and StrydMECH (i.e., determining
whether the Stryd running power metric can be used in isolation from energy expenditure
to approximate the RE), there is no indication that StrydMECH is related to RE (i.e., metabolic
power), suggesting that this approach cannot distinguish between more and less economi-
cal runners. Previous investigations have similarly concluded that Stryd running power
metrics may be insufficient for detecting differences in RE between trained runners [46,47]
or detecting worsened RE (i.e., increased
·
VO2 at a given running speed) after purposefully
altering running biomechanics [37].
Our Stryd-derived measures of mechanical efficiency (~21–25%) are lower than pre-
vious estimates of “apparent” running mechanical efficiency during level running (e.g.,
~50–70%) [6,7,39] but are similar to estimates of gross cycling efficiency (e.g., ~20–25%) [4].
Furthermore, in comparison with the up to ~20% difference in previously reported esti-
mates of running efficiency measurements at various running speeds [6,7,39], the Stryd
estimates of running mechanical efficiency for level running during MOD and heavy-
Sensors 2023, 23, 8729
16 of 19
intensity running were relatively small (i.e., ~4%). Consequently, our results indicate that
Stryd-based measures of mechanical running efficiency remain relatively stable at various
submaximal intensities and that the metabolic requirement per unit of Stryd running power
and the metabolic requirement per unit of cycling PO are similar.
Despite certain limitations related to the accurate detection of changing RE [37,47] and
the quantification of running mechanical PO [39], our data suggest that foot-worn running
power metrics can still be used to monitor training and quantify running performance.
Although these findings do question the ability of Stryd running power to accurately
represent the running mechanical PO, we suggest that a wearable running power device
need not evaluate running power in a manner that is true to the definition of mechanical PO
to be useful. Indeed, as the relationship between metabolic demand and measurements of
running mechanical PO may vary with running speed, incline, and surface [6–8], a running
training tool that provides a consistent and seemingly equivalent evaluation of metabolic
demand may be more useful than one that evaluates external work rate, particularly for
such applied uses.
4.5. Experimental Considerations
Several limitations warrant discussion. Firstly, as all testing was performed on a
treadmill with a fixed incline (1%), it remains unknown whether our findings can be
extended to outdoor running conditions under variable running gradients, surfaces, or
air resistances. With varying inclines, Stryd has shown evidence of repeatability [11] and
strong correlations with
·
VO2 [11,18,39], but the influence of variable running gradients and
surfaces on metabolic cost requires further investigation. Secondly, it remains unknown
whether Stryd power can adjust for changes in air resistance, such as changes in wind
speed. For example, changes in air resistance (e.g., wind, drafting, or drag) impact the
cycling
·
VO2–speed relationship [45] without influencing the
·
VO2–PO relationship. As
Stryd seemingly derives its estimate of running power by quantifying positive changes in
vertical displacement and horizontal velocities, whether it can account for the increases in
mechanical PO required to overcome greater air resistance is unclear [13]. Despite evidence
that Stryd may detect changes in wind speed [48] and has introduced a metric, “Air power”,
to adjust running power based on changes in air resistance from increasing or decreasing
wind speeds and/or running speeds [49], it remains unknown whether the Stryd power
metric–
·
VO2 relationship is linear in uncontrolled environments.
5. Conclusions
A wide variety of internal and external load-monitoring methods have been used
in endurance sports, such as running speed and pace, RPE, [BLa], HR, step count, step
frequency, and distance [50]; however, none of these variables provide a continuous, instan-
taneous, and reliable method to measure training intensity, and imprecise measurements of
training stress may negatively affect performance and elevate injury risk. With evidence
of stability, reliability, and sensitivity, our study suggests that Stryd’s foot-worn wearable
device can be used to monitor training intensity and quantify aerobic fitness. While the
impact of variable running gradients, surfaces, and air resistance on the Stryd running
power metric still needs to be assessed, our results support the use of Styrd running power
to delineate exercise intensity domains, guide training intensity, and assess aerobic fitness
during level treadmill running.
Supplementary Materials: The following supporting information can be downloaded at: https://
www.mdpi.com/article/10.3390/s23218729/s1, Table S1: Mean physiological and perceptual responses
to exercise near the maximal lactate steady state (MLSS), and indices of reliability between two runs at
the MLSS.
Sensors 2023, 23, 8729
17 of 19
Author Contributions: Conceptualization: C.R.v.R. and M.J.M.; data curation: C.R.v.R. and M.J.M.;
formal analysis: C.R.v.R., J.K.G. and M.J.M.; funding acquisition: M.J.M.; investigation: C.R.v.R., O.O.A.
and K.M.S.; methodology: all authors; project administration: C.R.v.R. and M.J.M.; resources: M.J.M.;
software: M.J.M.; supervision: M.J.M.; validation: C.R.v.R., O.O.A., K.M.S. and M.J.M.; visualization:
C.R.v.R., J.K.G. and M.J.M.; writing—original draft: C.R.v.R. and M.J.M.; writing—reviewing and editing:
all authors. All authors have read and agreed to the published version of the manuscript.
Funding: This work was supported by an operating grant from the Natural Sciences and Engineering
Research Council of Canada (NSERC; grant number RGPIN-2018-06424) and start-up funding from
the Faculty of Kinesiology (University of Calgary) received by M.J.M. C.V.R. was funded by NSERC,
the NSERC CREATE Wearable Technology and Collaboration (We-TRAC) Training Program, an Al-
berta Innovates Graduate Student Scholarship for Data-Enabled Innovation, and an Alberta Graduate
Excellence Scholarship. OOA was also funded by NSERC and the NSERC CREATE We-TRAC. The
authors would like to acknowledge the contributions of all participants, students, faculty, and staff,
who assisted and made this investigation possible.
Institutional Review Board Statement: The study was conducted in accordance with the Declaration
of Helsinki, except for pre-registration of the trial, and was approved by the University of Calgary
Conjoint Health Research Ethics Board (REB20-0111, approved 29 July 2020).
Informed Consent Statement: Informed consent was obtained from all participants involved in
the study.
Data Availability Statement: Individual data are shown where possible; however, other data are
available from the corresponding author upon reasonable request.
Conflicts of Interest: M.J.M. has received several foot pods from Stryd for research purposes; how-
ever, that equipment was not used in this study. Stryd had no involvement in the conduct, analysis,
or reporting of this study, and M.J.M. has no professional relationship with Stryd. All authors declare
no competing interests. The funders of this study (NSERC, University of Calgary) had no role in
the design of the study; in the collection, analyses, or interpretation of data; in the writing of the
manuscript; or in the decision to publish the results.
References
1.
Williams, K.R.; Cavanagh, P.R. A model for the calculation of mechanical power during distance running. J. Biomech. 1983,
16, 115–128. [CrossRef] [PubMed]
2.
Arampatzis, A.; Knicker, A.; Metzler, V.; Bruggemann, G.P. Mechanical power in running: A comparison of different approaches.
J. Biomech. 2000, 33, 457–463. [CrossRef] [PubMed]
3.
Willems, P.A.; Cavagna, G.A.; Heglund, N.C. External, internal and total work in human locomotion. J. Exp. Biol. 1995, 198,
379–393. [CrossRef] [PubMed]
4.
Ettema, G.; Loras, H.W. Efficiency in cycling: A review. Eur. J. Appl. Physiol. 2009, 106, 1–14. [CrossRef]
5.
Keir, D.A.; Paterson, D.H.; Kowalchuk, J.M.; Murias, J.M. Using ramp-incremental
·
VO2 responses for constant-intensity exercise
selection. Appl. Physiol. Nutr. Metab. 2018, 43, 882–892. [CrossRef]
6.
Rasica, L.; Porcelli, S.; Minetti, A.; Pavei, G. Biomechanical and metabolic aspects of backward (and forward) running on uphill
gradients: Another clue towards an almost inelastic rebound. Eur. J. Appl. Physiol. 2020, 120, 2507–2515. [CrossRef]
7.
Monte, A.; Maganaris, C.; Baltzopoulos, V.; Zamparo, P. The influence of Achilles tendon mechanical behaviour on “apparent”
efficiency during running at different speeds. Eur. J. Appl. Physiol. 2020, 120, 2495–2505. [CrossRef]
8.
Lejeune, T.M.; Willems, P.A.; Heglund, N.C. Mechanics and energetics of human locomotion on sand. J. Exp. Biol. 1998,
201, 2071–2080. [CrossRef]
9.
Jamnick, N.A.; By, S.; Pettitt, C.D.; Pettitt, R.W. Comparison of the YMCA and a custom submaximal exercise test for determining
·
VO2max. Med. Sci. Sports Exerc. 2016, 48, 254–259. [CrossRef]
10.
Passfield, L.; Hopker, J.G.; Jobson, S.; Friel, D.; Zabala, M. Knowledge is power: Issues of measuring training and performance in
cycling. J. Sports Sci. 2017, 35, 1426–1434. [CrossRef]
11.
Cerezuela-Espejo, V.; Hernández-Belmonte, A.; Courel-Ibáñez, J.; Conesa-Ros, E.; Mora-Rodríguez, R.; Pallarés, J.G. Are we
ready to measure running power? Repeatability and concurrent validity of five commercial technologies. Eur. J. Sport. Sci. 2021,
21, 341–350. [CrossRef]
12.
Jaen-Carrillo, D.; Roche-Seruendo, L.E.; Carton-Llorente, A.; Ramirez-Campillo, R.; Garcia-Pinillos, F. Mechanical power in
endurance running: A scoping review on sensors for power output estimation during running. Sensors 2020, 20, 6482. [CrossRef]
[PubMed]
Sensors 2023, 23, 8729
18 of 19
13.
Stryd, T. How to Lead the Pack: Running Power Meters & Quality Data. 2022. Available online: https://blog.stryd.com/2017/1
2/07/how-to-lead-the-pack-running-power-meters-quality-data/ (accessed on 11 January 2022).
14.
García-Pinillos, F.; Roche-Seruendo, L.E.; Marcén-Cinca, N.; Marco-Contreras, L.A.; Latorre-Román, P.A. Absolute reliability and
concurrent validity of the Stryd system for the assessment of running stride kinematics at different velocities. J. Strength. Cond.
Res. 2021, 35, 78–84. [CrossRef] [PubMed]
15.
Navalta, J.W.; Montes, J.; Bodell, N.G.; Aguilar, C.D.; Radzak, K.; Manning, J.W.; DeBeliso, M. Reliability of trail walking and
running tasks using the Stryd power meter. Int. J. Sports Med. 2019, 40, 498–502. [CrossRef] [PubMed]
16.
Cartón-Llorente, A.; Roche-Seruendo, L.E.; Mainer-Pardos, E.; Nobari, H.; Rubio-Peirotén, A.; Jaén-Carrillo, D.; García-Pinillos, F.
Acute effects of a 60-min time trial on power-related parameters in trained endurance runners. BMC Sports Sci. Med. Rehabil.
2022, 14, 142. [CrossRef] [PubMed]
17.
García-Pinillos, F.; Latorre-Román, P.Á.; Roche-Seruendo, L.E.; García-Ramos, A. Prediction of power output at different running
velocities through the two-point method with the Stryd™ power meter. Gait Posture 2019, 68, 238–243. [CrossRef]
18.
Taboga, P.; Giovanelli, N.; Spinazze, E.; Cuzzolin, F.; Fedele, G.; Zanuso, S.; Lazzer, S. Running power: Lab based vs. portable
devices measurements and its relationship with aerobic power. Eur. J. Sport. Sci. 2022, 22, 1555–1568. [CrossRef]
19.
McKay, A.K.A.; Stellingwerff, T.; Smith, E.S.; Martin, D.T.; Mujika, I.; Goosey-Tolfrey, V.L.; Sheppard, J.; Burke, L.M. Defining
training and performance caliber: A participant classification framework. Int. J. Sports Physiol. Perform. 2022, 17, 317–331.
[CrossRef]
20.
Van Rassel, C.R.; Ajayi, O.O.; Sales, K.M.; Azevedo, R.A.; Murias, J.M.; MacInnis, M.J. A “Step-Ramp-Step” protocol to identify
running speed and power associated with the maximal metabolic steady state. Med. Sci. Sports Exerc. 2022, 55, 534–547. [CrossRef]
21.
Iannetta, D.; Inglis, E.C.; Pogliaghi, S.; Murias, J.M.; Keir, D.A. A “Step-Ramp-Step” protocol to identify the maximal metabolic
steady state. Med. Sci. Sports Exerc. 2020, 52, 2011–2019. [CrossRef]
22.
Beneke, R.; von Duvillard, S.P. Determination of maximal lactate steady state response in selected sports events. Med. Sci. Sports
Exerc. 1996, 28, 241–246. [CrossRef] [PubMed]
23.
Jones, A.M.; Doust, J.H. A 1% treadmill grade most accurately reflects the energetic cost of outdoor running. J. Sports Sci. 1996,
14, 321–327. [CrossRef] [PubMed]
24.
Beaver, W.L.; Wasserman, K.; Whipp, B.J. A new method for detecting anaerobic threshold by gas exchange. J. Appl. Physiol. 1986,
60, 2020–2027. [CrossRef] [PubMed]
25.
Whipp, B.J.; Davis, J.A.; Wasserman, K. Ventilatory control of the ‘isocapnic buffering’ region in rapidly-incremental exercise.
Respir. Physiol. 1989, 76, 357–367. [CrossRef] [PubMed]
26.
Borg, G.A. Psychophysical bases of perceived exertion. Med. Sci. Sports Exerc. 1982, 14, 377–381. [CrossRef]
27.
Billat, V.L.; Sirvent, P.; Py, G.; Koralsztein, J.-P.; Mercier, J. The concept of maximal lactate steady state: A bridge between
biochemistry, physiology and sport science. Sports Med. 2003, 33, 407–426. [CrossRef]
28.
Nixon, R.J.; Kranen, S.H.; Vanhatalo, A.; Jones, A.M. Steady-state
·
VO2 above MLSS: Evidence that critical speed better represents
maximal metabolic steady state in well-trained runners. Eur. J. Appl. Physiol. 2021, 121, 3133–3144. [CrossRef]
29.
Wilcox, S.L.; Broxterman, R.M.; Barstow, T.J. Constructing quasi-linear
·
VO2 responses from nonlinear parameters. J. Appl. Physiol.
2016, 120, 121–129. [CrossRef]
30.
Peronnet, F.; Massicotte, D. Table of nonprotein respiratory quotient: An update. Can. J. Sport. Sci. 1991, 16, 23–29.
31.
Koo, T.K.; Li, M.Y. A guideline of selecting and reporting intraclass correlation coefficients for reliability research. J. Chiropr. Med.
2016, 15, 155–163. [CrossRef]
32.
Hoffman, L. Longitudinal Analysis: Modeling within-Person Fluctuation and Change; Routledge: Abingdon, UK, 2015.
33.
García-Pinillos, F.; Soto-Hermoso, V.M.; Latorre-Román, P.Á.; Párraga-Montilla, J.A.; Roche-Seruendo, L.E. How does power
during running change when measured at different time intervals? Int. J. Sports Med. 2019, 40, 609–613. [CrossRef] [PubMed]
34.
Cerezuela-Espejo, V.; Hernandez-Belmonte, A.; Courel-Ibanez, J.; Conesa-Ros, E.; Martinez-Cava, A.; Pallares, J.G. Running
power meters and theoretical models based on laws of physics: Effects of environments and running conditions. Physiol. Behav.
2020, 223, 112972. [CrossRef] [PubMed]
35.
Iannetta, D.; Inglis, E.C.; Fullerton, C.; Passfield, L.; Murias, J.M. Metabolic and performance-related consequences of exercising
at and slightly above MLSS. Scand. J. Med. Sci. Sports 2018, 28, 2481–2493. [CrossRef] [PubMed]
36.
Iannetta, D.; Inglis, E.C.; Mattu, A.T.; Fontana, F.Y.; Pogliaghi, S.; Keir, D.A.; Murias, J.M. A Critical Evaluation of Current Methods
for Exercise Prescription in Women and Men. Med. Sci. Sports Exerc. 2020, 52, 466–473. [CrossRef]
37.
Baumgartner, T.; Held, S.; Klatt, S.; Donath, L. Limitations of foot-worn sensors for assessing running power. Sensors 2021,
21, 4952. [CrossRef] [PubMed]
38.
Ruiz-Alias, S.A.; Olaya-Cuartero, J.; Nancupil-Andrade, A.A.; Garcia-Pinillos, F. 9/3-Minute Running Critical Power Test:
Mechanical Threshold Location With Respect to Ventilatory Thresholds and Maximum Oxygen Uptake. Int. J. Sports Physiol.
Perform. 2022, 17, 1111–1118. [CrossRef]
39.
Imbach, F.; Candau, R.; Chailan, R.; Perrey, S. Validity of the Stryd power meter in measuring running parameters at submaximal
speeds. Sports 2020, 8, 103. [CrossRef]
40.
Poole, D.C.; Jones, A.M. Oxygen uptake kinetics. Compr. Physiol. 2012, 2, 933–996. [CrossRef]
41.
Bassett, D.R., Jr.; Howley, E.T. Limiting factors for maximum oxygen uptake and determinants of endurance performance. Med.
Sci. Sports Exerc. 2000, 32, 70–84. [CrossRef]
Sensors 2023, 23, 8729
19 of 19
42.
Jones, A.M.; Doust, J.H. The validity of the lactate minimum test for determination of the maximal lactate steady state. Med. Sci.
Sports Exerc. 1998, 30, 1304–1313. [CrossRef]
43.
Hawley, J.A.; Noakes, T.D. Peak power output predicts maximal oxygen uptake and performance time in trained cyclists. Eur. J.
Appl. Physiol. Occup. Physiol. 1992, 65, 79–83. [CrossRef] [PubMed]
44.
Smith, J.; Dangelmaier, B.; Hill, D. Critical power is related to cycling time trial performance. Int. J. Sports Med. 1999, 20, 374–378.
[CrossRef] [PubMed]
45.
Faria, E.W.; Parker, D.L.; Faria, I.E. The science of cycling: Factors affecting performance—Part 2. Sports Med. 2005, 35, 313–337.
[CrossRef] [PubMed]
46.
Dearing, C.G.; Paton, C.D. Is Stryd critical power a meaningful parameter for runners? Biol. Sport 2022, 40, 657–664. [CrossRef]
47.
Austin, C.L.; Hokanson, J.F.; McGinnis, P.M.; Patrick, S. The relationship between running power and running economy in
well-trained distance runners. Sports 2018, 6, 142. [CrossRef]
48.
Bello, M.L.; Anglin, D.A.; Gillen, Z.M.; Smith, J.W. The use of wearable technology to quantify power and muscle load differences
during running against varying wind resistances. Int. J. Kines. Sports Sci. 2022, 10, 11–15. [CrossRef]
49.
Stryd, T. New Power-Based Metric Reveals the Wind So You Can Benefit from Its Effects. 2019. Available online: https://blog.stryd.
com/2019/08/16/introducing-air-power/ (accessed on 29 March 2023).
50.
Bourdon, P.C.; Cardinale, M.; Murray, A.; Gastin, P.; Kellmann, M.; Varley, M.C.; Gabbett, T.J.; Coutts, A.J.; Burgess, D.J.; Gregson,
W.; et al. Monitoring Athlete Training Loads: Consensus Statement. Int. J. Sports Physiol. Perform. 2017, 12, S2161–S2170.
[CrossRef]
Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual
author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to
people or property resulting from any ideas, methods, instructions or products referred to in the content.
| Is Running Power a Useful Metric? Quantifying Training Intensity and Aerobic Fitness Using Stryd Running Power Near the Maximal Lactate Steady State. | 10-26-2023 | van Rassel, Cody R,Ajayi, Oluwatimilehin O,Sales, Kate M,Griffiths, James K,Fletcher, Jared R,Edwards, W Brent,MacInnis, Martin J | eng |
PMC7552741 | sports
Article
Physiological and Race Pace Characteristics of
Medium and Low-Level Athens Marathon Runners
Aristides Myrkos 1, Ilias Smilios 1,*
, Eleni Maria Kokkinou 1, Evangelos Rousopoulos 2*
and
Helen Douda 1
1
School of Physical Education and Sport Science, Democritus University of Thrace, 69100 Komotini, Greece;
[email protected] (A.M.); [email protected] (E.M.K.); [email protected] (H.D.)
2
Ergoscan Physical Performance Evaluation Center, Ionias 110, 17456 Alimos, Greece; [email protected]
*
Correspondence: [email protected]; Tel.: +30-2531039723
Received: 6 July 2020; Accepted: 19 August 2020; Published: 21 August 2020
Abstract: This study examined physiological and race pace characteristics of medium- (finish
time < 240 min) and low-level (finish time > 240 min) recreational runners who participated in a
challenging marathon route with rolling hills, the Athens Authentic Marathon. Fifteen athletes (age:
42 ± 7 years) performed an incremental test, three to nine days before the 2018 Athens Marathon,
to determine maximal oxygen uptake (VO2 max), maximal aerobic velocity (MAV), energy cost of
running (ECr) and lactate threshold velocity (vLTh), and were analyzed for their pacing during the
race. Moderate- (n = 8) compared with low-level (n = 7) runners had higher (p < 0.05) VO2 max
(55.6 ± 3.6 vs. 48.9 ± 4.8 mL·kg−1·min−1), MAV (16.5 ± 0.7 vs. 14.4 ± 1.2 km·h−1) and vLTh (11.6 ± 0.8
vs. 9.2 ± 0.7 km·h−1) and lower ECr at 10 km/h (1.137 ± 0.096 vs. 1.232 ± 0.068 kcal·kg−1·km−1).
Medium-level runners ran the marathon at a higher percentage of vLTh (105.1 ± 4.7 vs. 93.8 ± 6.2%)
and VO2 max (79.7 ± 7.7 vs. 68.8 ± 5.7%). Low-level runners ran at a lower percentage (p < 0.05) of
their vLTh in the 21.1–30 km (total ascent/decent: 122 m/5 m) and the 30–42.195 km (total ascent/decent:
32 m/155 m) splits. Moderate-level runners are less affected in their pacing than low-level runners
during a marathon route with rolling hills. This could be due to superior physiological characteristics
such as VO2 max, ECr, vLTh and fractional utilization of VO2 max. A marathon race pace strategy
should be selected individually according to each athlete’s level.
Keywords: endurance; aerobic performance; lactate threshold; running economy; maximal oxygen
consumption; oxygen fractional utilization; running
1. Introduction
Marathon running is one of the most demanding races which requires well-organized mental and
physical preparation [1]. Today, marathon races have turned into very large events where thousands
of elite, high-level and recreational athletes participate in this 42.195 m race [2–4]. For many years,
the physiological demands of a marathon as well as the physiological characteristics of top-class athletes
were examined by researchers [5–11]. It is known that the most important parameters to sustain the
highest possible running velocity over a marathon are the maximal oxygen uptake (VO2 max), a high
fractional utilization of VO2 max and the energy cost of running (ECr) [7,12,13]. These parameters
explain 70% of the variance of the average running speed sustained during a marathon race [6,7]
and are good indicators of the endurance performance of individuals of different ages, genders and
disciplines [1]. A typical VO2 max value for male top-class marathoners is about 70–85 mL/kg/min,
for low-level athletes around 65 mL/kg/min and for recreational runners about 51–58 mL/kg/min [14–16].
Additionally, oxygen fractional utilization at lactate threshold (LTh) intensity, the point where blood
lactate concentrations increase from baseline, is higher for top-class marathoners compared with
Sports 2020, 8, 116; doi:10.3390/sports8090116
www.mdpi.com/journal/sports
Sports 2020, 8, 116
2 of 10
low-level athletes (65–80% vs. 50–80% of VO2 max, respectively) and is also higher at the lactate
turn-point (LTP), the point where an abrupt increase in blood lactate is observed (85–90% vs. 80–85%
of VO2 max, respectively) [6,11,12,17,18].
Few studies have examined in more detail the physiological characteristics of recreational marathon
runners, with finishing times >3 h, and how these characteristics affect performance in this group of
runners. It was shown that the better the level of recreational marathoners, the higher the VO2 max as
well as the velocity and the VO2 at LTh [2,19]. No differences were observed between the different
level of runners in the LTh expressed as a percentage of VO2 max and the oxygen cost of running at
LTh [2]. Regarding medium- and low-level recreational runners, however, no data exist about the
correspondence of race pace on the blood lactate curve, the fractional utilization of VO2 max at race
pace and if these differ according to the performance ability of the runners.
Most of the studies examined runners who participated in a marathon ran on a flat terrain
where they could sustain a relatively stable pace till the end of the race, although the lower the
level of the runners, the higher the variability in race pace [20]. The peculiarity of the terrain could
be an external factor that may affect the physiological and race pace characteristics of a marathon
race. The terrain at one of the most famous and challenging marathons in the world, the Athens
Authentic Marathon, is characterized by rolling hills and includes the toughest uphill climb of any
major marathon. The total ascent is 317 m (51.2% of the route is uphill), the total descent is 262 m
(40.5% of the route is downhill) and the steepest grading ranges from −6.2 to 3.8% [21,22]. It is possible
that the difficulty of the route may affect differently the race pace characteristics of medium- and
low-level recreational runners. A runner with a faster pace will cross the hill segments in a shorter
amount of time compared with a slower runner, altering probably the physiological requirements of
the run. Therefore, recreational runners of different levels may run the Athens Marathon at a rate
corresponding to different percentages of aerobic performance parameters. This may lead athletes
and coaches to over- or underestimate the potential performance and to the determination of a false
race pace strategy. Therefore, it would be useful to examine which are the physiological and race pace
characteristics of medium- and low-level recreational athletes participating in the Athens Marathon
and if they adopt different pace characteristics in relation to their physiological profile. Based on the
above, the aim of the present study was to compare the physiological and race pace characteristics
of medium- (finish time < 240 min) and low-level (finish time > 240 min) recreational runners who
participated in the Athens Marathon.
2. Materials and Methods
2.1. Participants
Fifteen recreational marathon runners (age: 42 ± 7 years, height: 174.9 ± 6.5 cm and body
mass: 72.8 ± 6.9 kg) volunteered to take part in the study. All participants were healthy and ran
approximately 1–2 years on a systematic basis with a structured program with an average weekly load
of 50–60 km. Based on their finishing time at the Athens Marathon 2018, athletes were divided into a
moderate-level group, with finishing times < 240 min (n = 8), and a low-level group with finishing
times > 240 min (n = 7). Before the start of the study, the institutional review board committee approved
the experimental protocol in accordance with the Helsinki Declaration.
2.2. Maximal Incremental Test
Three to nine days before participation in the Athens Marathon 2018 race, participants performed
a maximal incremental test on a treadmill (Technogym run race 1200, Italy) for the determination of
VO2 max, maximum aerobic velocity (MAV), maximum heart rate (HRmax), the relationship between
blood lactate concentration and running velocity, oxygen consumption and running velocity, heart rate
and running velocity, and the energy cost of running.
Sports 2020, 8, 116
3 of 10
The protocol started at 7 km·h−1 and was increased by 1.5 km·h−1 every 3 min until volitional
exhaustion. Treadmill grade was set at 1% throughout the protocol. Gas exchange was measured by
the open circuit Douglas bag method as described by Cooke (2009). The subject breathed through a
low-resistance 2-way Hans-Rudolph 2700 B valve (Shawnee, OK, USA). The concentrations of CO2
and O2 in the expired air were measured by using the Hi-tech (GIR 250) combined Oxygen and
Carbon Dioxide Analyzer. The gas analyzers were calibrated continuously against standardized gases
(15.35% O2, 5.08% CO2 and 100% N2). Expired volume was measured by means of a dry gas meter
(Harvard) previously calibrated against standard air flow with a 3 L syringe. Barometric pressure
and gas temperature were recorded and respiratory gas exchange data for each work load (i.e., VO2,
VCO2 and VE) were determined based on the computations described by Cooke [23] when VEatps,
FECO2 and FEO2 are known. The highest VO2 value obtained during a 30-sec time period during
the incremental exercise test was recorded as the subject’s
.
VO2max. HR was continuously measured
telemetrically (Polar RS400) and the highest 10 sec value was regarded as maximal. The test was
considered as maximal when at least 3 of the following criteria were achieved: (a) visual exhaustion
of the participants, (b) a plateau in oxygen consumption (<2 mL kg−1·min−1) despite an increase
in running velocity, (c) maximal HR higher than 90% of the predicted maximum (220-age) and (d)
maximum respiratory exchange ratio > 1.1. MAV was calculated using the following formula: MAV
(km·h−1) = Velocity of the last completed stage + (seconds run at last stage/180).
2.3. LTh and LTP Determination
At the end of each stage during the incremental test, approximately 0.3 µL of whole blood was
collected from the fingertip and immediately analyzed for lactate concentration with a portable analyzer
(Lactate Pro 2, Arkray Factory Inc., Koka-Shi, Japan) using an enzymatic-amperometric method. The
individual relationships between blood lactate concentrations and running velocities were determined
using an exponential model: y = a + b × exp(x/c), where y = lactate concentration, x = running
velocity and a, b and c are constants. The LTh and the LTP were identified as the velocities (km·h−1)
at which blood lactate concentrations were increased by 0.3 and 1.5 mmol·L−1 from baseline values,
respectively. Furthermore, LTh and LTP were expressed relative to MAV (%MAV) units, and based on
the relationship between VO2 and running velocity were also expressed in absolute (mL·kg−1·min−1)
and relative (%VO2 max) VO2 max values.
2.4. Energy Cost of Running
The gas exchange data (VO2, VCO2) collected during the final 30 s of every 3-min stage up to the
previous stage from the LTP were used for the calculations of the caloric cost of running. Substrate
oxidation rate (g·min−1) was estimated using nonprotein respiratory quotient equations [24]:
Fat oxidation (g·min−1) = 1.6947 × VO2 (L·min−1) − 1.7012 × VCO2 (L·min−1)
Carbohydrate oxidation (g·min−1) = 4.5851 × VCO2 (L·min−1) − 3.22259 × VO2 (L·min−1)
The energy produced from each substrate was calculated by assuming an energy equivalent for
1 g of fat and carbohydrate of 9.75 and 4.07 kcal, respectively [25]. Total ECr was quantified from the
sum of these values and was expressed in kcal·kg−1·km−1. The energy cost of running at 10 km/h and
at the velocities corresponding to LTh and marathon race pace were estimated from the relationship
between exergy cost and running velocity derived from the incremental test.
All the physiological data were analyzed after the completion of the marathon race to avoid any
pacing strategies from the participants and their coaches based on the results of testing.
2.5. Route Characteristics and Race Pace Analysis
The profile of the Athens Marathon route includes rolling uphills and downhills. More specifically,
when calculated in 450 m intervals, the total ascent, total descent, the percent of uphill distance,
the percent of downhill distance and the steepest uphill and downhill, respectively, are for: (a) the total
route: 317 m, 262 m, 51.2%, 40.5%, 3.8% and −6.2%, (b) the 0–10 km split: 19 m, 36 m, 36.4%, 50%, 1.3%
Sports 2020, 8, 116
4 of 10
and −2.0%, (c) the 10–21.1 km split: 143 m, 66 m, 66.7%, 25%, 3.6% and −6.2%, (d) the 21.1–30 km split:
122 m, 5 m, 95%, 5%, 3.3% and −1.1% and (e) the 30–42.195 km split: 32 m, 155 m, 15.4%, 76.9%, 3.8%
and −5.1% [22].
Finishing time and split times for each participant were exported from the official results posted
on the site of the organization [21]. The average running velocity of each runner was calculated by
dividing marathon distance to the time needed to complete the race. Race pace was expressed as a
percentage of VO2 max (index of fractional utilization of VO2 max), MAV and the velocities at LTh
(vLTh) and LTP (vLTP).
To determine the differences between the two groups in pacing during the race, average running
velocities for the distances of 0–10, 10–21.1, 21.1–30 and 30–42.195 km were calculated by dividing the
distance of the split to the time to complete the split. For the analysis of the data, mean velocity of each
split was expressed as a percentage of the vLTh. The velocity at LTh was selected as a reference point
because for the whole sample, average marathon running velocity was equal to vLTh.
2.6. Statistical Analysis
All data are presented as means ± SD. Normality of the distribution of the data was examined
with the Shapiro–Wilk’s W test. A t-test was used to examine the differences among the medium-level
and the low-level runners in the physiological parameters and race pace characteristics measured.
A two-way analysis of variance with repeated measures in the second factor was used to examine the
differences between the two groups in the mean velocity of each running split (0–10, 10–21.1, 21.1–30
and 30–42.195 km). Significant differences between means were located with the Newman–Keuls
post hoc test. Pearson product moment correlations were used to determine the association between
marathon time and the measured parameters. The statistical significance level was set for all tests at
p < 0.05.
3. Results
3.1. Physiological Characteristics
Medium-level runners had higher (p < 0.05) VO2 max, MAV, LTh (km·h−1), LTh (%MAV), LTh
(mL·kg−1·min−1), LTP (km·h−1), LTP (%MAV) and LTP (mL·kg−1·min−1) than the low-level group.
There were no significant differences (p > 0.05) between groups at HRmax, LTh (%VO2 max) and LTP
(%VO2 max) (Table 1). Medium-level runners had lower ECr at 10 km·h−1 (p = 0.05), at vLTh (p = 0.07)
and at marathon race pace (p = 0.09) (Table 1).
Table 1. Physiological and race pace characteristics of the medium-level, the low-level and of all runners.
Medium-Level
Runners
Low-Level
Runners
All Runners
p Value between
Groups
Age (years)
41.00 ± 7.69
42.14 ± 7.20
41.53 ± 7.22
0.36
Body height (m)
175.00 ± 6.44
174.71 ± 7.06
174.87 ± 6.49
0.94
Body mass (kg)
72.50 ± 6.58
73.17 ± 7.91
72.81 ± 6.97
0.86
VO2 max
(mL·kg−1·min−1)
55.56 ± 3.62
48.85 ± 4.77
52.43 ± 5.32
0.01
MAV (km·h−1)
16.45 ± 0.74
14.39 ± 1.24
15.49 ± 1.44
0.01
HRmax (b·min−1)
178.25 ± 9.54
183.71 ± 9.25
180.80 ± 9.5
0.28
vLTh (km·h−1)
11.58 ± 0.81
9.22 ± 0.72
10.48 ± 1.42
0.01
vLTP (km·h−1)
13.6 ± 0.87
11.1 ± 0.8
12.43 ± 1.52
0.01
vLT1% (%MAV)
70.37 ± 3.9
64.25 ± 4.15
67.51 ± 5.00
0.01
vLT2% (%MAV)
82.7 ± 4.83
77.37 ± 4.52
80.21 ± 5.29
0.05
Sports 2020, 8, 116
5 of 10
Table 1. Cont.
Medium-Level
Runners
Low-Level
Runners
All Runners
p Value between
Groups
VO2 LTh
(mL·kg−1·min−1)
41.76 ± 1.81
35.11 ± 2.80
38.65 ± 4.10
0.01
VO2 LTP
(mL·kg−1·min−1)
48.04 ± 2.41
40.80 ± 3.78
44.66 ± 4.80
0.01
%VO2 LTh (%VO2 max)
75.31 ± 3.64
72.14 ± 5.87
73.83 ± 4.91
0.22
%VO2 LTP (%VO2 max)
86.59 ± 3.90
83.63 ± 4.08
85.21 ± 4.13
0.17
ECr 10 km·h−1
(kcal·kg−1·km−1)
1.137 ± 0.096
1.232 ± 0.068
1.181 ± 0.096
0.05
ECr vLTh
(kcal·kg−1·km−1)
1.157 ± 0.079
1.232 ± 0.066
1.192 ± 0.081
0.07
ECr Race Pace
(kcal·kg−1·km−1)
1.160 ± 0.083
1.232 ± 0.068
1.194 ± 0.082
0.09
Race pace (km·h−1)
12.14 ± 0.60
8.63 ± 0.64
10.50 ± 1.91
0.01
Race Pace (%MAV)
73.82 ± 2.60
60.11 ± 3.13
67.42 ± 7.59
0.01
Race Pace (%vLTh)
105.08 ± 4.71
93.80 ± 6.20
99.82 ± 7.84
0.01
Race Pace (%vLTP)
89.45 ± 4.47
77.92 ± 6.14
84.07 ± 7.85
0.01
Race Pace (%HRmax)
83.91 ± 5.8
77.41 ± 5.40
80.87 ± 6.37
0.04
Race Pace (%VO2 max)
79.74 ± 7.65
68.80 ± 5.73
74.63 ± 8.68
0.01
HRmax: maximum heart rate, MAV: maximal aerobic velocity, vLTh: velocity (km·h−1) at lactate threshold, vLTP:
velocity (km·h−1) at lactate turn-point, ECr: energy cost of running.
3.2. Race Pace Characteristics
Medium-level runners had, by design, a lower (p < 0.05) marathon time (209.0 ± 10.4 min, range:
194–225 min) than the low-level runners (289.7 ± 25.1 min, range: 260–328 min). Marathon finishing
time was not related to the number of days between the maximal incremental test and the race day
(Figure 1). Medium-level runners had a higher (p < 0.05) race pace expressed as %MAV, %vLTh, %vLTP,
%VO2 max and %HRmax (Table 1). Medium- and low-level runners had a similar (p > 0.05) race
pace (expressed as %vLTh) at the first two running splits (0–10 and 10–21.1 km). However, low-level
runners had a lower (p < 0.05) race pace at the last two splits (21.1–30 and 30–42.195 km) compared to
the medium-level runners (Figure 2).
Sports 2020, 8, x FOR PEER REVIEW
5 of 10
%VO2 LTP (%VO2
max)
86.59 ± 3.90
83.63 ± 4.08
85.21 ± 4.13
0.17
ECr 10 km·h−1
(kcal·kg−1·km−1)
1.137 ± 0.096
1.232 ± 0.068
1.181 ±
0.096
0.05
ECr vLTh
(kcal·kg−1·km−1)
1.157 ± 0.079
1.232 ± 0.066
1.192 ±
0.081
0.07
ECr Race Pace
(kcal·kg−1·km−1)
1.160 ± 0.083
1.232 ± 0.068
1.194 ±
0.082
0.09
Race pace (km·h−1)
12.14 ± 0.60
8.63 ± 0.64
10.50 ± 1.91
0.01
Race Pace (%MAV)
73.82 ± 2.60
60.11 ± 3.13
67.42 ± 7.59
0.01
Race Pace (%vLTh)
105.08 ± 4.71
93.80 ± 6.20
99.82 ± 7.84
0.01
Race Pace (%vLTP)
89.45 ± 4.47
77.92 ± 6.14
84.07 ± 7.85
0.01
Race Pace (%HRmax)
83.91 ± 5.8
77.41 ± 5.40
80.87 ± 6.37
0.04
Race Pace (%VO2 max)
79.74 ± 7.65
68.80 ± 5.73
74.63 ± 8.68
0.01
HRmax: maximum heart rate, MAV: maximal aerobic velocity, vLTh: velocity (km·h−1) at lactate
threshold, vLTP: velocity (km·h−1) at lactate turn-point, ECr: energy cost of running.
3.2. Race Pace Characteristics
Medium-level runners had, by design, a lower (p < 0.05) marathon time (209.0 ± 10.4 min, range:
194–225 min) than the low-level runners (289.7 ± 25.1 min, range: 260–328 min). Marathon finishing
time was not related to the number of days between the maximal incremental test and the race day
(Figure 1). Medium-level runners had a higher (p < 0.05) race pace expressed as %MAV, %vLTh,
%vLTP, %VO2 max and %HRmax (Table 1). Medium- and low-level runners had a similar (p > 0.05)
race pace (expressed as %vLTh) at the first two running splits (0–10 and 10–21.1 km). However, low-
level runners had a lower (p < 0.05) race pace at the last two splits (21.1–30 and 30–42.195 km)
compared to the medium-level runners (Figure 2).
Figure 1. Plot of marathon finishing time vs. number of days between maximal incremental test and
race day for the low-(squares) and the medium-level runners (triangles).
Figure 1. Plot of marathon finishing time vs. number of days between maximal incremental test and
race day for the low-(squares) and the medium-level runners (triangles).
Sports 2020, 8, 116
6 of 10
Sports 2020, 8, x FOR PEER REVIEW
6 of 10
Figure 2. Race pace, expressed as a percentage of the velocity at lactate threshold (%vLTh), at the
running splits of 0–10, 10–21.1, 21.1–30 and 30–42.195 km (total ascent in meters/total decent in meters)
of the Athens Marathon, for the low-level, the medium-level and all runners. a: p < 0.05 significant
difference between low- and medium-level runners, b: p < 0.05 significantly different from the 0–10
and 10–21.1 km splits for the low-level runners.
3.3. Correlation between Marathon Time and Measured Variables
Marathon finish time correlated significantly (p < 0.05) with VO2 max (r = −0,76), MAV (r = −0.88),
vLTh (km·h−1; r = −0.91), vLTP (km·h−1; r = −0.88), LTh (%MAV; r = −0.58), LTh (mL·kg−1·min−1; r =
−0.86), LTP (mL·kg−1·min−1; r = −0.80), ECr 10 km·h−1 (r = 0.62), ECr vLTh (r = 0.59), ECr race pace (r =
0.55), race pace (%VO2 max; r = −0.62), race pace (%vLTh; r = −0.75), race pace (%vLTP; r = −0.81) and
race pace (%MAV; r = −0.90). Marathon finish time did not correlate significantly (p > 0.05) with LTP
(%MAV; r = −0.38), LTh (%VO2 max; r = −0.22) and LTP (%VO2 max; r = −0.21).
4. Discussion
The purpose of this study was to provide further insight into the physiological and race pace
characteristics of medium- and low-level marathon runners with a completion time < 240 min and >
240 min, respectively, of the Athens Authentic Marathon. This marathon race is famous, not only for
historical reasons but also for its level of difficulty due to the peculiarity of the terrain. The results of
the present study show that recreational medium-level runners compared to lower-level runners
have: (a) higher VO2 max, MAV and lactate threshold values in absolute velocity (km·h−1) and VO2
(mL·kg−1·min−1) units, (b) higher lactate threshold in relative velocity units (%MAV), (c) lower energy
cost of running at 10 km/h and (d) adopt a race pace corresponding to a higher percentage of their
lactate threshold velocity and fractional utilization of VO2 max and show no significant alterations in
their pace due to terrain alterations in contrast to the low-level runners to whom the uphill part of
the race leads to great reductions in race pace.
Previous studies have examined the importance of physiological parameters and race pace
characteristics of elite marathoners, but few studies provide data for recreational runners [2,19,20].
Maximal oxygen consumption, a high fractional utilization of VO2 max and the energy cost of running
are considered the determinants of endurance performance [26]. Indeed, in the present study the
medium-level runners had higher VO2 max than the low-level runners. This agrees with previous
reports where the better level marathoners had higher VO2 max than the lower level [2,7,18,19]. The
VO2 max values of the medium-level marathoners (55.56 ± 3.62 mL/kg/min) measured in the present
Figure 2. Race pace, expressed as a percentage of the velocity at lactate threshold (%vLTh), at the
running splits of 0–10, 10–21.1, 21.1–30 and 30–42.195 km (total ascent in meters/total decent in meters)
of the Athens Marathon, for the low-level, the medium-level and all runners. a: p < 0.05 significant
difference between low- and medium-level runners, b: p < 0.05 significantly different from the 0–10
and 10–21.1 km splits for the low-level runners.
3.3. Correlation between Marathon Time and Measured Variables
Marathon finish time correlated significantly (p < 0.05) with VO2 max (r = −0,76), MAV (r = −0.88),
vLTh (km·h−1; r = −0.91), vLTP (km·h−1; r = −0.88), LTh (%MAV; r = −0.58), LTh (mL·kg−1·min−1;
r = −0.86), LTP (mL·kg−1·min−1; r = −0.80), ECr 10 km·h−1 (r = 0.62), ECr vLTh (r = 0.59), ECr race pace
(r = 0.55), race pace (%VO2 max; r = −0.62), race pace (%vLTh; r = −0.75), race pace (%vLTP; r = −0.81)
and race pace (%MAV; r = −0.90). Marathon finish time did not correlate significantly (p > 0.05) with
LTP (%MAV; r = −0.38), LTh (%VO2 max; r = −0.22) and LTP (%VO2 max; r = −0.21).
4. Discussion
The purpose of this study was to provide further insight into the physiological and race pace
characteristics of medium- and low-level marathon runners with a completion time < 240 min and
> 240 min, respectively, of the Athens Authentic Marathon. This marathon race is famous, not only for
historical reasons but also for its level of difficulty due to the peculiarity of the terrain. The results
of the present study show that recreational medium-level runners compared to lower-level runners
have: (a) higher VO2 max, MAV and lactate threshold values in absolute velocity (km·h−1) and VO2
(mL·kg−1·min−1) units, (b) higher lactate threshold in relative velocity units (%MAV), (c) lower energy
cost of running at 10 km/h and (d) adopt a race pace corresponding to a higher percentage of their
lactate threshold velocity and fractional utilization of VO2 max and show no significant alterations in
their pace due to terrain alterations in contrast to the low-level runners to whom the uphill part of the
race leads to great reductions in race pace.
Previous studies have examined the importance of physiological parameters and race pace
characteristics of elite marathoners, but few studies provide data for recreational runners [2,19,20].
Maximal oxygen consumption, a high fractional utilization of VO2 max and the energy cost of running
are considered the determinants of endurance performance [26]. Indeed, in the present study the
medium-level runners had higher VO2 max than the low-level runners. This agrees with previous
reports where the better level marathoners had higher VO2 max than the lower level [2,7,18,19].
Sports 2020, 8, 116
7 of 10
The VO2 max values of the medium-level marathoners (55.56 ± 3.62 mL/kg/min) measured in the
present study are approximately the same (55.7 ± 4.8) as those reported by Gordon et al. [2] for
athletes who ran the marathon between 3:00 and 3:30 h as in the present study. The same holds
even for the low-level runners (VO2 max: 48.85 ± 4.77 mL/kg/min; finish time: 4:00–5:30 h) of the
present study and runners with approximately similar finishing times in Gordon et al.’s [2] (VO2 max:
46.5 ± 5.2 mL/kg/min; finish time: >4:30 h) and Chmura et al.’s [14] (VO2 max: 51 ± 2 mL/kg/min;
finish time: 4:17 ± 10.51 min) studies. It appears that certain levels of VO2 max are necessary to achieve
certain marathon times regardless of the level of the runner since VO2 max determines the upper limit
of aerobic performance. The high correlation between VO2 max and marathon performance which
has been previously reported for high-level to elite athletes [7,18,27,28] supports this notion. A large
correlation (r = −0.76) between VO2 max and marathon performance was observed as well in the
present study for medium- to low-level marathon runners, enriching the limited information available
for recreational athletes [2,19].
For most sports scientists, running economy or energy cost of running is a key factor for
performance in long distance events and becomes more important as running distance increases [29–32]).
In the present study, we examined ECr at a specific speed (10 km/h) and at the vLTh and we found
that medium-level runners had lower ECr than the lower-level runners. This was probably another
factor that allowed them to run the marathon at a faster pace. It should be noted that the ECr in the
medium-level runners tended to be lower in the race pace as well. This is of importance considering
that the medium-level runners sustained a faster running pace. Furthermore, ECr at 10 km/h and at
vLTh had large correlations with marathon time (r = 0.62 and r = 0.59, respectively). The results of
the current study reveal that running economy is a determinant of performance even for recreational
runners with limited training experience and supports the suggestion in the literature that athletes
should focus their training on the optimization of this parameter as well [32–34].
Besides the importance of VO2 max and energy cost of running for marathon performance,
stronger associations are observed between maximal aerobic velocity and the velocities at the lactate
threshold or any point on the blood lactate curve, and endurance performance [35–37]. Similarly, very
large correlations were observed in this study between marathon time and velocities at LTh and LTP
(r = −0.91 and −0.88) and MAV (r = −0.88) of recreational runners. The velocity at LTh was the stronger
single predictor of marathon finish time. This is not surprising considering that these indexes, when
expressed in velocity units, encompass both VO2 max and running economy [37]. When LTh and LTP
values were normalized to MAV and VO2 max, the relationship of these parameters with running
performance became lower (r = −0.22 to −0.58). This is because the effect of VO2 max and/or running
economy was diminished [37]. Medium-level runners had higher LTh values, expressed either in
velocity or VO2 units, than the lower-level runners. Even when LTh velocity was normalized to MAV,
medium-level runners had a higher LTh, indicating a higher ability of the fat oxidation rate to meet
ATP demands and the occurrence of a later increased stimulation of glycolysis and glycogenolysis
relative to their maximum performance. This probably reflects a greater aerobic capacity and increased
buffering capacity promoting the ability to achieve higher running velocities due to metabolic and/or
locomotor reasons.
Many studies declare that the fractional utilization of VO2 max at LTh, LTP and at race pace
is one of the most crucial parameters of aerobic performance along with VO2 max and running
economy [6,7,31]. Fractional utilization of VO2 max at LTh and LTP did not differ between the two
groups in the present study. Similarly, Gordon et al. [2] did not find any differences in fractional
utilization of VO2 max at LTh and LTP between recreational runners with different marathon finish
times. It could be that adaptations in the utilization of oxygen from working muscles may require
a significant amount of training load which was not achieved by our runners. In the present study,
however, we found that medium-level runners had a higher fractional utilization of VO2 max at
marathon race pace. This agrees with previous findings that in high-level athletes, increased levels of
fractional utilization of VO2 max at marathon race pace were associated with faster performance [1,2,7].
Sports 2020, 8, 116
8 of 10
In addition, a positive correlation of fractional utilization at race pace and marathon time was found (r
= −0.62), Therefore, our data reveal that even in recreational runners, fractional utilization of VO2 max
at marathon race pace appears to be a contributing factor to performance.
A main finding of the present study is that medium-level marathoners ran the marathon distance
with an average speed corresponding to higher percentages of vLTh and MAV. The better running
economy may allow them to adopt a higher running velocity. Furthermore, the higher LTh and LTP
velocities mean that the medium-level runners will cover a given distance at a shorter time. This may
allow them to run at a higher point on the blood lactate curve because they can sustain this pace
for the time needed to complete the race. On the contrary, the slower LTh and LTP velocities of the
low-level runners mean that they need to run for a longer time to complete the race having a lower
fractional oxygen utilization. It has been shown that as the duration of an endurance event increases,
fractional oxygen utilization decreases [38,39]. The lower running velocity of the low-level runners
made them spend more time running the uphill part of the Athens Marathon course. The total ascent
from the 21.1st km to the 30th km is about 122 m and almost all this split is uphill. This forced the
low-level runners to adopt an even lower velocity during this part of the route. Indeed, the split
analysis revealed that the low-level runners were more influenced by this uphill part than the medium
level. It appears that this specific segment has the greatest impact in the finish time between different
levels of athletes and makes the Athens Marathon a totally different terrain from other marathons. It is
worth noting that even at the last part of the route, which is mostly downhill, low-level runners were
not able to increase their speed. Probably, the accumulated fatigue after hours of running may increase
even more the stress placed on the musculoskeletal system, besides that induced by the increased
eccentric load during downhill running, which prevents an increase in running speed compared to the
previous uphill part. Therefore, the peculiarity of the terrain may affect differently the performance of
a marathon runner depending on his/her ability level. This is of importance for coaches and athletes
for the determination of the pace strategy to follow when running on a rolling hill terrain.
An advantage of the present study is that all recreational runners participated in the same
marathon race and not in different ones. This makes comparisons between different levels of runners
more reliable since all of them competed in the same route, on the same day and under the same
environmental conditions. In addition, physiological testing was performed at a time point very close
to the race day (three–nine days before) providing valid data about the relationship of physiological
determinants of endurance performance and actual marathon running performance. Limitations of the
present study, though, should also be acknowledged. A larger sample size would have given more
valid data about the different levels of marathon runners. It was difficult, however, to measure many
runners at a time close to the actual race. Furthermore, the energy cost of running at 10 km/h and at the
velocity corresponding to LTh was estimated from the relationship between exergy cost and running
velocity derived from the incremental test. Measurements at the exact velocities would have given
more precise values of the energy cost. Again, the execution of these submaximal measurements would
have increased the time of testing and it would not be possible to perform them near the race date.
5. Conclusions
The results of the present study enrich the existing literature regarding the physiological profile
and the race pace characteristics of recreational marathon runners competing in a difficult route,
the Athens Marathon. Medium-level runners (finish time range: 194–225 min) have higher VO2 max,
lactate threshold values, better running economy, greater oxygen fractional utilization at race pace and
adopt a faster race pace in relation to their lactate threshold velocity than low-level runners (finish
time range: 260–328 min). Furthermore, medium-level runners show no significant alterations in
their pace due to terrain alterations in contrast to the low-level runners to whom the uphill part of
the race leads to great reductions in race pace. Therefore, slower runners are more influenced by a
hilly terrain and they decrease more their running velocity to complete this part of the race. Thus,
careful planning of race pace should be considered so that pacing of the parts before the uphill would
Sports 2020, 8, 116
9 of 10
be of such an intensity to avoid a large decrease in running velocity at uphill. Therefore, besides the
focus on training for the improvement of important physiological parameters related to endurance
performance, it is recommended that the selected race pace strategy be applied individually according
to each athlete’s level.
Author Contributions: Conceptualization, A.M. and I.S.; methodology, A.M., I.S., E.M.K., E.R. and H.D.;
investigation, E.M.K. and E.R.; data analysis, A.M., I.S., E.M.K. and E.R.; writing—original draft preparation, A.M.,
I.S. and H.D.; writing—review and editing, A.M., I.S. and H.D.; supervision, I.S. and H.D. All authors have read
and agreed to the published version of the manuscript.
Funding: This research received no external funding.
Acknowledgments: The authors wish to thank the recreational runners who participated in the study.
Conflicts of Interest: The authors declare no conflict of interest.
References
1.
Zinner, C.; Sperlich, B. Marathon Running: Physiology, Psychology, Nutrition and Training Aspects; Springer
International Publishing: Berlin, Germany, 2016.
2.
Gordon, D.; Wightman, S.; Basevitch, I.; Johnstone, J.; Espejo-Sanchez, C.; Beckford, C.; Boal, M.; Scruton, A.;
Ferrandino, M.; Merzbach, V.; et al. Physiological and training characteristics of recreational marathon
runners. Open Access J. Sports Med. 2007, 8, 231–241. [CrossRef] [PubMed]
3.
Berndsen, J.; Smyth, B.; Lawlor, A. Pace my race: Recommendations for marathon running. In Proceedings
of the 13th ACM Conference on Recommender Systems, Copenhagen, Denmark, 16–20 September 2019.
4.
Fokkema, T.; van Damme, A.; Fornerod, M.; de Vos, R.J.; Bierma-Zeinstra, S.; van Middelkoop, M. Training
for a (half-)marathon: Training volume and longest endurance run related to performance and running
injuries. Scand. J. Med. Sci. Sports in press. [CrossRef] [PubMed]
5.
Foster, C.; Daniels, J.; Yarbrough, R. Physiological and training correlates of marathon running performance.
Aust. J. Sports Med. 1977, 9, 58–61.
6.
Di Prampero, P.E.; Atchou, G.; Brückner, J.-C.; Moia, C. The energetics of endurance running. Eur. J. Appl.
Physiol. Occup. Physiol. 1986, 55, 259–266. [CrossRef]
7.
Billat, V.; Demarle, A.; Slawinski, J.; Paiva, M.; Koralsztein, J.-P. Physical and training characteristics of
top-class marathon runners. Med. Sci. Sports Exerc. 2001, 33, 2089–2097. [CrossRef]
8.
Billat, V.; Demarle, A.; Paiva, M.; Koralsztein, J.P. Effect of Training on the Physiological Factors of Performance
in Elite Marathon Runners (Males and Females). Int. J. Sports Med. 2002, 23, 336–341. [CrossRef]
9.
Billat, V.; Petot, H.; Landrain, M.; Meilland, R.; Koralsztein, J.P.; Mille-Hamard, L. Cardiac Output and
Performance during a Marathon Race in Middle-Aged Recreational Runners. Sci. World J. 2012, 2012, 1–9.
[CrossRef]
10.
Schmid, W.; Knechtle, B.; Knechtle, P.; Barandun, U.; Rüst, C.A.; Rosemann, T.; Lepers, R. Predictor Variables
for Marathon Race Time in Recreational Female Runners. Asian J. Sports Med. 2012, 3, 90–98. [CrossRef]
11.
Vernillo, G.; Schena, F.; Galvani, C.; Maggioni, M.A. A thropometric characteristics of top class Kenyan
marathon runners. J. Sports Med. Physic. Fit. 2013, 53, 403–408.
12.
Joyner, M.J.; Coyle, E.F. Endurance exercise performance: The physiology of champions. J. Physiol. 2008, 586,
35–44. [CrossRef]
13.
Jones, A.M. The Physiology of the World Record Holder for the Women’s Marathon. Int. J. Sports Sci. Coach.
2006, 1, 101–116. [CrossRef]
14.
Chmura, J.; Chmura, P.; Konefat, M.; Batra, A.; Mroczek, D.; Kosowski, M.; Mlynarska, K.; Andrzejewski, D.;
Rokita, A.; Ponikowski, P. The effects of a marathon effort on psychomotor performance ad catecholamine
concentration in runners over 50 years of age. Appl. Sci. 2020, 10, 2067. [CrossRef]
15.
Hernando, C.; Hernando, C.; Collado, E.J.; Panizo, N.; Martinez-Navarro, I.; Hernando, B. Establishing
cut-points for physical activity classification using triaxial accelerometer in middle-aged recreational
marathoners. PLoS ONE 2018, 13, e0202815. [CrossRef] [PubMed]
16.
Hernando, C.; Hernando, C.; Martinez-Navaro, I.; Collado-Boira, E.; Panizo, N.; Hrenando, B.; Hernando, B.
Estimation of energy consumed by midlle aged recreational marathoners during a marathon using
accelerometry-based devices. Sci. Rep. 2020, 10, 1523. [CrossRef] [PubMed]
Sports 2020, 8, 116
10 of 10
17.
Beneke, R.; Leithäuser, R.M.; Ochentel, O. Blood Lactate Diagnostics in Exercise Testing and Training. Int. J.
Sports Physiol. Perform. 2011, 6, 8–24. [CrossRef]
18.
Sjödin, B.; Svedenhag, J. Applied Physiology of Marathon Running. Sports Med. 1985, 2, 83–99. [CrossRef]
19.
Christensen, C.L.; Ruhling, R.O. Physical characteristics of novice and experienced women marathon runners.
Br. J. Sports Med. 1983, 17, 166–171. [CrossRef]
20.
Haney, T.A.; Mercer, J. A Description of Variability of Pacing in Marathon Distance Running. Int. J. Exerc. Sci.
2011, 4, 133–140.
21.
Available online: www.athensauthenticmarathon.gr (accessed on 2 August 2020).
22.
Available online: www.plotaroute.com/routeprofile/20092 (accessed on 2 August 2020).
23.
Cooke, C. Maximal oxygen uptake, economy and efficiency. In Kinanthropometry and Exercise Physiology
Laboratory Manual, 3rd ed.; Eston, R., Reilly, T., Eds.; Taylor & Francis Group: London, UK, 2009; pp. 145–184.
24.
Péronnet, F.; Massicotte, D. Table of nonprotein respiratory quotient: An update. Can. J. Sport Sci. 1991, 16,
23–29.
25.
Jeukendrup, A.E.; Wallis, G.A. Measurement of Substrate Oxidation during Exercise by Means of Gas
Exchange Measurements. Int. J. Sports Med. 2005, 26, S28–S37. [CrossRef]
26.
Joyner, M.J.; Hunter, S.K.; Lucia, A.; Jones, A.M. Last Word on Viewpoint: Physiology and fast marathons.
J. Appl. Physiol. 2020, 128, 1086–1087. [CrossRef] [PubMed]
27.
Joyner, M.J. Modeling: Optimal marathon performance on the basis of physiological factors. J. Appl. Physiol.
1991, 70, 683–687. [CrossRef] [PubMed]
28.
Lucia, A.; Esteve, J.; Olivan, J.; Gallego-Gomez, F. Physiological characteristics of the best Eritrean
runners—Exceptional economy. Appl. Physiol. Nutr. Met. 2006, 31, 1–11. [CrossRef] [PubMed]
29.
Foster, C.; Lucia, A. Running economy: The forgotten factor in elite performance. Sports Med. 2007, 37,
316–319. [CrossRef]
30.
Saunders, P.; Pyne, D.; Telford, R.; Hawley, J. Factors affecting running economy in trained distance runners.
Sports Med. 2004, 34, 465–485. [CrossRef]
31.
Barnes, K.; Kilding, A. Running economy: Measurement, norms, and determining factors. Sports Med. Open
2015, 1, 8. [CrossRef]
32.
Hoogkamer, W.; Kram, R.; Arellano, C.J. How Biomechanical Improvements in Running Economy Could
Break the 2-h Marathon Barrier. Sports Med. 2017, 47, 1739–1750. [CrossRef]
33.
Lee, E.; Snyder, E.; Lundstrom, C. Effects of marathon training in maximal aerobic capacity and running
economy in experienced marathon runners. J. Hum. Sport Exerc. 2020, 15, 79–93.
34.
Barnes, K.R.; Hopkins, W.G.; McGuigan, M.R.; Northuis, M.E.; Kilding, A.E. Effects of Resistance Training on
Running Economy and Cross-country Performance. Med. Sci. Sports Exerc. 2013, 45, 2322–2331. [CrossRef]
35.
Kolbe, T.; Dennis, S.; Selley, E.; Noakes, T.; Lambert, M. The relationship between critical power and running
performance. J. Sports Sci. 1995, 13, 265–269. [CrossRef]
36.
Noakes, T.D.; Myburgh, K.H.; Schall, R. Peak treadmill running velocity during theVO2max test predicts
running performance. J. Sports Sci. 1990, 8, 35–45. [CrossRef] [PubMed]
37.
Tokmakidis, S.P.; Léger, L.A.; Pilianidis, T.C. Failure to obtain a unique threshold on the blood lactate
concentration curve during exercise. Eur. J. Appl. Physiol. Occup. Physiol. 1998, 77, 333–342. [CrossRef]
[PubMed]
38.
Davies, C.T.M.; Thompson, M.W. Aerobic performance of female marathon and male ultramarathon athletes.
Eur. J. Appl. Physiol. Occup. Physiol. 1979, 41, 233–245. [CrossRef] [PubMed]
39.
Maughan, R.J.; Leiper, J.B. Aerobic capacity and fractional utilization of aerobic capacity in elite and non-elite
male and female marathon runners. Eur. J. Appl. Physiol. Occup. Physiol. 1983, 52, 80–87. [CrossRef]
[PubMed]
© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access
article distributed under the terms and conditions of the Creative Commons Attribution
(CC BY) license (http://creativecommons.org/licenses/by/4.0/).
| Physiological and Race Pace Characteristics of Medium and Low-Level Athens Marathon Runners. | 08-21-2020 | Myrkos, Aristides,Smilios, Ilias,Kokkinou, Eleni Maria,Rousopoulos, Evangelos,Douda, Helen | eng |
PMC7085788 | sensors
Article
Muscle Activation in Middle-Distance Athletes with
Compression Stockings
Diego Moreno-Pérez 1, Pedro J. Marín 2, Álvaro López-Samanes 3,*
, Roberto Cejuela-Anta 4 and
Jonathan Esteve-Lanao 5
1
Departament of Education, Research and Evaluation Methods, Comillas Pontifical University, 28015 Madrid,
Spain; [email protected]
2
CYMO Research Institute, 47140 Valladolid, Spain; [email protected]
3
School of Physiotherapy, Faculty of Health Sciences, Universidad Francisco de Vitoria, 28223 Madrid, Spain
4
Department of Physical Education and Sports, University of Alicante, 03690 Alicante, Spain;
[email protected]
5
All in Your Mind Training System, Mérida 97134, Mexico; [email protected]
*
Correspondence: [email protected]; Tel.: +34-91-709-14-00 (ext. 1955)
Received: 18 January 2020; Accepted: 24 February 2020; Published: 26 February 2020
Abstract: The aim of this study was to evaluate changes in electromyographic activity with the
use of gradual compression stockings (GCSs) on middle-distance endurance athletes’ performance,
based on surface electromyography measurement techniques. Sixteen well-trained athletes were
recruited (mean ± SD: age 33.4 ± 6.3 years, VO2max 63.7 ± 6.3 mL·kg−1·min−1, maximal aerobic speed
19.7 ± 1.5 km·h). The athletes were divided into two groups and were assigned in a randomized
order to their respective groups according to their experience with the use of GCSs. Initially, a
maximum oxygen consumption (VO2max) test was performed to standardize the athletes’ running
speeds for subsequent tests. Afterward, electromyographic activity, metabolic, and performance
variables for each group were measured with surface electromyography. In addition, blood lactate
concentration was measured, both with and without GCSs, during 10 min at 3% above VT2 (second
ventilatory threshold), all of which were performed on the track. Next, surface electromyography
activity was measured during a 1 km run at maximum speed. No significant changes were found in
electromyography activity, metabolic and performance variables with GCSs use (p > 0.164) in any
of the variables measured. Overall, there were no performance benefits when using compression
garments against a control condition.
Keywords: surface electromyography; compression garment; gradual-elastic compression stockings;
muscular fatigue; endurance athletes
1. Introduction
Graduated compression stockings (GCSs) are socks that create a compressive pressure around
the muscle, bone, and connective tissue, with this pressure higher in the ankle area and gradually
decreasing until the knee [1]. In addition, compression garments were originally used to treat deep vein
thrombosis [2] and venous insufficiencies [1,3]. Thus, several studies have demonstrated an increase in
the venous velocity, a reduction of venous pooling, and improvement in venous return in hospital
patients who wore GCSs [1,4]. Although there are no significant changes in heart rate associated with
the use of GCSs in endurance events, [5–8] the interest in the sports sciences field in GCS application
and commercialization is highly increasing [9].
A decrease in the concentration of metabolites associated with compression garment use may
have benefits during submaximal running efforts [10]. Berry and McMurray [11] hypothesized that a
Sensors 2020, 20, 1268; doi:10.3390/s20051268
www.mdpi.com/journal/sensors
Sensors 2020, 20, 1268
2 of 8
reduced blood lactate concentration associated with the use of compression stockings could be due
to greater blood flow removal during exercise with CGSs. In addition, it has been described that
GCSs lead to improvements in blood lactate concentration clearance during continuous sports, such as
cycling [12]. However, other studies reported no changes in blood lactate concentration with the use of
GCSs in endurance efforts [6–8,13]. The differences between studies could be attributed to the different
methodologies used [6,7]. Thus, the benefits of using GCSs while running are not entirely clear at the
metabolic and cardiovascular levels. Moreover, the possible improvements in muscle recruitment with
the use of GCSs during dynamic actions are unknown.
One method to measure the fiber recruitment is electromyography (EMG). This type of
measurement technique comprises the sum of the electrical contributions made by the active motor
units (MUs), that are detected by electrodes placed on the skin overlying the muscle. The information
extracted from the surface EMG is often considered a global measure of MU activity because of the
inability of the traditional (two electrode) recording configuration to detect activity at the level of
single MUs, which allows the measurement of the electrical signal during a muscular action [14].
Raez et al. [15] defined EMG as the acquisition, recording, and analysis of electrical activity produced
by nerves and muscles through electrode surface electromyography, which is a noninvasive method
that allows the evaluation of muscle recruitment during dynamic efforts [16]. Most recent research
on compression garments by means of EMG has mainly been focused on the relationship between
intramuscular pressure and EMG responses during concentric isokinetic muscle contractions [17].
Likewise, previous research underlined the relationship between the use of compression garments and
the perception of lower muscle pain [5,18], greater comfort, and a lower subjective perception of effort
(RPE; rating of perceived exertion) [18]. According to Varela-Sanz [13] there seems to be a tendency to
run faster with a lower perception of effort. If there are no clear metabolic or cardiovascular benefits,
the benefits may be found in a change in muscle recruitment between the thigh muscle and the leg (i.e.,
triceps surae).
The aim of this study is to evaluate changes in electromyographic activity with the use of the
use of gradual compression stockings (GCSs) on middle-distance endurance athletes’ performance,
based on surface electromyography measurement techniques. We hypothesize an improvement in
submaximal (i.e., 10 min at 3% above VT2 (second ventilatory threshold)) and maximal conditions (i.e.,
1 km at full speed) with the use of gradual compression stockings compared to control condition.
2. Materials and Methods
2.1. Participants
Fourteen male and two female athletes reported to the laboratory three times with 72 h hours
between protocols (mean ± SD, age 33.2 ± 7.2 years, VO2max 63.7 ± 6.3 mL·kg−1·min−1, maximal
aerobic speed 19.7 ± 1.5 km·h−1, 4 min and 18 s at 1500 mL). All athletes had competed in the
Spanish Track and Field Championships, and some of them had won medals at the National Track
Veterans’ Championships.
Before the beginning of the study, all subjects gave written informed consent in accordance with
the Declaration of Helsinki [19]. The protocol was approved by the Ethics Committee of the University.
The athletes were randomly assigned to either an experimental group, with GCSs (EXP), or control
group without CGSs (CNT). There were no significant changes in the descriptive variables between
groups (p < 0.050).
2.2. Experimental Design
Day 1: A maximum oxygen consumption (VO2max) test was performed in order to define the
subjects’ running speeds for consecutive tests. After a standardized warm-up of 20 min of continuous
running on a treadmill (Technogym Run Race 1400 HC, Gambettola, Italy) at 60% of their maximum
heart rate and a block of dynamic warm-up [20], subjects performed a VO2max test with a gas analyzer
Sensors 2020, 20, 1268
3 of 8
(VO2000, Medical Graphics Corporation, St. Paul, MN, USA). The variables that were measured were
oxygen uptake (VO2), pulmonary ventilation (VE), ventilatory equivalents for oxygen (VE·VO2−1) and
carbon dioxide (VE·CO2 VE·VO2−1), and end-tidal partial pressure of oxygen (PETO2) and carbon
dioxide (PETCO2). VO2max was recorded as the highest VO2 value obtained for any continuous 30 s
period during the test. The VT1 was determined using the criteria of an increase in both VE·VO2−1
and PETO2 with no increase in VE·VCO2−1, whereas the VT2 was determined using the criteria of an
increase in both VE·VO2−1and VE·VCO2−1 and a decrease in PETCO2 [21]. Two independent observers
detected VT1 and VT2. If there was disagreement, a third investigator was consulted. The maximal
aerobic speed was associated with the last completed 30 s stage before the exhaustion, which was
associated with VO2max [21]. The protocol started with a gradient of 1% at a speed of 10 km·h−1, with
increments of 0.3 km·h−1 every 30 s until the maximum exhaustion [21]. The tests were performed in
the Exercise Physiology Laboratory of the Universidad Europea de Madrid (i.e., 600 m altitude). All
evaluations were performed at the same time of day (i.e., evening, between 7:00 p.m. and 9:00 p.m.)
and under similar environmental conditions (i.e., 20–22 ◦C temperature, 60–65% relative humidity) to
avoid effects associated with circadian rhythms on performance [22].
Days 2 and 3: Each group had to perform the same training session with compression garments
(EXP) and without GCSs (CNT), with a recovery period of 72 h between the two sessions. One group
was assigned to use GCSs only on the first day, and the other group was assigned to use GCSs only
on the second day (the athletes served as their own controls). On the day that GCSs were not used,
the athletes used traditional socks. The participants wore GCSs (Medilast Sport, Lleida, Spain) with
degressive pressure (15–20 mm Hg at the ankle; 88% Polyamid, 12% Elasthane) from the ankle to the
calf area (always under the supervision of a member of the investigators’ team.). The compression was
similar to that used in the medical field [23].
2.3. Surface Electromyographic Activity (EMG)
EMG was measured according to the electrical activity (EA) recorded with a telemetric system
(BTS Pocket EMG, Garbagnate M.se, Italy). The information extracted from the surface EMG give
global and, rarely, individual indications of motor units activity [24]. A sampling frequency of 1 kHz
was used. Preamplifiers placed next to the measuring electrode allowed ruling out the influence of
likely movements of the wires on the measurement. Signals from the EMG were band-pass filtered
(10–400 Hz), and the root mean square (RMS) was analyzed. Bipolar surface EMG electrodes (Al/AgCl,
discs of 10 mm diameter) with an inter-electrode distance of 24 mm were placed on the bellies of the
vastus lateralis (VAL), vastus medialis (VAM), rectus femoris (RF), biceps femoris (BF), gastrocnemius
(GAM), and soleus (SOL) in accordance with the Surface EMG for Non-invasive Assessment of
Muscles [25].
We evaluated EMG during two footraces: (a) 10 min at 3% above VT2 (t > VT2) (i.e., represents a
submaximal effort), (b) 1 km at full speed (t1km) (i.e., represents a maximal effort). These runs were
performed on the athletics track with a 3 min break in between. All evaluations were performed at the
same time of day (i.e., evening, between 7:00 p.m. and 9:00 p.m.) and under similar environmental
conditions (i.e., 22–24 ◦C temperature, 55% relative humidity).
2.4. Metabolic, Perceptual and Performance Variables
The concentration of blood lactate concentration (mmol/L−1) was measured at t > VT2 with a
blood lactate analyzer (Lactate Pro Arkray INIC, Amstelveen, NED). The subjective perception of effort
(RPE) was measured using the Borg scale [26]. For the performance variable, a stopwatch was used to
measure the time (min) subjects took to run t1km.
2.5. Statistical Analysis
The data set obtained was analyzed with the SPSS Statistics 19 software (SPSS Inc., Chicago,
IL, USA). T-tests were applied to related samples, both to verify that there were no differences in
Sensors 2020, 20, 1268
4 of 8
matching the subjects and to observe the differences in the sports performance variables. All data were
expressed as mean (M) and standard deviation (SD). Homogeneity of variance was tested with the use
of a Kolmogorov–Smirnov test and Lilliefors correction. The level of statistical significance was set at
p < 0.05. The significance level was set at 0.05. Cohen’s formula for effect size (ES) was used and the
results were based on the following criteria: trivial (0–0.19), small (0.20–0.49), medium (0.50–0.79), and
large (0.80 and greater) [27].
3. Results
3.1. Metabolic and Perceptual Variables at Submaximal Efforts (t > VT2)
According to the metabolic demands and perceptual variables, no statistical differences were
founded in the different conditions measured in the study between GCSs and CNT conditions, such as
heart rate (182.6 ± 10.1 versus 182.6 ± 10.0, p = 1.000, ES < 0.01, trivial), blood lactate concentration
(mmol·L−1) (8.3 ± 2.1 versus 7.9 ± 2.4, p = 0.476, ES = 0.20, small), and RPE (8.5 ± 1.0 versus 9.0 ± 0.6,
p = 0.301, ES = 0.33, small).
3.2. Perceptual and Performance Variables at Maximal Effort (t1km)
Perceptual and performance variables did not reach statistical significance during t1km—GCS
versus CNT: RPE (9.9 ± 0.3 versus 10.0 ± 0.0; p = 0.164; ES = 0.39, small), speed (19.2 ± 1.7 versus 19.1 ± 1.7
km·h−1; p = 0.847; ES = 0.00, trivial), % maximal aerobic speed (97.2 ± 3.0 versus 97.0 ± 3.6 km·h−1;
p = 0.823; ES = 0.00, trivial).
3.3. Surface Electromyography
Muscular activity did not any reach statistical significance (Table 1). However, according to effect
sizes the electromyographic activity was greater in the calf musculature when not using GCSs while
running at submaximal effort, while descriptive changes were observed in effect size (ES). EA was
lower in the leg during submaximal efforts (GAM and SOL, ES = 0.10, trivial) compared to in the thigh
(VAL, VAM, BF, and RF, ES = 0.24, small) with GCS use versus CNT; leg (ES = 0.25, small) and leg
(ES = 0.25, small).
Table 1. EMG (electromyography) variables base on wearing or not wearing a graduate compression
garment during the t > Vt2 and t1km.
EMGrms (mV)
GCSs
CNT
t > VT2
t1km
t > VT2
t1km
BF
0.320
±
0.181
0.314
±
0.196
0.323
±
0.104
0.304
±
0.304
RF
0.199
±
0.058
0.194
±
0.081
0.220
±
0.085
0.192
±
0.094
VAL
0.176
±
0.075
0.168
±
0.087
0.195
±
0.086
0.189
±
0.088
VAM
0.223
±
0.053
0.212
±
0.077
0.212
±
0.077
0.235
±
0.064
SOL
0.196
±
0.116
0.164
±
0.062
0.296
±
0.115
0.295
±
0.141
GAM
0.388
±
0.289
0.331
±
0.257
0.268
±
0.101
0.233
±
0.092
Abbreviations: vastus lateralis (VAL), vastus medialis (VAM), rectus femoris (RF), biceps femoris (BF), gastrocnemius
(GAM), and soleus (SOL).
During running maximal effort (i.e., t1km), EA was higher in the leg (GAM and SOL, ES = 0.24,
small) compared to in the thigh (VAL, VAM, BF, and RF, ES = 0.11, trivial) with GCS use versus CNT:
thigh (ES = 0.17, trivial) and leg (ES = 0.18, trivial). (Table 1)
Sensors 2020, 20, 1268
5 of 8
4. Discussion
The aim of this study was to evaluate changes in electromyographic activity with the use of the
use of gradual compression stockings (GCSs) on endurance athletes’ performance, with the use of
electromyography techniques. At the metabolic domain, our results are consistent with previous
studies, where GCS use did not affect blood lactate concentration measurement [6–8,13]. Thus,
performance variables during t1km reported no benefits from GCS use with regards to the time
required to run the specified distance, and our data are in agreement with studies previously published
by Ali [5,7], in which no differences were found in a 10 km test. Therefore, in our study there were no
significant changes in the perceptual variables with GCS use, and small effect sizes were found for
GCS use during submaximal efforts (t > VT2; ES = 0.33) and maximal efforts (t1km; ES = 0.39). This is
consistent with previous studies, where lower muscle pain [5,18], greater comfort [6,7], and a lower
subjective perception of effort [13,18,28] were observed.
With regard to electromyography activity, it was evaluated by means of changes in amplitude
(electromyography amplitude), since it has been described as a variable that provides knowledge
on the degree of muscle fatigue [29]. No significant changes were found between treatments when
measuring EA by surface electromyography, null and small effect sizes were observed (ES = 0.11–0.25),
so that muscle activation changed according to whether CGSs were used or not. Thus, in submaximal
efforts (t > VT2), the activation of the calf muscles (GAM, SOL) was observed to be less than that of the
thigh muscles (VAL, VAM, RF, BF). This would have advantages for performance, as fatigue in the
leg muscles frequently tends to limit performance more than that of the thigh muscles, because the
gastrocnemius and soleus are the greatest contributors to propulsion and support during submaximal
running [30]. According to our data, Lucas-Cuevas AG [31] reported a decrease in the muscular
contribution of the GAM using GCSs in the rest situation and at the beginning of submaximal effort in
the running race. However, one limitation of this study was that they did not analyze the possible
changes in muscle recruitment between the thigh and muscular fatigue between GAM and SOL that
we measured in our study.
During maximal efforts (t1km), muscle recruitment differs from that of submaximal efforts
(t > VT2). The use of GCSs reduces electromyographic activity in the thigh and increases in the
gastrocnemius muscles (sum GAM and SOL). If we analyze EA by muscle group in our study during
the maximum effort of race (t1km), there was less recruitment using GCSs in both the BF and the
RF than where there was no use of GCSs. This could be beneficial since in foot race efforts—as the
speed increases, the RF and BF are the muscle groups that increase their muscular contribution the
most [32,33].
During the present study, the study subjects realized a protocol for detecting electromyography
activity performance, and effort perception variables during two different endurance tests. There are
a few limitations that need to be addressed. First, it could be that the different tests used could not
represent the endurance effort made by the athletes in a race. Thus, it could be necessary to gather more
studies. Also, the individual data were variable, and the sample size was small and medium. Therefore,
the lack of statistical significance could be due to a Type II error. Secondly, quantification of muscle
activity from surface EMG signals is problematic when movement is involved and motion artifacts
and other electromagnetic noises may influence the signal levels. We tried to minimize disturbances
by the applied signal processing and filtering routines; still, artifacts may have small impacts on the
derived maximum muscle activations. Thirdly, we could not use a multi-channel approach to provide
access to a set of physiologically relevant variables on the global muscle level or on the level of single
motor units, opening new fronts for the study of muscle fatigue; however, in the present study, we did
not have this electromyographic analysis technology.
From our point of view, the various findings reported on previous studies with CGSs may be due
to the different gradual compression socks used [5–8,12]. Other authors did not specify which type of
GCSs they used in their studies; most likely they used stockings with a uniform degree of compression.
Sensors 2020, 20, 1268
6 of 8
All the studies conducted to date assessed the potential benefits of using compression means
(GCS) during the efforts in the race on metabolic or cardiovascular variables. According to a study
developed by Varela-Sanz et al. [13], there seems to be a tendency to run faster with a lower perception
of effort. If there are no clear metabolic or cardiovascular benefits, the benefits can be found in a
change in muscle recruitment between the thigh and leg muscle (i.e., triceps surae). Only a previous
study developed by Lucas Cuevas et al. [31] analyzed muscle fatigue using GCSs by EMG techniques.
However, Lucas Cuevas et al. [31] did not compare the possible recruitment changes between the
muscles of the thigh; the study was realized with lower performance level athletes and lower intensities
(75% maximal speed). The novelty of our study is the analysis of EMG recruitment on thigh muscle in
high-level athletes at submaximal/maximal intensities.
5. Conclusions
Endurance athletes perform much of their training and competitions at submaximal intensities
Contrary to our initial hypothesis, no differences were reported on any of the variables analyzed on
this study in EMG recruitment in well-trained athletes between the different conditions (GCSs versus
CNT) at submaximal/maximal efforts. Future studies should be developed based on research findings
to confirm our data or even explore other possibilities, such as the analysis of EMG recruitment during
the recovery process, that are essential for performance in high-level athletes.
Author Contributions: Conceptualization, D.M.-P. and J.E.-L.; methodology, D.M.-P. and J.E.-L.; software, P.J.M.;
validation, D.M.-P., P.J.M. and J.E.-L.; formal analysis, D.M.-P. and P.J.M.; investigation, D.M.-P., R.C.-A. and
J.E.-L.;; resources, Á.L.-S.; data curation, D.M.-P., Á.L.-S.; writing—original draft preparation, D.M., Á.L.-S. and
J.E.-L.; writing—review and editing, D.M., P.J.M., Á.L.-S., R.C.-A., and J.E.-L.; visualization, D.M.-P. and J.E.-L.;
supervision, P.J.M. and J.E.-L.; project administration, D.M.-P. and J.E.-L.; funding acquisition, D.M.-P. All authors
have read and agreed to the published version of the manuscript.
Funding: This research received no external funding.
Acknowledgments: We would like to thank the “Allinyourmind training group” for their participation in this
project. We also acknowledge Medilast S. A. (Lleida, Spain), who supplied GCSs for the study participants. The
authors of this study contacted the brand Medilast Sport (Lleida, Spain) with the sole purpose of supplying sports
garments to carry out the investigation.
Conflicts of Interest: The authors declare no conflict of interest.
References
1.
Lawrence, D.; Kakkar, V.V. Graduated, static, external compression of the lower limb: A physiological
assessment. Br. J. Surg. 1980, 67, 119–121. [CrossRef]
2.
Byrne, B. Deep vein thrombosis prophylaxis: The effectiveness and implications of using below-knee or
thigh-length graduated compression stockings. Heart Lung 2001, 30, 277–284. [CrossRef]
3.
Agu, O.; Baker, D.; Seifalian, A.M. Effect of graduated compression stockings on limb oxygenation and
venous function during exercise in patients with venous insufficiency. Vascular 2004, 12, 69–76. [CrossRef]
4.
O’Donnell, T.F., Jr.; Rosenthal, D.A.; Callow, A.D.; Ledig, B.L. Effect of elastic compression on venous
hemodynamics in postphlebitic limbs. JAMA 1979, 242, 2766–2768. [CrossRef]
5.
Ali, A.; Caine, M.P.; Snow, B.G. Graduated compression stockings: Physiological and perceptual responses
during and after exercise. J. Sports Sci. 2007, 25, 413–419. [CrossRef]
6.
Ali, A.; Creasy, R.H.; Edge, J.A. Physiological effects of wearing graduated compression stockings during
running. Eur. J. Appl. Physiol. 2010, 109, 1017–1025. [CrossRef]
7.
Ali, A.; Creasy, R.H.; Edge, J.A. The effect of graduated compression stockings on running performance.
J. trength Cond. Res. 2011, 25, 1385–1392. [CrossRef]
8.
De Glanville, K.M.; Hamlin, M.J. Positive effect of lower body compression garments on subsequent 40-kM
cycling time trial performance. J. Strength Cond. Res. 2012, 26, 480–486. [CrossRef]
9.
Gill, N.D.; Beaven, C.M.; Cook, C. Effectiveness of post-match recovery strategies in rugby players. Br. J.
Sports Med. 2006, 40, 260–263. [CrossRef]
Sensors 2020, 20, 1268
7 of 8
10.
Bangsbo, J.; Madsen, K.; Kiens, B.; Richter, E.A. Effect of muscle acidity on muscle metabolism and fatigue
during intense exercise in man. J. Physiol. 1996, 495 Pt 2, 587–596. [CrossRef]
11.
Berry, M.J.; McMurray, R.G. Effects of graduated compression stockings on blood lactate following an
exhaustive bout of exercise. Am. J. Phys. Med. 1987, 66, 121–132. [CrossRef] [PubMed]
12.
Chatard, J.C.; Atlaoui, D.; Farjanel, J.; Louisy, F.; Rastel, D.; Guezennec, C.Y. Elastic stockings, performance
and leg pain recovery in 63-year-old sportsmen. Eur. J. Appl. Physiol. 2004, 93, 347–352. [CrossRef]
13.
Varela-Sanz, A.; Espana, J.; Carr, N.; Boullosa, D.A.; Esteve-Lanao, J. Effects of gradual-elastic compression
stockings on running economy, kinematics, and performance in runners. J. Strength Cond. Res. 2011, 25,
2902–2910. [CrossRef]
14.
Farina, D.; Merletti, R.; Enoka, R.M. The extraction of neural strategies from the surface EMG. J. Appl. Physiol.
2004, 96, 1486–1495. [CrossRef]
15.
Raez, M.B.; Hussain, M.S.; Mohd-Yasin, F. Techniques of EMG signal analysis: Detection, processing,
classification and applications. Biol. Proced. Online 2006, 8, 11–35. [CrossRef]
16.
Drost, G.; Verrips, A.; van Engelen, B.G.; Stegeman, D.F.; Zwarts, M.J. Involuntary painful muscle contractions
in Satoyoshi syndrome: A surface electromyographic study. Mov. Disord. 2006, 21, 2015–2018. [CrossRef]
17.
Fu, W.; Liu, Y.; Zhang, S.; Xiong, X.; Wei, S. Effects of local elastic compression on muscle strength,
electromyographic, and mechanomyographic responses in the lower extremity. J. Electromyogr. Kinesiol.
2012, 22, 44–50. [CrossRef]
18.
Rugg, S.; Sternlicht, E. The effect of graduated compression tights, compared with running shorts, on
counter movement jump performance before and after submaximal running. J. Strength Cond. Res. 2013, 27,
1067–1073. [CrossRef]
19.
Holt, G.R. Declaration of Helsinki-the world’s document of conscience and responsibility. South. Med J. 2014,
107, 407. [CrossRef]
20.
Ayala, F.; Moreno-Perez, V.; Vera-Garcia, F.J.; Moya, M.; Sanz-Rivas, D.; Fernandez-Fernandez, J. Acute and
Time-Course Effects of Traditional and Dynamic Warm-Up Routines in Young Elite Junior Tennis Players.
PLoS ONE 2016, 11, e0152790. [CrossRef]
21.
Esteve-Lanao, J.; Foster, C.; Seiler, S.; Lucia, A. Impact of training intensity distribution on performance in
endurance athletes. J. Strength Cond. Res. 2007, 21, 943–949. [CrossRef]
22.
Lopez-Samanes, A.; Moreno-Perez, D.; Mate-Munoz, J.L.; Dominguez, R.; Pallares, J.G.; Mora-Rodriguez, R.;
Ortega, J.F. Circadian rhythm effect on physical tennis performance in trained male players. J. Sports Sci.
2017, 35, 2121–2128. [CrossRef]
23.
Kemmler, W.; von Stengel, S.; Kockritz, C.; Mayhew, J.; Wassermann, A.; Zapf, J. Effect of compression
stockings on running performance in men runners. J. Strength Cond. Res. 2009, 23, 101–105. [CrossRef]
24.
Kamavuako, E.N.; Rosenvang, J.C.; Horup, R.; Jensen, W.; Farina, D.; Englehart, K.B. Surface versus
untargeted intramuscular EMG based classification of simultaneous and dynamically changing movements.
IEEE Trans. Neural Syst. Rehabil. Eng. 2013, 21, 992–998. [CrossRef]
25.
Hermens, H.J.; Freriks, B.; Disselhorst-Klug, C.; Rau, G. Development of recommendations for SEMG sensors
and sensor placement procedures. J. Electromyogr. Kinesiol. 2000, 10, 361–374. [CrossRef]
26.
Borg, G.A. Psychophysical bases of perceived exertion. Med. Sci. Sports Exerc. 1982, 14, 377–381. [CrossRef]
27.
Cohen, J. A power primer. Psychol. Bull. 1992, 112, 155–159. [CrossRef]
28.
Goh, S.S.; Laursen, P.B.; Dascombe, B.; Nosaka, K. Effect of lower body compression garments on submaximal
and maximal running performance in cold (10 degrees C) and hot (32 degrees C) environments. Eur. J.
Appl. Physiol. 2011, 111, 819–826. [CrossRef]
29.
Viitasalo, J.H.; Komi, P.V. Signal characteristics of EMG during fatigue. Eur. J. Appl. Physiol. Occup. Physiol.
1977, 37, 111–121. [CrossRef]
30.
Hamner, S.R.; Seth, A.; Delp, S.L. Muscle contributions to propulsion and support during running. J. Biomech.
2010, 43, 2709–2716. [CrossRef]
31.
Lucas-Cuevas, A.G.; Priego Quesada, J.I.; Gimenez, J.V.; Aparicio, I.; Cortell-Tormo, J.M.; Perez-Soriano, P.
Can Graduated Compressive Stockings Reduce Muscle Activity During Running? Res. Q. Exerc. Sport 2017,
88, 223–229. [CrossRef]
Sensors 2020, 20, 1268
8 of 8
32.
Albertus-Kajee, Y.; Tucker, R.; Derman, W.; Lamberts, R.P.; Lambert, M.I. Alternative methods of normalising
EMG during running. J. Electromyogr. Kinesiol. 2011, 21, 579–586. [CrossRef] [PubMed]
33.
Rimaud, D.; Messonnier, L.; Castells, J.; Devillard, X.; Calmels, P. Effects of compression stockings during
exercise and recovery on blood lactate kinetics. Eur. J. Appl. Physiol. 2010, 110, 425–433. [CrossRef] [PubMed]
© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access
article distributed under the terms and conditions of the Creative Commons Attribution
(CC BY) license (http://creativecommons.org/licenses/by/4.0/).
| Muscle Activation in Middle-Distance Athletes with Compression Stockings. | 02-26-2020 | Moreno-Pérez, Diego,Marín, Pedro J,López-Samanes, Álvaro,Cejuela, Roberto,Esteve-Lanao, Jonathan | eng |
PMC8540834 | sensors
Article
The Influence of Time Winning and Time Losing on
Position-Specific Match Physical Demands in the Top One
Spanish Soccer League
José C. Ponce-Bordón 1
, Jesús Díaz-García 1,*
, Miguel A. López-Gajardo 1
, David Lobo-Triviño 1,
Roberto López del Campo 2
, Ricardo Resta 2
and Tomás García-Calvo 1
Citation: Ponce-Bordón, J.C.;
Díaz-García, J.; López-Gajardo, M.A.;
Lobo-Triviño, D.; López del Campo,
R.; Resta, R.; García-Calvo, T. The
Influence of Time Winning and Time
Losing on Position-Specific Match
Physical Demands in the Top One
Spanish Soccer League. Sensors 2021,
21, 6843. https://doi.org/10.3390/
s21206843
Academic Editor: Angelo Maria
Sabatini
Received: 17 September 2021
Accepted: 11 October 2021
Published: 14 October 2021
Publisher’s Note: MDPI stays neutral
with regard to jurisdictional claims in
published maps and institutional affil-
iations.
Copyright: © 2021 by the authors.
Licensee MDPI, Basel, Switzerland.
This article is an open access article
distributed
under
the
terms
and
conditions of the Creative Commons
Attribution (CC BY) license (https://
creativecommons.org/licenses/by/
4.0/).
1
Faculty of Sport Sciences, University of Extremadura, Boulevard of the University s/n, 10003 Caceres, Spain;
[email protected] (J.C.P.-B.); [email protected] (M.A.L.-G.); [email protected] (D.L.-T.);
[email protected] (T.G.-C.)
2
LaLiga Sport Research Section, 28043 Madrid, Spain; [email protected] (R.L.d.C.); [email protected] (R.R.)
*
Correspondence: [email protected]; Tel.: +31-927-257-460
Abstract: The aim of the present study was to analyze the influence of time winning and time losing
on position-specific match physical demands with and without ball possession in the top Spanish
professional soccer league. All matches played in the First Spanish soccer league over four consecutive
seasons (from 2015/16 to 2018/19) were recorded using an optical tracking system (i.e., ChyronHego),
and the data were analyzed via Mediacoach®. Total distance (TD), and TD > 21 km·h−1 covered
with and without ball possession were analyzed using a Linear Mixed Model, taking into account
the contextual variables time winning and losing. Results showed that TD and TD > 21 km·h−1
covered by central midfielders (0.01 and 0.005 m/min, respectively), wide midfielders (0.02 and
0.01 m/min, respectively), and forwards (0.03 and 0.02 m/min, respectively) significantly increased
while winning (p < 0.05). By contrast, TD and TD > 21 km·h−1 covered by central defenders
(0.01 and 0.008 m/min, respectively) and wide defenders (0.06 and 0.008 m/min, respectively)
significantly increased while losing (p < 0.05). In addition, for each minute that teams were winning,
total distance with ball possession (TDWP) decreased, while, for each minute that teams were
losing, TDWP increased. Instead, TDWP > 21 km·h−1 obtained opposite results. Total distance
without ball possession increased when teams were winning, and decreased when teams were losing.
Therefore, the evolution of scoreline significantly influences tactical–technical and physical demands
on soccer matches.
Keywords: contextual variables; match running performance; ball possession; positional; professional
soccer
1. Introduction
Context-related variables are considered the most influencing variables on match
physical demands in soccer [1]. Time–motion analysis research has reported a large amount
of information about context-related variables such as match status, match location, and
opponent level [2]. Specifically, it has been previously shown that final and partial match
status (analyzed by epochs of time—i.e., 15-min periods or half-time) modify match physi-
cal demands as well as ball possession [3,4]. However, this method of analysis is probably
to do with the interaction of other variables such as the evolution of the match-scoreline [5].
Regardless of final match status, the time each team was leading, drawing, or trailing
during a match could be different, with matches taking place where a team has won
throughout 70 min (i.e., team A scored a goal in the 20th min and team B did not score) or
matches where a team has won throughout 1 min (i.e., teams were drawing over the match,
and one team scored a goal in the 89th min). Moreover, it is possible that one team that was
ahead for a long time would end up losing on the final scoreline. Match physical demands
Sensors 2021, 21, 6843. https://doi.org/10.3390/s21206843
https://www.mdpi.com/journal/sensors
Sensors 2021, 21, 6843
2 of 9
are believed to depend on evolving scoreline (i.e., whether a team is winning or losing)
since, when a team is losing, players try to reach their maximal physical capacity in order to
draw or win the match [2]. Therefore, this limitation could be solved by taking into account
the minutes that teams were ahead and behind during a match separately. According to
our knowledge, the influence of time winning and time losing on position-specific match
physical demands according to the evolution of the scoreline is less known.
Match status has been arguably analyzed enough to prove that it influences soccer
teams’ behavior [6]. In this vein, Lago–Peñas [7] reported that losing teams frequently
increase their percentage of possession; meanwhile, certain winning teams preferred
counterattacking or playing directly. In addition, match status clearly impacts teams
playing style [5], and both variables (i.e., match status and playing style) also influence
match physical demands; however, several studies about this topic have drawn the opposite
conclusions. For instance, elite Spanish soccer players performed less high-intensity
distance (19 km·h−1) when winning than when they were losing, since winning is a
comfortable status, it is possible that players assume a ball contention strategy, keeping the
game slower, which results in lower speeds [8]. Moreover, Castellano et al. [2] analyzed
one Spanish team of LaLiga and they showed that the distances covered at high intensity
by the reference team were greater when the result was adverse. Moalla et al. [3] obtained
the same results in a study from the Stars League during 2013/14 and 2014/15 seasons.
Conversely, during qualifying round matches of the World Cup 2010, drawing teams
covered a lower average speed than the winning and losing teams [9]. In this line, variables
that determine the intensity of the game (maximum speed and frequency of high-intensity
activity) in a professional Brazilian football team were significantly lower when the team
lost [10], meanwhile greater intensity running distances were observed in matches that the
team won as opposed to losing [11]. Therefore, coaches should take into account the match
status to analyze the external load implied by the match [12].
However, soccer players could have not been affected by this contextual variable in
the same way due to playing style, since playing style changes associated with match status
could affect different players differently in ways that are position-specific [13]. For example,
in the German Bundesliga during the 2014/15 season, central defenders and full-backs
covered shorter distances at high intensity in won matches than in lost matches (p < 0.01);
however, forwards covered significantly longer total distance in won matches than in
drawn and lost matches (p < 0.05) [14]. Despite these conclusions, position-specific match
physical demands can also vary depending on the evolution of scoreline status. In this vein,
when a team was winning, during preseason matches of the 2011/12 season Australian
League soccer, the average speed was 4.17% lower than when the team was drawing
(p < 0.05) [15]. Regarding player position-specific data, in the English Premier League,
Redwood-Brown et al. [16] found midfielders covered more distance at high intensity when
level, defenders more when losing, and attackers more when winning. Similarly, losing
status increased the total distance covered by defenders from Spanish First soccer league,
while attacking players showed the opposite trend [17]. Therefore, these previous studies
have suggested that, due to the player position, players perform different tactical roles,
and match status and the associated playing style changes could have different influences
on match running performance by positions.
The knowledge about the influence of time that teams were ahead or behind on
position-specific match physical demands could have important practical applications
during the competitive season to program the training load in a more strategic way based
on physical data [12]. In addition, less is known about the influence of time which teams
spend winning or losing on position-specific match physical demands. Therefore, the main
objective of the present study was to analyze the influence of time winning and losing
on position-specific match physical demands in the top Spanish soccer league across four
seasons (2015/2016–2018/2019). As a secondary objective, the study also aimed to analyze
the match physical demands with and without ball possession according to time winning
and time losing.
Sensors 2021, 21, 6843
3 of 9
Based on previous findings obtained by the aforementioned studies, the following
hypotheses were proposed. Firstly, concerning match physical demands with and without
ball possession, it was expected that total distance with ball possession would be less
during time winning [7]. Secondly, concerning position-specific match physical demands,
it was expected that total distance would be greater in attackers during time winning and
defenders during time losing, based on previous results [16,17].
2. Materials and Methods
2.1. Participants
The sample comprised 36,883 individual match observations of 1037 professional
soccer players who competed in the First Spanish professional soccer league (i.e., LaLiga
Santander) over four consecutive seasons (from 2015/16 to 2018/19). All players who
participated in matches (starters and non-starters) and played 10 min at least were included.
Only goalkeepers were excluded. According to previous studies [18], players were divided
into five position-specific groups: Central Defenders (CD; n = 6787 observations), Wide
Defenders (WD; n = 6530 observations); Central Midfielders (CM; n = 6826 observations);
Wide Midfielders (WM; n = 8394 observations); Forwards (FW; n = 8346 observations).
Data were provided to the authors by LaLigaTM, and the study received ethical approval
from the University of Extremadura; Vice-Rectorate of Research, Transfer and Innovation-
Delegation of the Bioethics and Biosafety Commission (Protocol number: 239/2019).
2.2. Procedure and Variables
Match physical demands data were collected by an optical tracking system
(ChyronHego®; TRACAB, New York, NY, USA). This multi-camera tracking system con-
sists of 8 different super 4K-High Dynamic Range cameras situated strategically to follow
and track the 22 players on the field throughout the match. These cameras film from several
angles and analyze X and Y coordinates of each player, providing real-time tracking with
data recorded at 25 Hz. Mediacoach® is also based on data correction of the semi-automatic
video technology (the manual part of the process) [19]. The validity and reliability of the
Tracab® video tracking system has been analyzed, reporting average measurement errors
of 2% for total distance covered [20–22].
The physical performance variables used for this study were categorized according
to the ball possession as follows [23,24]: with possession (WP) and without possession
(WOP). The following variables were studied for each of these categories: total distance
(m) covered by players (i.e., TD) and total distance covered at more than 21 km·h−1 (i.e.,
TD > 21 km·h−1).
To determine if the scoreline influenced position-specific match physical demands, the
cumulative time that each team was losing or winning during a match was included in the
analysis (not the final match result). For example, if team A scored a goal in the 20th min
and team B equalized in the final minute, team A was classified as losing for 0 min and
winning for 70 min, while team B was classified as losing for 70 min and winning for 0 min
2.3. Data Analysis
All statistical analyses were performed using R-studio [25]. A Linear Mixed Model
(LMM) was conducted for each of the physical variables using the lme4 package [26]. This
model allows for the analyzing of data with a hierarchical structure in nested units and
has demonstrated its ability to cope with unbalanced and repeated-measures data [27]. For
example, variables related to the distance covered in matches are nested for players (i.e.,
each player has a record for every match they have participated in), and players are nested
into teams. Also, cumulative times spent winning or losing are nested into matches and
these matches can also be nested into teams. This represents a threefold levels structure,
where teams are the topmost unit in the hierarchy.
A general multilevel-modelling strategy was applied [27], where fixed and random
effects had been included in different steps from the simplest to the most complex. First,
Sensors 2021, 21, 6843
4 of 9
unconditional models were analyzed exclusively including dependent variables (i.e., dis-
tance variables) to check if the grouping variables at levels 2 and 3 (i.e., players and teams)
significantly affected the intercept (mean) of each dependent variable. These models may
be used as baselines for comparing more complex models. Later, different models were
performed for each of the dependent variables, setting as fixed effects the position of
the players and the time winning/losing. Following the procedure proposed by Heck &
Thomas [27], models with different random effects (intercepts and slope) were created
for each variable. A model comparison for each step was performed using the Akaike
Information Criterion (AIC) [28] and a chi-square likelihood ratio test [29], where a lower
value represented a fitter model. Final models presented in Tables 1 and 2 (with random
intercept and slope effect) were chosen according to better values of AIC, log-likelihood,
and significant effect of variables. Maximum Likelihood (ML) estimation for model com-
parison and for the final model of each physical variable was used, the best model, again,
using Restricted Maximum Likelihood (REML) estimation, was refitted. Marginal and
conditional R2 metrics [30] for each LMM to provide some measure of effect sizes were
reported. Significance level was set at p < 0.05.
For a suitable interpretation of the results, the time winning/losing was group-mean
centered, being centered to the team’s mean in each season.
Table 1. Differences by ball possession of position-specific total distance covered according to the scoreline evolution.
CD
WD
CM
WM
FW
TD
(m/min)
Intercept
107.30
b, c, d, e
109.90
a, c, d, e
116.10
a, b
115.90
a, b
115.60
a, b, c
Slope Time Winning
−0.006
c, d, e
−0.005
c, d, e
0.01
a, b, d, e
0.02
a, b, c, e
0.03
a, b, c, d
Slope Time Lossing
0.01
c, d, e
0.006
c, d, e
−0.003
a, b, d, e
−0.02
a, b, c, e
−0.02
a, b, c, d
TDWP
(m/min)
Intercept
35.93
b, c, d, e
38.41
a, c, d, e
42.21
a, b, e
42.54
a, b, e
43.04
a, b, c, d
Slope Time Winning
−0.03
−0.04
e
−0.04
e
−0.04
−0.03
b, c
Slope Time Lossing
0.03
0.03
0.03
e
0.02
0.02
c
TDWOP
(m/min)
Intercept
42.50
b, c, d
43.80
a, c, d, e
47.39
a, b, d, e
45.36
a, b, c, e
42.16
b, c, d
Slope Time Winning
0.01
c, d, e
0.02
d, e
0.03
a, d, e
0.04
a, b, c
0.01
a, b, c
Slope Time Lossing
0.002
d, e
−0.003
d, e
−0.003
d, e
−0.01
a, b, c
−0.02
a, b, c
Note. CD = Central defenders; WD = Wide defenders; CM = Central midfielders; WM = Wide midfielders; FW = Forwards; TD = Total
distance; TDWP = Total distance with ball possession; TDWOP = Total distance without ball possession; a = significant differences compared
to central defenders; b = significant differences compared to wide defenders; c = significant differences compared to central midfielders;
d = significant differences compared to wide midfielders; e = significant differences compared to forwards.
Sensors 2021, 21, 6843
5 of 9
Table 2. Differences by ball possession of position-specific total distance covered at more than 21 km·h−1 according to the
scoreline evolution.
CD
WD
CM
WM
FW
Total distance > 21 km·h−1
(m/min)
Intercept
5.74
b, c, d, e
6.68
a, d, e
6.68
a, d, e
7.24
a, b, c, e
7.57
a, b, c, d
Slope Time Winning
−0.008
c, d, e
−0.005
c, d, e
0.005
a, b, d, e
0.01
a, b, c, e
0.02
a, b, c, d
Slope Time Lossing
0.008
c, d, e
0.008
c, d, e
−0.004
a, b, d, e
−0.001
a, b, c, e
−0.01
a, b, c, d
Total distance with ball
possession > 21 km·h−1
(m/min)
Intercept
2.21
b, c, d, e
2.81
a, c, d, e
3.14
a, b, d, e
3.57
a, b, c, e
4.01
a, b, c, d
Slope Time Winning
0.001
c, d, e
−0.001
c, d, e
0.005
a, b, e
0.007
a, b, e
0.01
a, b, c, d
Slope Time Lossing
−0.001
a, c, d, e
0.001
a, b, d, e
−0.006
a, b, c
−0.009
a, b, c
−0.009
a, c, d, e
Total distance without ball
possession > 21 km·h−1
(m/min)
Intercept
3.54
b, c, d, e
3.80
a, c, d, e
3.37
a, b, e
3.40
a, b, e
3.16
a, b, c, d
Slope Time Winning
−0.008
b, c, d, e
−0.005
a, c, d, e
−0.001
a, b, d, e
0.002
a, b, c
0.004
a, b, c
Slope Time Lossing
−0.009
b, c, d, e
0.007
a, c, d, e
0.002
a, b, d, e
−0.001
a, b, c, e
−0.002
a, b, c
Notes. CD = Central defenders; WD = Wide defenders; CM = Central midfielders; WM = Wide midfielders; FW = Forwards; a = significant
differences compared to central defenders; b = significant differences compared to wide defenders; c = significant differences compared to
central midfielders; d = significant differences compared to wide midfielders; e = significant differences compared to forwards.
3. Results
Firstly, the Wald test and intraclass correlation coefficient (ICC) suggested statistically
significant variability in the distances covered by players according to time winning and
losing (ICC > 0.10); therefore, LMM was justified for the purpose of the study. Also, AIC
suggested that the twofold levels model was the one fitter for this purpose.
Secondly, Table 1 shows the differences of TD covered according to ball possession
and to the scoreline evolution by player positions. Regardless of scoreline, CM covered
significantly greater TD than the rest of the players (p < 0.05). TD covered by CD and WD
decreased significantly with respect to CM, WM, and FW (p < 0.05) for each minute that
teams were ahead. By contrast, for each minute that teams were trailing, TD covered by
CD and WD increased significantly with respect to CM, WM, and FW (p < 0.05).
During the match, FW covered significantly greater TDWP than CD, WD, CM, and
WM (p < 0.05). However, for each minute that teams were ahead, TDWP decreased for
all positions. Significant differences were found between WD and CM with respect to FW
(p < 0.01). On the contrary, for each minute that teams were trailing, TDWP increased for
all positions. Significant differences between CM and FW were observed (p < 0.05).
On the other hand, CM covered TDWOP significantly greater than the rest of the
players (p < 0.05). However, for each minute that teams were ahead, TDWOP increased for
all positions. CM and WM significantly increased TDWOP with respect to CD, WD, and
FW (p < 0.01). By contrast, for each minute that teams were trailing, TDWOP significantly
decreased for all positions, except CD (p < 0.05).
Thirdly, Table 2 shows the differences by ball possession of TD > 21 km·h−1 according
to the scoreline evolution by player positions. Regardless of the scoreline, FW covered
TD > 21 km·h−1 significantly greater than the rest of the players (p < 0.05). However, for
each minute that teams were ahead, CD and WD significantly decreased TD > 21 km·h−1
with respect to CM, WM, and FW (p < 0.05). By contrast, for each minute that teams were
trailing, CD and WD significantly increased TD > 21 km·h−1 with respect to CM, WM, and
FW (p < 0.05).
During the match, FW covered TD > 21 km·h−1 with ball possession significantly
greater than CD, WD, CM, and WM (p < 0.05). Moreover, for each minute that teams were
Sensors 2021, 21, 6843
6 of 9
ahead, TD > 21 km·h−1 significantly increased for all positions, except WD (p < 0.05). By
contrast, for each minute that teams were trailing, TD > 21 km·h−1 significantly decreased
for all positions, except WD (p < 0.05).
Finally, WD covered significantly greater TD > 21 km·h−1 without ball possession than
the rest of the players (p < 0.01). For each minute that teams were ahead, TD > 21 km·h−1
without ball possession covered by WM and FW increased significantly, with respect to
CD, WD, and CM (p < 0.05). Likewise, for each minute that teams were trailing, WD and
CM significantly increased TD > 21 km·h−1 without ball possession, with respect to CD,
WD and FW (p < 0.05).
4. Discussion
The aim of the present study was to analyze the influence of time winning and losing
on position-specific match physical demands in the top Spanish professional soccer league
across four seasons (from 2015/2016 to 2018/2019). Subsequently, match physical demands
with and without ball possession according to time winning and losing were examined.
The main results showed that TDWP was less while teams were winning, while it was
greater while teams were losing. In addition, TDWOP increased while teams were winning,
while it decreased while teams were losing. Finally, TD and TD > 21 km·h−1 covered by
CM, WD, and FW were greater while teams were winning, while TD and TD > 21 km·h−1
covered by CD and WD were greater while teams were losing.
Firstly, it had been hypothesized that TDWP would be less during time winning
(Hypothesis 1). Our results showed that TDWP decreased during time winning for all player
positions and increased during time losing, therefore Hypothesis 1 was confirmed. These
results suggest that during time winning, teams frequently decrease their percentage of
possession, which could be associated to defending closer to the goal, counterattacking,
or playing directly [7]. It has also been shown that teams that were ahead performed a
higher number of defensive actions which, in turn, are related to lower ball possession
levels during a match [31]. By contrast, during time losing, these results suggest teams
frequently increase their percentage of possession, attacking closer to the other team’s
goal [7]. Evidence has reported that successful teams normally have longer possession
times than less successful teams [23] or that ball possession might increase in teams that
are either losing or trying to tie the match [32]. Therefore, our results reported the need to
take into account the evolution of the scoreline.
In contrast, TDWP >21 km·h−1 increased during time winning for all player positions
and decreased during time losing. A possible reason to explain this result could be the fact
that teams adopt an indirect playing style to perform counterattacks [7]. Therefore, players
need to execute high-intensity specific technical and tactical tasks on the pitch when they
are in ball possession, such as receiving passes and crosses on the run, followed by dribbling
the ball in the opponent’s area to obtain a goal [33]. In fact, it has been demonstrated that
high-intensity actions are important within decisive situations in professional football [34].
Furthermore, Yang et al. [24] reported that total sprint distance was significantly greater
for the best-ranked teams compared to lower-ranked teams, highlighting the importance
of sprinting for tactical teamwork that generates offensive actions. A systematic review
conducted by Lago–Peñas & Sanromán–Álvarez [35] pointed out that successful teams
covered greater high-intensity running distance in ball possession. Therefore, our results
suggest that teams perform a greater number of high-intensity actions in ball possession
while winning or trying to maintain the advantage; meanwhile, TDWP decreases.
During time winning, TDWOP increased for all player positions and decreased dur-
ing time losing for all player positions, except CD. One potential reason for this situa-
tion could be the fact of ball possession decrease, like Lago-Peñas [7] reported, showing
that winning teams preferred counterattacking or playing directly. On the other hand,
research has shown that lower-ranked teams covered significantly greater TDWOP com-
pared with better-ranked teams, which likely represented a greater match time under-
taking defensive activities by these teams [24]. Our findings disagree with those from
Sensors 2021, 21, 6843
7 of 9
other studies where ball possession increased when teams were ahead [36]. Similarly,
TDWOP > 21 km·h−1 covered by WM and FW significantly increased when teams were
ahead, and TDWOP > 21 km·h−1 covered by WD and CM increased when teams were
behind. This fact could be explained due to that, when the team is not in possession of
the ball, the forwards often perform high-intensity activities (high pressing), attempting to
recover the lost ball [14,37].
Secondly, it had been hypothesized that TD would be greater in attackers while
teams were ahead and in defenders when teams were losing (Hypothesis 2). The results
showed that TD and TD > 21 km·h−1 covered by CM, WM, and FW significantly increased
when teams were ahead (p < 0.05), and TD and TD > 21 km·h−1 covered by CD and WD
significantly increased when teams were losing (p < 0.05). These findings are in line with
previous studies that found that attackers covered more distance at high intensity when
winning and defenders more when losing [16]. Lago–Peñas et al. [17] also reported that
losing status increased total distance covered by defenders, while attacking players showed
the opposite trend. Thus, it seems to be confirmed that CM, WM, and FW cover greater
TD when winning and CD and WD when losing, therefore Hypothesis 2 was accepted.
A possible explanation may be due to attackers´ work rate of the opposing team since,
when the opposing team is ahead or chasing a goal, attackers maintain a high work rate,
implying a defenders´ high work rate [16,38]. In addition, similar results were obtained by
Andrzejewski et al. [14], showing that defenders covered shorter distances at high intensity
in lost matches, while forwards covered longer total distances in won matches. Another
possible reason could be the playing style that the team adopted, for example, a direct style
of play when teams are winning can induce higher match intensity in running from the
attackers [7,11]. In particular, these findings indicate that physical demands vary according
to position-specificities and the evolution of the scoreline.
4.1. Study Limitations and Future Directions
The present study increases the knowledge about this research topic; however, a
number of limitations could be recognized with a view to future research. First, other
context-related variables such as match location or opposing team level were not consid-
ered. Ball possession percentages of teams have been also not considered, and it would be
interesting to analyze this variable when teams are winning or losing with the interaction of
match physical demands. Moreover, the comparison between five players’ positions accord-
ing to previous studies was analyzed [18]; however, the existence of more player positions
is possible than have been previously analyzed, therefore it would be interesting to conduct
a comparison between more player positions. In addition, further research is required,
considering several factors such as the playing style, since the player position could depend
on the playing style of teams. Finally, research has reported that external load variables
such as accelerations and decelerations belong to match physical demands [39], in which
case it would be necessary to know the full player´s work rate, including these variables.
4.2. Practical Applications
The findings of this study provide useful information on the variability of match
physical demands for practitioners in Spanish professional soccer. In particular, the study
extends previous research demonstrating that time that teams were winning or losing
influences both match physical demands and ball possession. This information could help
strength and conditioning coaches with personalizing recovery work after match play,
according to the different physical efforts performed in matches. Finally, goals scored are
the most important of all critical events, therefore the evolution of the scoreline should
be taken into account during training sessions to optimize physical aspects of soccer
performance. In this vein, it is necessary to know how these situations influence the
player’s capacity to deal with critical events in a match [40].
Sensors 2021, 21, 6843
8 of 9
5. Conclusions
The main findings reported that the evolution of scoreline significantly influences
match tactical–technical and physical demands. First, TDWP was less while teams were
winning, while it was greater while teams were losing, and TDWOP evolved conversely;
therefore, teams modify their playing style and tactical behavior according to the demands
of matches. Secondly, attackers covered greater distances when winning, and defenders
covered greater distances when losing; therefore, professional soccer players regulate their
physical efforts according to the periods of the game. Finally, the influence of scoreline is
reflected in changes in the teams and players’ tactical–technical and physical demands as a
response to the evolution of match outcome.
Author Contributions: Conceptualization, T.G.-C.; methodology, T.G.-C., J.C.P.-B. and J.D.-G.; formal
analysis, T.G.-C.; investigation, J.C.P.-B., J.D.-G., M.A.L.-G., D.L.-T., R.L.d.C., R.R. and T.G.-C.; re-
sources, R.L.d.C. and R.R.; writing—original draft preparation, J.C.P.-B., J.D.-G. and D.L.-T.; writing—
review and editing, M.A.L.-G. and T.G.-C.; funding acquisition, R.L.d.C. and R.R. All authors have
read and agreed to the published version of the manuscript.
Funding: This research was funded by the European Regional Development Fund (ERDF), the
Government of Extremadura (Department of Economy and Infrastructure) and LaLiga Research and
Analysis Sections.
Institutional Review Board Statement: The study was conducted according to the guidelines of
the Declaration of Helsinki, and approved by the Institutional Review Board of the University of
Extremadura; Vice-Rectorate of Research, Transfer and Innovation-Delegation of the Bioethics and
Biosafety Commission (Protocol number: 239/2019).
Informed Consent Statement: Informed consent was obtained from all subjects involved in the study.
Data Availability Statement: Restrictions apply to the availability of these data. Data was obtained
from LaLiga and are available with the permission of corresponding author.
Conflicts of Interest: The authors declare no conflict of interest. Also, the funders had no role in
the design of the study; in the collection, analyses, or interpretation of data; in the writing of the
manuscript, or in the decision to publish the results.
References
1.
Lago-Peñas, C. The Role of Situational Variables in Analysing Physical Performance in Soccer by. J. Hum. Kinet. 2012, 35, 89–95.
[CrossRef] [PubMed]
2.
Castellano, J.; Blanco-Villaseñor, A.; Álvarez, D. Contextual variables and time-motion analysis in soccer. Int. J. Sports Med. 2011,
32, 415–421. [CrossRef] [PubMed]
3.
Moalla, W.; Fessi, M.S.; Makni, E.; Dellal, A.; Filetti, C.; Salvo, V.D.I.; Chamari, K. Association of physical and technical activities
with partial match status in a soccer professional team. J. Strength Cond. Res. 2018, 32, 1708–1714. [CrossRef]
4.
Sullivan, C.; Bilsborough, J.C.; Cianciosi, M.; Hocking, J.; Cordy, J.; Coutts, A.J. Match score affects activity profile and skill
performance in professional Australian Football players. J. Sci. Med. Sport 2014, 17, 326–331. [CrossRef] [PubMed]
5.
Lago-Peñas, C.; Dellal, A. Ball possession strategies in elite soccer according to the evolution of the match-score: The influence of
situational variables. J. Hum. Kinet. 2010, 25, 93–100. [CrossRef]
6.
Fernandez-Navarro, J.; Fradua, L.; Zubillaga, A.; McRobert, A.P. Influence of contextual variables on styles of play in soccer. Int. J.
Perform. Anal. Sport 2018, 18, 423–436. [CrossRef]
7.
Lago-Peñas, C. The influence of match location, quality of opposition, and match status on possession strategies in professional
association football. J. Sports Sci. 2009, 27, 1463–1469. [CrossRef]
8.
Lago-Peñas, C.; Casais, L.; Dominguez, E.; Sampaio, J. The effects of situational variables on distance covered at various speeds in
elite soccer. Eur. J. Sport Sci. 2010, 10, 103–109. [CrossRef]
9.
Casamichana, D.; Castellano, J. Situational variables and distance covered during the FIFA Wold Cup South Africa 2010. Rev. Int.
Med. Ciencias Act. Fis. Deport. 2014, 14, 603–617.
10.
Aquino, R.; Martins, G.; Vieira, L.H.P.; Menezes, R.P. Influence of match location, quality of opponents, and match status on
movement patterns in brazilian professional football players. J. Strength Cond. Res. 2017, 31, 2155–2161. [CrossRef]
11.
Aquino, R.; Carling, C.; Palucci Vieira, L.H.; Martins, G.; Jabor, G.; Machado, J.; Santiago, P.; Garganta, J.; Puggina, E. Influence
of situational variables, team formation, and playing position on match running performance and social network analysis in
Brazilian professional soccer players. J. Strength Cond. Res. 2020, 34, 808–817. [CrossRef]
Sensors 2021, 21, 6843
9 of 9
12.
Oliva-Lozano, J.M.; Rojas-Valverde, D.; Gómez-Carmona, C.D.; Fortes, V.; Pino-Ortega, J. Impact of contextual variables on the
representative external load profile of Spanish professional soccer match-play: A full season study. Eur. J. Sport Sci. 2021, 21,
497–506. [CrossRef] [PubMed]
13.
Rago, V.; Brito, J.; Figueiredo, P.; Costa, J.; Barreira, D.; Krustrup, P.; Rebelo, A. Methods to collect and interpret external training
load using microtechnology incorporating GPS in professional football: A systematic review. Res. Sports Med. 2019, 28, 437–458.
[CrossRef] [PubMed]
14.
Andrzejewski, M.; Konefał, M.; Chmura, P.; Kowalczuk, E.; Chmura, J. Match outcome and distances covered at various speeds
in match play by elite German soccer players. Int. J. Perform. Anal. Sport 2016, 16, 817–828. [CrossRef]
15.
Wehbe, G.; Hartwig, T.; Duncan, C. Movement analysis of Australian national league soccer players using global positioning
system technology. J. Strength Cond. Res. 2014, 28, 834–842. [CrossRef]
16.
Redwood-Brown, A.; O’Donoghue, P.; Robinson, G.; Neilson, P. The effect of score-line on work-rate in English FA Premier
League soccer. Int. J. Perform. Anal. Sport 2012, 12, 258–271. [CrossRef]
17.
Lago-Peñas, C.; Kalén, A.; Lorenzo-Martínez, M. Do elite soccer players cover longer distance when losing? Differences between
attackers and defenders. Int. J. Sports Sci. Coach. 2020, 22, 1–12.
18.
Lorenzo-Martinez, M.; Kalén, A.; Rey, E.; López-Del Campo, R.; Resta, R.; Lago-Peñas, C. Do elite soccer players cover less
distance when their team spent more time in possession of the ball? Sci. Med. Footb. 2020, 1–7. [CrossRef]
19.
Felipe, J.L.; Garcia-Unanue, J.; Viejo-Romero, D.; Navandar, A.; Sánchez-Sánchez, J. Validation of a video-based performance
analysis system (Mediacoach®) to analyze the physical demands during matches in LaLiga. Sensors 2019, 19, 4113. [CrossRef]
20.
Linke, D.; Link, D.; Lames, M. Football-specific validity of TRACAB’s optical video tracking systems. PLoS ONE 2020, 15, e0230179.
[CrossRef]
21.
Pons, E.; García-Calvo, T.; Resta, R.; Blanco, H.; López del Campo, R.; Díaz García, J.; Pulido, J.J. A comparison of a GPS device
and a multi-camera video technology during official soccer matches: Agreement between systems. PLoS ONE 2019, 14, e0220729.
[CrossRef]
22.
Pons, E.; García-Calvo, T.; Cos, F.; Resta, R.; Blanco, H.; López del Campo, R.; Díaz-García, J. Integrating video tracking and GPS
to quantify accelerations and decelerations in elite soccer. Sci. Rep. 2021, 11, 18531. [CrossRef]
23.
Brito de Souza, D.; López-Del Campo, R.; Blanco-Pita, H.; Resta, R.; Del Coso, J. Association of match running performance with
and without ball possession to football performance. Int. J. Perform. Anal. Sport 2020, 20, 483–494. [CrossRef]
24.
Yang, G.; Leicht, A.S.; Lago, C.; Gómez, M.Á. Key team physical and technical performance indicators indicative of team quality
in the soccer Chinese super league. Res. Sport. Med. 2018, 26, 158–167. [CrossRef]
25.
R-Studio Team (Ed.) RStudio: Integrated Development for R; R-Studio Team: Boston, MA, USA, 2020.
26.
Bates, D.; Machler, M.; Bolker, B.; Walker, S. Fitting linear mixed-effects models using lme4. J. Stat. Softw. 2015, 67, 1–48. [CrossRef]
27.
Heck, R.H.; Thomas, S.L. An Introduction to Multilevel Modeling Techniques: MLM and SEM Approaches Using Mplus; Routledge:
New York, NY, USA, 2015.
28.
Akaike, H. A New Look at the Statistical Model Identification. IEEE Trans. Automat. Contr. 1974, 19, 716–723. [CrossRef]
29.
Field, A. Discovering Statistics Using IBM SPSS Statistics, 4th ed.; SAGE Editorial: London, UK, 2013.
30.
Nakagawa, S.; Schielzeth, H. A general and simple method for obtaining R2 from generalized linear mixed-effects models.
Methods Ecol. Evol. 2013, 4, 133–142. [CrossRef]
31.
Morgans, R.; Adams, D.; Mullen, R.; Williams, M.D. Changes in physical performance variables in an English Championship
League team across the competitive season: The effect of possession. Int. J. Perform. Anal. Sport 2014, 14, 493–503. [CrossRef]
32.
Bradley, P.; Lago-Peñas, C.; Rey, E.; Sampaio, J. The influence of situational variables on ball possession in the English Premier
League. J. Sports Sci. 2014, 32, 1867–1873. [CrossRef] [PubMed]
33.
Andrzejewski, M.; Chmura, J.; Pluta, B. Analysis of motor and technical activities of professional soccer players of the UEFA
Europa league. Int. J. Perform. Anal. Sport 2014, 14, 504–523. [CrossRef]
34.
Faude, O.; Koch, T.; Meyer, T. Straight sprinting is the most frequent action in goal situations in professional football. J. Sports Sci.
2012, 30, 625–631. [CrossRef] [PubMed]
35.
Lago-Peñas, C.; Sanromán-Álvarez, P. La influencia de la posesión del balón en el rendimiento físico en el fútbol profesional. Una
revisión sistemática. JUMP 2020, 68–80. [CrossRef]
36.
Casal, C.A.; Maneiro, R.; Ardá, T.; Marí, F.J.; Losada, J.L. Possession zone as a performance indicator in football. The game of the
best teams. Front. Psychol. 2017, 8, 1176. [CrossRef]
37.
Dellal, A.; Wong, D.P.; Moalla, W.; Chamari, K. Physical and technical activity of soccer players in the French first league- with
special reference to their playing position. Int. Sport. J. 2010, 11, 278–290.
38.
Lago-Peñas, C.; Rey, E.; Lago-Ballesteros, J.; Casais, L.; Domínguez, E. Analysis of work-rate in soccer according to playing
positions. Int. J. Perform. Anal. Sport 2009, 9, 218–227. [CrossRef]
39.
Dalen, T.; Jorgen, I.; Gertjan, E.; Havard, H.; Ulrik, W. Player load, acceleration and deceleration during forty-five competitive
matches of elite soccer. J. Strength Cond. Res. 2016, 30, 351–359. [CrossRef]
40.
Higham, A.; Harwood, C.; Cale, A. Momentum in Soccer: Controlling the Game; Kelly, A., Ed.; Coachwise Ltd.: Leeds, UK, 2005.
| The Influence of Time Winning and Time Losing on Position-Specific Match Physical Demands in the Top One Spanish Soccer League. | 10-14-2021 | Ponce-Bordón, José C,Díaz-García, Jesús,López-Gajardo, Miguel A,Lobo-Triviño, David,López Del Campo, Roberto,Resta, Ricardo,García-Calvo, Tomás | eng |
PMC9685973 | Fiedler et al. BMC Research Notes (2022) 15:351
https://doi.org/10.1186/s13104-022-06247-1
RESEARCH NOTE
Daytime fluctuations of endurance
performance in young soccer players:
a randomized cross-over trial
Janis Fiedler1* , Stefan Altmann1,2, Hamdi Chtourou3,4, Florian A. Engel5, Rainer Neumann6 and
Alexander Woll1
Abstract
Objectives: Fluctuations of physical performance and biological responses during a repetitive daily 24-h cycle are
known as circadian rhythms. These circadian rhythms can influence the optimal time of day for endurance perfor-
mance and related parameters which can be crucial in a variety of sports disciplines. The current study aimed to
evaluate the daytime variations in endurance running performance in a 3.000-m field run and endurance running
performance, blood lactate levels, and heart rate in an incremental treadmill test in adolescent soccer players.
Results: In this study, 15 adolescent male soccer players (age: 18.0 ± 0.6 years) performed a 3.000-m run and an
incremental treadmill test at 7:00–8:00 a.m. and 7:00–8:00 p.m. in a randomized cross-over manner. No significant
variations after a Bonferroni correction were evident in endurance running performance, perceived exertion, blood
lactate levels, and heart rates between the morning and the evening. Here, the largest effect size was observed for
maximal blood lactate concentration (9.15 ± 2.18 mmol/l vs. 10.64 ± 2.30 mmol/l, p = .110, ES = 0.67). Therefore,
endurance running performance and physiological responses during a field-based 3.000-m run and a laboratory-
based test in young male soccer players indicated no evidence for daytime variations.
Keywords: Circadian rhythm, Soccer, Aerobic exercise, Endurance, Lactate, Heart rate
© The Author(s) 2022. Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which
permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the
original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or
other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line
to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory
regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this
licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/. The Creative Commons Public Domain Dedication waiver (http:// creat iveco
mmons. org/ publi cdoma in/ zero/1. 0/) applies to the data made available in this article, unless otherwise stated in a credit line to the data.
Introduction
Circadian rhythms describe periodic changes in physi-
ological parameters for an approximately 24-h cycle [1].
They are well established for a range of biological param-
eters like core body temperature, heart rate (HR), blood
pressure, and different hormones [1] and are also pre-
sent in physical performance and related responses [2, 3].
These circadian rhythms are influenced by other parame-
ters like age, light hours, sleeping pattern, or type of exer-
cise but are overall stable [2]. For coaches and athletes
(i.e. soccer players), it might be important to consider
circadian rhythms as determinants of exercise capacity
as well as performance for the best results in competi-
tions [4]. As endurance running performance is related
to overall performance in soccer players, and elite play-
ers run about 10 km during one game, this motor fitness
parameter is of particular interest [5]. Previous research
including soccer players found heterogenic results con-
cerning the presence of daytime variation for endurance
performance and related physiological responses like lac-
tate or HR [2, 4, 6–13].
Therefore, this study aimed to examine potential day-
time variation (morning vs. evening) in i) endurance
running performance during a 3.000-m field run and an
incremental treadmill test; and ii) blood lactate concen-
tration and HR during the incremental treadmill test.
Open Access
BMC Research Notes
*Correspondence: [email protected]
1 Institute of Sports and Sports Science, Karlsruhe Institute of Technology,
Engler-Bunte-Ring 15, 76131 Karlsruhe, Germany
Full list of author information is available at the end of the article
Page 2 of 6
Fiedler et al. BMC Research Notes (2022) 15:351
According to the literature, we hypothesized that (i)
endurance running performance during the 3.000-m
run and the treadmill test would be higher in the even-
ing than in the morning [7, 12, 14, 15] and (ii) that both
blood lactate levels and HR during the incremental tread-
mill test would be different between the morning and
evening [4, 16–21].
Material and Methods
Participants
Fifteen male soccer players (age = 18.0 ± 0.6 years;
height = 178.7 ± 5.3 cm; weight = 71.1 ± 6.6 kg), with
a regular training volume of three training sessions per
week and one soccer match on the weekend, volunteered
to participate in this study.
Procedures
All 15 participants performed a 3.000-m run and an
incremental treadmill test (see [22]) on two occasions
at two-day times (one in the morning between 7:00 and
8:00 a.m. as well as one in the evening between 7:00 and
8:00 p.m.). Using a cross-over design, the participants
were randomly assigned to two groups. Both groups per-
formed the 3.000-m run first and the incremental tread-
mill test trials second. However, group 1 performed the
first trial for both tests (3.000-m run and incremental
treadmill test, respectively) in the morning and the sec-
ond in the evening, while the timing was switched for
group 2. All tests were separated by 36 h.
In a first step, all participants performed the two field-
based 3.000-m runs on a 400 m running track. The par-
ticipants were familiar with the 3.000-m run and were
instructed to perform the whole 3.000-m run as fast as
possible. Time to completion and ratings of perceived
exertion (RPE) [23] were recorded after each trial in
order to control for exhaustion criteria [24].
In a second step, all participants performed two labo-
ratory-based incremental treadmill tests on a Woodway
treadmill (Woodway GmbH, Weil am Rhein, Germany)
with a slope of 1%. Each trial started at a running speed
of 6 km/h, increasing by 2 km/h every 3 min. After each
3 min-stage, participants rested for 30-s for collection
of capillary blood from the earlobe; two participants
provided no consent for blood withdrawal and lactate
thresholds were estimated using the Ergonizer Software
(Ergonizer, Freiburg, Germany). HR was monitored
using a Polar system (Polar Electro Oy, Kempele, Fin-
land) throughout the whole test. Athletes were instructed
to complete as many stages as possible, and the test
was finished at volitional exhaustion. Blood lactate
concentration for each stage was analyzed utilizing Bio-
sen C-Line Sport (EKF-diagnostic GmbH, Barleben,
Germany).
Data analysis
Time to completion and RPE were recorded as param-
eters for the 3.000-m run. Regarding the incremental
treadmill test, the following measurement points were
chosen to measure one or multiple of the following
parameters: blood lactate concentration, HR, and run-
ning speed (see [25]):
• rest: immediately before the beginning of the test in
a standing position
• individual aerobic threshold (LT): running velocity
at which blood lactate concentration begins to rise
above baseline levels
• individual anaerobic threshold (IAT): running
velocity at LT + blood lactate concentration of
1.5 mmol/l
• maximal running speed (max): running velocity at
the point of volitional exhaustion
The following parameters were included for the incre-
mental treadmill test:
• blood lactate concentration (at rest, LT, IAT, and
max)
• HR (at rest, LT, IAT, and max)
• running speed (LT, IAT, and max)
Statistical analysis
Because of the cross-over study design, the existence of
possible sequencing effects was calculated by perform-
ing an independent t-test between the sum scores (day
1 + day 2 group 1 vs day 1 + day 2 group 2) for each
parameter in addition to a sufficient washout period [26].
All Data are available in the Additional file 1.
Daytime variations in all measured variables were cal-
culated using paired t-tests. To correct for multiple test-
ing, the results were adapted by multiplying the p-value
with the number of comparisons of the parameter follow-
ing the Bonferroni correction [27]. In addition, Cohen’s
d effect sizes (ES) were calculated to quantify the mag-
nitude of differences between the morning and even-
ing trials: 0.2 ≤ ES < 0.5 was considered a small effect;
0.5 ≤ ES < 0.8 was considered a moderate effect; ES ≥ 0.8
was considered a large effect [28]. Statistical analyses
were performed using SPSS statistical software version
26.0 (SPSS, Inc., Chicago, IL). The level for significance
was set a priori to 0.05 after the Bonferroni correction.
Page 3 of 6
Fiedler et al. BMC Research Notes (2022) 15:351
Results
The investigation of potential sequencing effects, ana-
lyzed using an independent t-test, showed no signifi-
cant differences between the two groups.
For the 3.000-m run, neither time for completion
(see Fig. 1a) nor RPE (see Fig. 1b) differed significantly
between the morning and evening trials. For the incre-
mental treadmill test, no significant differences after
the Bonferroni correction were found for blood lac-
tate (maximal blood lactate concentration see Fig. 2a)
or running speed (maximal running speed see Fig. 2b)
between the morning and evening trials (see Table 1
for detailed results).
Discussion and conclusion
This study aimed to evaluate daytime variation in aerobic
endurance performance in a 3.000-m run and an incre-
mental treadmill test in young soccer players. Addition-
ally, blood lactate concentrations and HR during the
incremental treadmill test were analyzed for daytime
differences. Hypothesis (i) that aerobic endurance per-
formance would be better in the evening than in the
morning could not be verified for the 3.000-m run and
the incremental treadmill test. Hypothesis (ii) that blood
lactate levels and HR during exercise would be higher in
the evening could also not be verified.
Fig. 1 The individual values of all participants (lines) and the mean value (box) for the parameters a time to completion, and b perceived exertion
during the morning and evening trial of the 3.000-m run
Fig. 2 The individual values of all participants (lines) and the mean value (box) for a maximal blood lactate concentration, and b maximal running
speed during the morning and evening trial of the incremental treadmill test
Page 4 of 6
Fiedler et al. BMC Research Notes (2022) 15:351
Aerobic endurance performance in the incremen-
tal treadmill test indicated no evidence for differences
between the evening and the morning. This is in line with
some previous studies in untrained participants [20] and
competitive cyclists [29] while others reported increased
endurance performance in an incremental cycle ergome-
ter test in students [15] and a Yo-Yo intermittent recovery
test in young soccer players [12]. While there is a good
theoretical basis for performance differences due to hor-
monal control of glucose metabolism [13], results from
laboratory and field studies yield heterogenic findings.
Additionally, no differences in endurance performance
and RPE were found for the field test (i.e., 3.000-m run).
One possible explanation for the results of the 3.000-m
run is that the self-selected pacing is a crucial factor for
maximum performance in the 3.000-m run [30]. This
is supported by the reported mean RPE which did not
reach the range of exhaustion criteria (RPE > 16) in the
3.000-m run, while exhaustion criteria were reached
(mean max lactate > 9 mmol/l) [24] in the incremental
treadmill test.
Furthermore, no evidence for a daytime variation in
any physiological parameter was found in our study. Con-
trasting, previous studies found higher blood lactate lev-
els for various exercises [4, 21]. Additionally, one study
reported higher blood lactate levels at rest in the morn-
ing compared to the afternoon and evening [20], and
another study found no differences in blood lactate levels
throughout the day [29]. Reasons for the different results
between the aforementioned studies and the results of
the present study can be found in different test proce-
dure and population. Concerning daytime variations of
HR during endurance exercise, the overall results seem
to be inconsistent [30]. While some studies reported evi-
dence for the presence of daytime variation in HR [16–
18], Chtourou and Souissi described equivocal results for
daytime variation of HR in their recent review [30].
Overall, our hypotheses that daytime variations are
present in endurance performance and related physi-
ological parameters of youth soccer players could not
be confirmed by this study. While circadian rhythms are
considered an important factor related to physical per-
formance and physiological parameters in competitive
sports, the importance of circadian rhythms for aerobic
endurance performance remains unclear.
Limitations
Some limitations must be acknowledged concerning
this study. First, the use of only two times of the day (i.e.,
morning and evening) might not be sufficient because
the time window for optimal performance differs for each
individuum [2]. However, the choice of the selected times
of the day in our study did incorporate the optimum time
of day for soccer players’ performance between 04:00
Table 1 Results for endurance running performance, blood lactate levels, and heart rate differences between morning and evening
Means (standard deviations) and results of the paired t-tests for daytime differences at the incremental treadmill test before the start (rest), at the onset of lactate
accumulation (LT), the individual anaerobic threshold (IAT), and immediately after volitional exhaustion (max) and at the end of the 3.000-m run for time to
completion (Time) and rating of perceived exertion (RPE). p-values were corrected using the Bonferroni method
Incremental treadmill test
Parameter
Morning
Evening
Mean difference
corrected p-value
(original)
Cohen’s d (t-value)
df
Heart rate [1/min]
Rest
86.60 (9.68)
85.73 (10.88)
0.87 (10.33)
1.00 (0.750)
− 0.09 (0.35)
14
LT
150.93 (12.13)
153.47 (10.29)
− 2.53 (9.81)
1.00 (0.334)
0.23 (− 1)
14
IAT
177.47 (7.85)
179.27 (6.31)
− 1.80 (4.87)
0.872 (0.174)
0.25 (− 1.43)
14
Max
197.13 (6.29)
198.73 (6.08)
− 1.60 (4.21)
0.814 (0.163)
0.26 (− 1.47)
14
Lactate concentration [mmol/l]
Rest
0.84 (0.21)
0.83 (0.31)
0.01 (0.31)
1.00 (0.930)
− 0.04 (0.09)
12
LT
1.52 (0.67)
1.66 (0.61)
− 0.13 (0.40)
1.00 (0.250)
0.22 (− 1.20)
12
IAT
3.02 (0.67)
3.16 (0.61)
− 0.14 (0.40)
0.962 (0.241)
0.22 (− 2.51)
12
Max
9.15 (2.18)
10.64 (2.30)
− 1.49 (2.15)
0.110 (0.028)
0.67 (− 2.51)
12
Running speed [km/h]
LT
8.67 (1.17)
9.00 (1.10)
− 3.30 (0.74)
0.429 (0.107)
0.29 (− 1.72)
14
IAT
11.94 (1.28)
12.12 (1.18)
− 0.19 (0.68)
1.00 (0.317)
0.15 (− 1.04)
13
Max
15.81 (1.62)
16.31 (1.59)
− 0.49 (0.76)
0.100 (0.025)
0.31 (− 2.51)
14
3.000-m test
Time [min:sec]
12:59:00 (1:29)
13:06:00 (1:30)
− 0:07 (0:22)
0.228
0.08 (− 1.26)
15
RPE
15.31 (1.82)
15.44 (1.09)
− 0.13 (1.78)
0.783
0.09 (− 0.28)
15
Page 5 of 6
Fiedler et al. BMC Research Notes (2022) 15:351
p.m. and 08:00 p.m. [6] to compensate for this shortcom-
ing. Secondly, the RPE used in the 3.000-m run has not
been used in the incremental treadmill test, while blood
lactate testing has only been performed during the incre-
mental treadmill test and not after the 3.000-m run and
therefore limits the interpretation concerning exhaus-
tion criteria. Other important factors might be that this
study did not control for sleeping patterns, sleep dura-
tion, naps, and morning or evening type of participants
which is known to influence the circadian rhythm [2, 31].
Here, the relation between the chronotype and the per-
formance of athletes at certain daytimes is particularly
interesting but evidence in the literature is heterogenic
[32, 33]. Finally, a larger sample size would have reduced
the beta error and would lead to more robust results.
Future studies should address these shortcomings by
adding physiological parameters to control for exhaus-
tion criteria with parameters like blood lactate, HR,
and RPE. Additionally, sleep related variables, and
chronotype of participants should be considered. This
may enable researchers to distinguish between physio-
logical and psychological aspects of aerobic endurance
performance and to better determine if and why day-
time variations are present for the different outcome
parameters. Finally, if studies aim to determine sport-
specific (i.e., soccer) daytime variation, a field test rep-
resenting the sport-specific requirements seems more
appropriate compared to generic endurance tests like
the 3.000-m run.
Abbreviations
ES: Effect size (Cohen’s d); HR: Heart rate; IAT: Individual anaerobic threshold; LT:
Individual aerobic threshold; max: Maximal; RPE: Rate of perceived exertion.
Supplementary Information
The online version contains supplementary material available at https:// doi.
org/ 10. 1186/ s13104- 022- 06247-1.
Additional file 1. Exercise data of soccer players.
Acknowledgments
The authors would like to thank the participants for their enthusiastic
participation and the students for their support during data collection. We
acknowledge support by the KIT-Publication Fund of the Karlsruhe Institute of
Technology.
Author contributions
Conceptualization, JF, SA, FE; Data curation, JF; Formal analysis, JF; Investi-
gation, RN, FE, SA; Methodology, JF, SA, RN, FE, SA; Writing—original draft,
JF; Writing—review & editing, SA, HC, FE, RN, and AW All authors read and
approved the final manuscript.
Funding
Open Access funding enabled and organized by Projekt DEAL. No funding
was provided for this study.
Availability of data and materials
All data generated or analysed during this study are included in this published
article [and its Additional file 1].
Declarations
Ethics approval and consent to participate
All participants provided written informed consent before the start. The study
was approved by the Institutional Reviewer Board of the Institute of Sport and
Sport Science at the Karlsruhe Institute of Technology and all methods were
carried out according to the countries guidelines and regulations.
Consent for publication
Not applicable.
Competing interests
The authors declare no competing interests.
Author details
1 Institute of Sports and Sports Science, Karlsruhe Institute of Technology,
Engler-Bunte-Ring 15, 76131 Karlsruhe, Germany. 2 TSG ResearchLab gGmbH,
74939 Zuzenhausen, Germany. 3 Institut Supérieur du Sport et de l’Education
Physique de Sfax, Université de Sfax, 3000 Sfax, Tunisie. 4 Activité Physique,
Sport et Santé, UR18JS01, Observatoire National du Sport, 1003 Tunis, Tunisie.
5 Institute of Sport Science, Integrative & Experimental Exercise Science & Train-
ing, Würzburg University, 97070 Würzburg, Germany. 6 Institute of Movement
and Sport, University of Education Karlsruhe, 73133 Karlsruhe, Germany.
Received: 18 July 2022 Accepted: 11 November 2022
References
1.
Minors DS, Waterhouse JM. Circadian Rhythms and the Human. Oxford:
Elsevier Science; 2014.
2.
Reilly T, Atkinson G, Waterhouse J. Biological rhythms and exercise.
Oxford: Oxford University Press; 1997.
3.
Drust B, Waterhouse J, Atkinson G, Edwards B, Reilly T. Circadian rhythms
in sports performance–an update. Chronobiol Int. 2005;22(1):21–44.
4.
Hammouda O, Chtourou H, Chaouachi A, Chahed H, Bellimem H,
Chamari K, et al. Time-of-day effects on biochemical responses to
soccer-specific endurance in elite Tunisian football players. J Sports Sci.
2013;31(9):963–71.
5.
Stølen T, Chamari K, Castagna C, Wisløff U. 2005 Physiology of soc-
cer: an update. Sports Med. 35(6):501–36. https://link.springer.com/
article/https:// doi. org/ 10. 2165/ 00007 256- 20053 5060- 00004.
6.
Reilly T, Atkinson G, Edwards B, Waterhouse J, Farrelly K, Fairhurst E. Diur-
nal variation in temperature, mental and physical performance, and tasks
specifically related to football (soccer). Chronobiol Int. 2007;24(3):507–19.
7.
Hammouda O, Chtourou H, Chahed H, Ferchichi S, Chaouachi A, Kallel
C, et al. Diurnal variations in physical performances related to football in
young soccer players. Int J Sports Med. 2012;33(11):886–91.
8.
Hammouda O, Chahed H, Chtourou H, Ferchichi S, Miled A, Souissi N.
Morning-to-evening difference of biomarkers of muscle injury and
antioxidant status in young trained soccer players. Biol Rhythm Res.
2012;43(4):431–8.
9.
Chtourou H, Hammouda O, Souissi N, Chaouachi A. Temporal specific-
ity of training: an update. J Athl Enhancement. 2014. https:// doi. org/ 10.
4172/ 2324- 9080. 10001 53.
10. Chtourou H, Chaouachi A, Driss T, Dogui M, Behm DG, Chamari K, et al.
The effect of training at the same time of day and tapering period on the
diurnal variation of short exercise performances. J Strength Cond Res.
2012;26(3):697–708.
11. Hill DW, Cureton KJ, Collins MA. Circadian specificity in exercise training.
Ergonomics. 1989;32(1):79–92.
12. Hammouda O, Chtourou H, Farjallah MA, Davenne D, Souissi N. The effect
of Ramadan fasting on the diurnal variations in aerobic and anaero-
bic performances in Tunisian youth soccer players. Biol Rhythm Res.
2012;43(2):177–90.
Page 6 of 6
Fiedler et al. BMC Research Notes (2022) 15:351
•
fast, convenient online submission
•
thorough peer review by experienced researchers in your field
•
rapid publication on acceptance
•
support for research data, including large and complex data types
•
gold Open Access which fosters wider collaboration and increased citations
maximum visibility for your research: over 100M website views per year
•
At BMC, research is always in progress.
Learn more biomedcentral.com/submissions
Ready to submit your research
Ready to submit your research ? Choose BMC and benefit from:
? Choose BMC and benefit from:
13. Kusumoto H, Ta C, Brown SM, Mulcahey MK. Factors contributing to
diurnal variation in athletic performance and methods to reduce within-
day performance variation: a systematic review. J Strength Cond Res.
2021;35(Suppl 12):S119–35.
14. Souissi W, Hammouda O, Ammar A, Ayachi M, Bardiaa Y, Daoud O, et al.
Higher evening metabolic responses contribute to diurnal variation of
self-paced cycling performance. Biol Sport. 2021. https:// doi. org/ 10. 5114/
biols port. 2021. 102930.
15. Hill DW, Cureton KJ, Ma Collins, Grisham SC. Diurnal variations in
responses to exercise of morning types and evening types. J Sports Med
Phys Fitness. 1988;28(3):213–9.
16. Cohen CJ, Muehl GE. Human circadian rhythms in resting and exercise
pulse rates. Ergonomics. 1977;20(5):475–9.
17. Cohen CJ. Human circadian rhythms in heart rate response to a maximal
exercise stress. Ergonomics. 1980;23(6):591–5.
18. Wahlberg I, Åstrand I. Physical work capacity during the day and at night.
Work, Environment, Health. 1973;10(2):65–8.
19. Winget CM, DeRoshia CW, Holley DC. Circadian rhythms and athletic
performance. Med Sci Sports Exerc. 1985;17(5):498–516.
20. Dschenes MR, Sharma JV, Brittingham KT, Casa DJ, Le Armstrong, Maresh
CM. Chronobiological effects on performance and selected physiological
responses. Eur J Appl Physiol. 1998;77(3):249–56.
21. Forsyth JJ, Reilly T. Circadian rhythms in blood lactate concentra-
tion during incremental ergometer rowing. Eur J Appl Physiol.
2004;92(1–2):69–74.
22. Altmann S, Neumann R, Woll A, Härtel S. Endurance capacities in profes-
sional soccer players: are performance profiles position specific? Front
Sports Act Living. 2020;2:549897.
23. Borg G. Perceived exertion as an indicator of somatic stress. Scand J
Rehabil Med. 1970;2(2):92–8.
24. Dickhuth H-H, Röcker K, Gollhofer A, König D, Mayer F. 2011 Einführung in
die Sport-und Leistungsmedizin. Überarbeitete und aktualisierte Auflage.
Schorndorf: Hofmann
25. Faude O, Kindermann W, Meyer T. Lactate threshold concepts: how valid
are they? Sports Med. 2009;39(6):469–90.
26. Wellek S, Blettner M. On the proper use of the crossover design in clinical
trials: part 18 of a series on evaluation of scientific publications. Dtsch
Arztebl Int. 2012;109(15):276–81.
27. Shaffer JP. Multiple hypothesis testing. Annu Rev Psychol. 1995;46:561–84.
28. Cohen J. 1988 Statistical Power Analysis for the Behavioral Sciences. L.
Erlbaum Associates, Hillsdale, NJ.
29. Dalton B, McNaughton L, Davoren B. Circadian rhythms have no effect on
cycling performance. int J Sports Med. 1997;18(7):538–42.
30. Chtourou H, Souissi N. The effect of training at a specific time of day: a
review. J Strength Cond Res. 2012;26(7):1984–2005.
31. Lastella M, Halson SL, Vitale JA, Memon AR, Vincent GE. To nap or not to
nap? a systematic review evaluating napping behavior in athletes and
the impact on various measures of athletic performance. Nat Sci Sleep.
2021;13:841–62.
32. Rae DE, Stephenson KJ, Roden LC. 2015 Factors to consider when
assessing diurnal variation in sports performance: the influence of
chronotype and habitual training time-of-day. Eur J Appl Physiol.
115(6):1339–49. https://link.springer.com/article/https:// doi. org/ 10. 1007/
s00421- 015- 3109-9.
33. Vitale JA, Weydahl A. 2017 Chronotype, Physical Activity, and Sport Per-
formance: A Systematic Review. Sports Med. 47(9):1859–68. https://link.
springer.com/article/https:// doi. org/ 10. 1007/ s40279- 017- 0741-z. Decla
ratio ns
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in pub-
lished maps and institutional affiliations.
| Daytime fluctuations of endurance performance in young soccer players: a randomized cross-over trial. | 11-24-2022 | Fiedler, Janis,Altmann, Stefan,Chtourou, Hamdi,Engel, Florian A,Neumann, Rainer,Woll, Alexander | eng |
PMC4388468 | RESEARCH ARTICLE
The Correlation between Running Economy
and Maximal Oxygen Uptake: Cross-Sectional
and Longitudinal Relationships in Highly
Trained Distance Runners
Andrew J. Shaw1,2*, Stephen A. Ingham1, Greg Atkinson3, Jonathan P. Folland2
1 English Institute of Sport, Loughborough University, Loughborough, United Kingdom, 2 School of Sport,
Exercise and Health Sciences, Loughborough University, Loughborough, United Kingdom, 3 Health and
Social Care Institute, Teesside University, Middlesbrough, United Kingdom
* [email protected]
Abstract
A positive relationship between running economy and maximal oxygen uptake (V̇O2max)
has been postulated in trained athletes, but previous evidence is equivocal and could have
been confounded by statistical artefacts. Whether this relationship is preserved in response
to running training (changes in running economy and V̇O2max) has yet to be explored. This
study examined the relationships of (i) running economy and V̇O2max between runners,
and (ii) the changes in running economy and V̇O2max that occur within runners in response
to habitual training. 168 trained distance runners (males, n = 98, V̇O2max 73.0 ± 6.3
mLkg-1min-1; females, n = 70, V̇O2max 65.2 ± 5.9 mL kg-1min-1) performed a discontinuous
submaximal running test to determine running economy (kcalkm-1). A continuous incre-
mental treadmill running test to volitional exhaustion was used to determine V̇O2max 54 par-
ticipants (males, n = 27; females, n = 27) also completed at least one follow up assessment.
Partial correlation analysis revealed small positive relationships between running economy
and V̇O2max (males r = 0.26, females r = 0.25; P<0.006), in addition to moderate positive re-
lationships between the changes in running economy and V̇O2max in response to habitual
training (r = 0.35; P<0.001). In conclusion, the current investigation demonstrates that
only a small to moderate relationship exists between running economy and V̇O2max in highly
trained distance runners. With >85% of the variance in these parameters unexplained
by this relationship, these findings reaffirm that running economy and V̇O2max are primarily
determined independently.
Introduction
Running economy (RE) and maximal oxygen uptake (V̇O2max) are two of the primary determi-
nants of endurance running performance [1–4]. The combination of RE and V̇O2max, defined as
the velocity at V̇O2max (vV̇O2max), has been found to account for ~94% of the inter-individual
PLOS ONE | DOI:10.1371/journal.pone.0123101
April 7, 2015
1 / 10
OPEN ACCESS
Citation: Shaw AJ, Ingham SA, Atkinson G, Folland
JP (2015) The Correlation between Running
Economy and Maximal Oxygen Uptake: Cross-
Sectional and Longitudinal Relationships in Highly
Trained Distance Runners. PLoS ONE 10(4):
e0123101. doi:10.1371/journal.pone.0123101
Academic Editor: Oyvind Sandbakk, Norwegian
University of Science and Technology, NORWAY
Received: October 31, 2014
Accepted: February 27, 2015
Published: April 7, 2015
Copyright: This is an open access article, free of all
copyright, and may be freely reproduced, distributed,
transmitted, modified, built upon, or otherwise used
by anyone for any lawful purpose. The work is made
available under the Creative Commons CC0 public
domain dedication.
Data Availability Statement: All relevant data are
within the paper.
Funding: These authors have no support or funding
to report.
Competing Interests: The authors have declared
that no competing interests exist.
variance in running performance over 16.1 km [5]. Consequently, exceptional values of both RE
and V̇O2max are considered requirements for success in elite endurance competitions, and en-
durance runners strive to improve both parameters through training in order to maximise
performance. As the margin of success is extremely small in elite distance running, subtle en-
hancements in either parameter could result in substantial performance gains. Therefore,
understanding the relationship of RE and V̇O2max both between and within individuals is neces-
sary to understand and optimise performance.
Within cohorts of trained [6,7] and elite [8] distance runners, it has been suggested that a
superior RE, quantified as the submaximal oxygen uptake, is associated with a lower V̇O2max.
These findings have been used to postulate that superior economy compensates for a lower
V̇O2max in some individual to achieve a similar performance level [3,8,9]. However, these inves-
tigations have often been restricted to small sample sizes (<25 participants [3,6,8]), and the va-
lidity of their statistical techniques has been questioned due to the expression both variables
relative to body mass (i.e. mLkg-1min-1); creating a common divisor that is known to produce
spurious correlations [10]. Partial correlation analysis would provide an appropriate method to
account for the influence of body mass on both variables whilst avoiding statistical artefacts,
however this method has yet to be used to examine the relationship between RE and V̇O2max.
Furthermore, studies have solely employed oxygen cost (OC) as a measure of RE, rather than
the more valid and comprehensive measurement of energy cost (EC; [11]). Thus, whether a
genuine association exists between RE and V̇O2max remains unclear from the limited cross-
sectional observations to date.
Moreover, the concurrent alterations in RE and V̇O2max that occur within athletes over time
with training might further reveal if there is an inherent association between these variables,
whilst also informing the optimisation of both variables and thus performance. Previous inves-
tigations in well trained athletes have noted enhancements in cycling efficiency following
short-term, intensive endurance training, but with no change in V̇O2max evident [9,12,13]. In
contrast, a recent investigation reported an association between individual changes in cycling
efficiency and V̇O2max in response to endurance training and across a competitive season; de-
spite no change in mean group V̇O2max [14]. These preliminary findings highlight the signifi-
cance of this relationship for elite endurance athletes, as enhancements in either RE or V̇O2max
might only be achievable at the expense of the other variable. However, this previous investiga-
tion was limited to measurements of gross efficiency, with no data presented on movement
economy. Moreover, analysis of this longitudinal relationship was restricted to observations
within small cohorts of athletes, and with responses to run training yet to be explored.
The primary aim of the current investigation was to explore the cross-sectional relationship
between V̇O2max and RE, quantified as EC (OC data are also presented for comparative pur-
poses), within a large cohort of highly trained distance runners. The secondary aim was to ex-
amine the longitudinal relationship between the changes in V̇O2max and RE occurring within
athletes in response to endurance training.
Materials and Methods
Overview
The cross-sectional investigation involved retrospective analysis of data from 168 healthy en-
durance trained athletes with competitive distances ranging from 800m to the marathon
(males, n = 98; females, n = 70; Table 1), who undertook testing and monitoring as part of their
sport science support from the English Institute of Sport. The following tests were performed
after written informed consent was obtained as a part of sports science support provision,
with procedures approved by the Internal Review Board of English Institute of Sport. Of the
Correlation between Running Economy and V̇O2max
PLOS ONE | DOI:10.1371/journal.pone.0123101
April 7, 2015
2 / 10
participants assessed, 97 (males, n = 57; females, n = 40) were classed as middle distance run-
ners, defined by a primary competitive distance3000m [15], with 71 classed as long distance
runners (males, n = 41; females, n = 30). During the season following their final visit, athlete’s
best performance times in their primary competitive distance were 89.1±6.1% and 91.2±4.4%
of the current British record for males and females, respectively. Data were collected from two
laboratories, with all tests conducted as part of athlete support services between November
2004 and April 2013. Participants provided informed consent prior to physiological assess-
ments, in addition to an athlete agreement providing permission for the use of their data in
anonymous retrospective analysis. During each visit to the laboratory, participants completed
first submaximal and then maximal running assessments (detailed below). Participants wore
appropriate clothing (shorts and a vest or t-shirt) and racing shoes, and laboratory conditions
were similar throughout all running assessments (temperature 20.6±1.9°C, relative humidity
45.9±9.8%). As differences in RE and V̇O2max have been noted between sexes [16–18], males
and females were analysed separately for cross sectional analyses.
The longitudinal aspect of the study was based on 54 participants (males, n = 27; females,
n = 27) from amongst the larger cohort of 168 runners, that had completed at least one follow
up assessment, with a median trial separation of 203 days (range: 37–2567 days) in order to as-
sess within-athlete changes in both RE and V̇O2max over time. The number of repeat assessments
in the longitudinal analysis varied between participants, with a median of 3 visits per athlete
(range: 2–10 visits), summating to 182 assessments in total. No evidence is currently available re-
garding sex differences in the concurrent alterations in RE and V̇O2max in response to habitual
endurance training, thus data for males and females were combined for longitudinal analysis.
Protocol
Submaximal running assessments.
Following a warm-up (~10 min at 10–12 kmh-1),
participants completed a discontinuous submaximal incremental test consisting of six to nine
stages of 3 minutes continuous running, with increments of 1 kmh-1 on a motorised treadmill
of known belt speeds (HP cosmos Saturn, Traunstein, Germany) interspersed by 30 s rest peri-
ods for blood sampling. As the speeds assessed were typically between 10.5 kmh-1 and 18
kmh-1, treadmill gradient was maintained at 1% throughout submaximal assessments in order
to reflect the energetic cost of outdoor running [19]. This protocol has been shown to reliable
measures of running economy when quantified as both EC and OC (typical error ~3%; [20]).
Moreover, the controlled laboratory environment enabled assessments of EC whilst avoiding
the confounding influence of air resistance that is evident during outdoor running as speed
Table 1. Physiological and anthropometrical characteristics of athletes within the cross sectional and longitudinal investigations.
Cross sectional
Longitudinal sub-group
Females
Males
Females
Males
(n = 70)
(n = 98)
(n = 27)
(n = 27)
Age (yrs)
23±4
23±6
23±5
21±3
Body mass (kg)
55.2±4.7
67.1±7.1
55.4±4.3
66.6±6.0
Stature (cm)
169±5
179±7
168±4
179±6
V̇O2max (mLkg-1min-1)
65.2±5.9
73.0±6.3
64.5±4.9
73.6±5.9
vLTP (kmh-1)
15.5±1.2
17.2±1.3
15.7±1.2
17.6±1.1
Running economy (kcalkg-1km-1)
1.15±0.09
1.14±0.09
1.13±0.06
1.13±0.07
V̇O2max, maximal oxygen uptake; vLTP, velocity at lactate turnpoint.
doi:10.1371/journal.pone.0123101.t001
Correlation between Running Economy and V̇O2max
PLOS ONE | DOI:10.1371/journal.pone.0123101
April 7, 2015
3 / 10
increases [21]. Recent performance times of participants were used to determine an appropri-
ate starting speed to provide ~4 speeds prior to lactate turnpoint (LTP). Increments were
continued until blood lactate concentration had risen exponentially, typically defined as an in-
crease in blood lactate of ~2 mmolL-1 from the previous stage. HR (s610i, Polar, Finland) and
pulmonary gas exchange (detailed below) were monitored throughout the test.
Maximal running assessments.
V̇O2max was determined by a continuous incremental
treadmill running ramp test to volitional exhaustion. After a warm-up, participants initially
ran at a speed 2 kmh-1 below the final speed of the submaximal test and at a 1% gradient. Each
minute, the incline was increased by 1% until volitional exhaustion. The test duration was typi-
cally 6–8 minutes.
Measurements
Anthropometry.
Prior to exercise on laboratory visits, body mass was measured using dig-
ital scales (Seca 700, Seca, Hamburg, Germany) to the nearest 0.1 kg. Stature was recorded to
the nearest 1 cm using a stadiometer (Harpenden Stadiometer, Holtain Limited, UK).
Pulmonary gas exchange.
Breath-by-breath gas exchange data was quantified via an auto-
mated open circuit metabolic cart (Oxycon Pro, Carefusion, San Diego, USA). Participants
breathed through a low dead-space mask, with air sampled at 60 mLmin-1. Prior to each test,
two point calibrations of both gas sensors were completed, using a known gas mixture (16%
O2, 5% CO2) and ambient air. Ventilatory volume was calibrated using a 3 L (±0.4%) syringe.
This system has previously been shown to be a valid apparatus for the determination of oxygen
consumption (V̇O2) and carbon dioxide production (V̇CO2) at both low and maximal exercise
intensities [22]. As previous data from our laboratory has demonstrated a steady state V̇O2,
and V̇CO2 is achieved within the first 2 minutes of each stage for highly trained endurance run-
ners [20], mean values from breathe-by-breathe measures over the final 60 seconds of each
stage were used to quantify V̇O2, carbon dioxide production V̇CO2, and RER.
Blood lactate.
A 20μL capillary blood sample was taken from the earlobe for analysis of
blood lactate ([La]b) (Biosen C-line, EKF diagnostics, Germany). The LTP was identified via
the modified Dmax method [23]. LTP was quantified as the point on the third order polynomi-
al curve fitted to the speed-lactate relationship that generated the greatest perpendicular dis-
tance to the straight line formed between the stage proceeding an increase in [La]b greater than
0.4 mmol.L-1 (lactate threshold) and the final stage. The four stages prior to LTP were identi-
fied for each participant, with an average of these four stages used to quantify OC and EC.
Calculation of running economy.
V̇O2 and V̇CO2 during the final minute of each sub-
maximal stage were used to calculate EC. Updated nonprotein respiratory quotient equations
[24] were used to estimate substrate utilisation (gmin-1) during the monitored period. The
energy derived from each substrate was then calculated by multiplying fat and carbohydrate
usage by 9.75 kcal and 4.07 kcal, respectively, reflecting the mean energy content of the
metabolised substrates during moderate to high intensity exercise [25]. EC was quantified as
the sum of these values, expressed in kcalkm-1. V̇O2 during the final minute of each submaxi-
mal stage was used to determine oxygen cost (OC) in mLkm-1 to enable comparisons to
previous investigations.
Statistical analyses. Data are presented as mean±SD for all dependant variables. Data
analysis was conducted using SPSS for windows (v21; IBM Corporation, Armonk, NY). When
an individual visited the laboratory for repeated assessments, an average of the assessments
was calculated and used for the cross sectional analysis. Pearsons product-moment coefficients
were calculated to assess the relationship between body mass and EC, OC and V̇O2max. As body
mass was strongly related to both RE measures (EC, OC) and V̇O2max, partial correlations
Correlation between Running Economy and V̇O2max
PLOS ONE | DOI:10.1371/journal.pone.0123101
April 7, 2015
4 / 10
controlling for body mass, and associated 95% confidence intervals (CI), were used to assess
the relationship between absolute V̇O2max and both EC and OC. This method removes the in-
fluence of body mass on both RE and V̇O2max whilst avoiding spurious correlations created by
correlating two variables with a common divisor [26]. For graphical display of these relation-
ships, values of EC and V̇O2max adjusted for body mass for each individual were calculated
based on individual residuals. This involved summating the individual’s residual, in compari-
son to the cohort relationship with body mass (e.g. EC vs body mass), with the group mean for
that variable [27]. For the longitudinal analysis, in order to assess any relationships between
the changes over time in absolute V̇O2max and the changes in both EC and OC over repeat visits,
partial correlation coefficients were calculated using ANCOVA [28]; providing a comprehen-
sive model that accounts variations in both body mass and the number of visits per athlete.
Cohen's d effect size descriptors (trivial 0.0–0.1, small 0.1–0.3, moderate 0.3–0.5, large 0.5–0.7,
very large 0.7–0.9, nearly perfect 0.9–1, perfect 1) were used to infer correlation magnitude
[29]. Significance was accepted at P0.05.
Results
Participant Characteristics
Participant characteristics are shown in Table 1. The well trained status of the participants was
emphasised by the high V̇O2max and vLTP values for both males and females.
Cross-sectional analysis
Partial correlation analysis controlling for body mass, revealed small positive relationships be-
tween EC and V̇O2max (males r = 0.26, CI 0.07–0.44, P = 0.009; females r = 0.25, CI 0.02–0.46,
P = 0.036; Fig 1), and a moderate positive relationship between OC and V̇O2max (males r = 0.33,
CI 0.14–0.50, P = 0.001; females r = 0.33, CI 0.10–0.52, P = 0.006).
Longitudinal analysis
Partial correlation analysis from ANCOVA revealed moderate positive relationships between
the changes in EC and V̇O2max over time (r = 0.35; CI 0.19–0.49, P < 0.001; Fig 2), and changes
in OC and V̇O2max over time (r = 0.44; CI 0.29–0.57, P < 0.001).
Discussion
The present investigation explored the cross-sectional and longitudinal relationships between
RE and V̇O2max in a large cohort of highly trained distance runners. The major contribution
of this study to the field is that only a small to moderate association exists between RE and
V̇O2max (R2 ~ 12%) when body mass is appropriately accounted for. With >85% of the vari-
ance in these parameters unexplained by this relationship, these findings reaffirm that RE and
V̇O2max are primarily determined independently.
Cross-sectional analysis revealed a small positive between-participant relationships between
V̇O2max and the metabolic cost of running, when quantified as both EC (r ~ 0.25) and OC
(r ~ 0.33). These results support the findings of Pate et al. [7], who reported a similar relation-
ship (r = 0.29) between submaximal V̇O2 and V̇O2max in a similarly large cohort of habitual
distance runners. Conversely, a stronger, moderate positive relationship has been reported be-
tween submaximal V̇O2 and V̇O2max in smaller cohorts of elite distance runners (r = 0.59; [8])
and physically active individuals (r = 0.48; [30]). However, all aforementioned investigations
are confounded by statistical artefacts that arise when correlating two variables with common
divisors [10,26], and thus should be regarded with caution. Within the current study, spurious
Correlation between Running Economy and V̇O2max
PLOS ONE | DOI:10.1371/journal.pone.0123101
April 7, 2015
5 / 10
correlations between RE and V̇O2max were avoided by removing the influence of body mass
with partial correlations, which enabled the true relationship between these variables to be ex-
amined. As a lower metabolic cost is reflective of a more economical runner, our findings con-
firm the existence of a small inverse association between RE and V̇O2max in endurance runners.
The longitudinal analysis of the relationship between the changes in RE and the changes in
V̇O2max within participants in response to training has not previously been documented. Sup-
porting the findings from our cross sectional analysis, a moderate positive relationship
(r = 0.35) was observed between the changes in EC and V̇O2max over repeated assessments.
Fig 1. Scatter plot of energy cost (Kcalkm-1) adjusted for body mass (BM) vs V̇O2max (Lmin-1)
adjusted for BM for both females (A; n = 70; r = 0.25; P = 0.036) and males (B; n = 98; r = 0.26; P = 0.009)
within the cross-sectional analysis.
doi:10.1371/journal.pone.0123101.g001
Correlation between Running Economy and V̇O2max
PLOS ONE | DOI:10.1371/journal.pone.0123101
April 7, 2015
6 / 10
Moreover, these findings support recent observations from competitive road cyclists that
highlighted a similar moderate relationship (r = 0.44) between changes in gross efficiency and
V̇O2max across a training season [14].
It has been postulated that variations in lipid oxidation rates between individuals might, in
part, explain the relationship between OC and V̇O2max that some previous studies have docu-
mented; with a higher V̇O2max facilitating greater lipid oxidation and consequently a greater
OC during sub-maximal exercise [7]. Whilst OC may be sensitive to lipid oxidation, the calcula-
tion of EC includes the RER and thus is insensitive to differences in substrate metabolism. The
influence of substrate metabolism could conceivably explain the marginally stronger relation-
ship observed between OC and V̇O2max, than EC and V̇O2max, in both the cross sectional
(r ~ 0.33 vs r ~ 0.25) and longitudinal observations (r = 0.44 vs r = 0.35). More importantly, a
positive relationship was documented between EC and V̇O2max that is clearly independent of
variations in lipid metabolism.
The mechanisms that underpin the small relationship between EC and V̇O2max remain un-
clear. It has been argued that for athletes of a similar, high performance level, there would be
an inevitable relationship between EC and V̇O2max in order to produce a similar velocity at
V̇O2max [31]. However, we have found no evidence for this possibility, despite all the partici-
pants in this study being highly trained and high performing runners, perhaps in part because
of the variable performance ability of the athletes. It is also possible that less economical run-
ners recruit a larger muscle mass (braking, oscillation etc.) and it is conceivable that this could
contribute to a higher V̇O2max. However there is considerable evidence that V̇O2max during
whole body exercise such as running is largely dependent on oxygen delivery rather than utili-
sation [32], which might question this explanation. Thus, further investigation would be re-
quired to identify the factors driving the interdependence of EC and V̇O2max.
Though reaching statistical significance, the association between RE and V̇O2max was
small. The current study found only ~ 7% (between-participant cross sectional data) and 12%
Fig 2. Scatter plot of the changes over time in energy cost (Kcalkm-1) adjusted for body mass (BM) vs
the changes over time in V̇O2max (Lmin-1) adjusted for BM (r = 0.35; P < 0.001) within the longitudinal
analysis.
doi:10.1371/journal.pone.0123101.g002
Correlation between Running Economy and V̇O2max
PLOS ONE | DOI:10.1371/journal.pone.0123101
April 7, 2015
7 / 10
(within-participant longitudinal data) of the variance in RE was explained by V̇O2max. This
small association likely reflects the distinct nature of these variables and their physiological
determinants. V̇O2max is known to be determined by factors such as cardiac output [33], total
haemoglobin mass [34], and mitochondrial capacity [1]. Conversely, RE is thought to be
closely associated to multiple biomechanical and anthropometrical factors, including effective
storage and re-utilisation of elastic energy [35,36], vertical oscillation [37] and ground contact
time [38]. As there are few common determinants of both RE and V̇O2max, adaptations
that lead to enhancements in one of these variables are unlikely to directly influence in the
opposing variable.
In conclusion, the current investigation demonstrates that only a small to moderate rela-
tionship exists between running economy and V̇O2max in highly trained distance runners.
With >85% of the variance in these parameters unexplained by this relationship, these findings
reaffirm that running economy and V̇O2max are primarily determined independently.
Acknowledgments
The authors would like to thank Dr Barry Fudge, Dr Jamie Pringle, Dr Charles Pedlar and Kate
Spilsbury for their time and efforts collecting data on behalf of the English Institute of Sport
that enabled this retrospective analysis, in addition to David Green for his time and assistance
during data collation.
Author Contributions
Conceived and designed the experiments: AJS JPF SAI. Performed the experiments: AJS JPF
SAI. Analyzed the data: AJS JPF SAI GA. Contributed reagents/materials/analysis tools: AJS
JPF SAI GA. Wrote the paper: AJS JPF SAI GA.
References
1.
Di Prampero P (2003) Factors limiting maximal performance in humans. Eur J Appl Physiol 90:
420–429. doi: 10.1007/s00421-003-0926-z PMID: 12910345
2.
Ingham S, Whyte G, Pedlar C, Bailey D, Dunman N, Nevill A (2008) Determinants of 800-m and 1500-
m running performance using allometric models. Med Sci Sports Exerc 40: 345–350. doi: 10.1249/
mss.0b013e31815a83dc PMID: 18202566
3.
Lucía A, Hoyos J, Pérez M, Santalla A, Chicharro J (2002) Inverse relationship between VO2max and
economy/efficiency in world-class cyclists. Med Sci Sports Exerc 34: 2079–2084. doi: 10.1249/01.
MSS.0000039306.92778.DF PMID: 12471319
4.
Joyner MJ (1991) Modeling: optimal marathon performance on the basis of physiological factors. J Appl
Physiol 70: 683–687. PMID: 2022559
5.
McLaughlin JE, Howley ET, Bassett DR, Thompson DL, Fitzhugh EC (2010) Test of the classic model
for predicting endurance running performance. Med Sci Sports Exerc 42: 991–997. doi: 10.1249/MSS.
0b013e3181c0669d PMID: 19997010
6.
Fletcher JR, Esau SP, Macintosh BR (2009) Economy of running: beyond the measurement of oxygen
uptake. J Appl Physiol 107: 1918–1922. doi: 10.1152/japplphysiol.00307.2009 PMID: 19833811
7.
Pate R, Macera C, Bailey S, Bartoli W, Powell K (1992) Physiological, anthropometric, and training cor-
relates of running economy. Med Sci Sports Exerc 24: 1128–1133. PMID: 1435160
8.
Morgan DW, Daniels JT (1994) Relationship between VO2max and the aerobic demand of running in
elite distance runners. Int J Sports Med 15: 426–429. doi: 10.1055/s-2007-1021082 PMID: 8002123
9.
Santalla A, Naranjo J, Terrados N (2009) Muscle efficiency improves over time in world-class cyclists.
Med Sci Sports Exerc 41: 1096–1101. doi: 10.1249/MSS.0b013e318191c802 PMID: 19346977
10.
Atkinson G, Davison R, Passfield L, Nevill A (2003) Could the correlation between maximal oxygen up-
take and “economy” be spurious? Med Sci Sports Exerc 35: 1242–1243. doi: 10.1249/01.MSS.
0000074560.08128.5D PMID: 12840650
11.
Shaw AJ, Ingham SA, Folland JP (2014) The valid measurement of running economy in runners. Med
Sci Sports Exerc 46: 1968–1973. doi: 10.1249/MSS.0000000000000311 PMID: 24561819
Correlation between Running Economy and V̇O2max
PLOS ONE | DOI:10.1371/journal.pone.0123101
April 7, 2015
8 / 10
12.
Hopker J, Coleman D, Passfield L (2009) Changes in cycling efficiency during a competitive season.
Med Sci Sports Exerc 41: 912–919. doi: 10.1249/MSS.0b013e31818f2ab2 PMID: 19276841
13.
Iaia F, Hellsten Y, Nielsen J, Fernstrom M, Sahlim K, Bangsbo J (2009) Four weeks of speed endur-
ance training reduces energy expenditure during exercise and maintains muscle oxidative capacity de-
spite a reduction in training volume. J Appl Physiol 106: 73–80. doi: 10.1152/japplphysiol.90676.2008
PMID: 18845781
14.
Hopker J, Coleman D, Jobson S, Passfield L (2012) Inverse Relationship between V_O2max and Gross
Efficiency. Int J Sports Med 33: 789–794. PMID: 22562732
15.
Brandon LJ (1995) Physiological factors associated with middle distance running performance. Sport
Med 19: 268–277. PMID: 7604199
16.
Daniels J, Daniels N (1992) Running economy of elite male and elite female runners. Med Sci Sports
Exerc 24: 483–489. PMID: 1560747
17.
Helgerud J, Støren O, Hoff J (2010) Are there differences in running economy at different velocities for
well-trained distance runners? Eur J Appl Physiol 108: 1099–1105. doi: 10.1007/s00421-009-1218-z
PMID: 20024579
18.
Helgerud J (1994) Maximal oxygen uptake, anaerobic threshold and running economy in women and
men with similar performances level in marathons. Eur J Appl Physiol Occup Physiol 68: 155–161.
PMID: 8194545
19.
Jones A, Doust J (1996) A 1% treadmill grade most accurately reflects the energetic cost of outdoor
running. J Sports Sci 14: 321–327. PMID: 8887211
20.
Shaw AJ, Ingham SA, Fudge BW, Folland JP (2013) The reliability of running economy expressed as
oxygen cost and energy cost in trained distance runners. Appl Physiol Nutr Metab 38: 1268–1272. doi:
10.1139/apnm-2013-0055 PMID: 24195628
21.
Pugh L (1970) Oxygen intake in track and treadmill running with observations on the effect of air resis-
tance. J Physiol: 823–835. PMID: 4322767
22.
Rietjens G, Kuipers H, Kester A, Keizer H (2001) Validation of a computerized metabolic measurement
system (Oxycon-Pro) during low and high intensity exercise. Int J Sports Med 22: 291–294. PMID:
11414673
23.
Bishop D, Jenkins D, Mackinnon L (1998) The relationship between plasma lactate parameters, Wpeak
and 1-h cycling performance in women. Med Sci Sport Exerc Exerc 30: 1270–1275. PMID: 9710868
24.
Péronnet F, Massicotte D (1991) Table of nonprotein respiratory quotient: an update. Can J Sport Sci
16: 23–29. PMID: 1645211
25.
Jeukendrup AE, Wallis GA (2005) Measurement of substrate oxidation during exercise by means of
gas exchange measurements. Int J Sports Med 26: S28–37. doi: 10.1055/s-2004-830512 PMID:
15702454
26.
Pearson K (1896) On a form of spurious correlation which may arise when indices are used in the mea-
surement of organs. Proc R Soc London: 489–498.
27.
Moya-Laraño J, Corcobado G (2008) Plotting partial correlation and regression in ecological studies.
Web Ecol 8: 35–46. doi: 10.5194/we-8-35-2008
28.
Bland JM, Altman DG (2009) Statistics notes Calculating correlation coefficients with repeated observa-
tions: Part 1—correlation within subjects. Br Med J 446: 1–7.
29.
Hopkins WG, Marshall SW, Batterham AM, Hanin J (2009) Progressive statistics for studies in sports
medicine and exercise science. Med Sci Sports Exerc 41: 3–13. doi: 10.1249/MSS.
0b013e31818cb278 PMID: 19092709
30.
Sawyer B, Blessinger J (2010) Walking and running economy: inverse association with peak oxygen
uptake. Med Sci Sports 42: 2122–2127. doi: 10.1249/MSS.0b013e3181de2da7.Walking PMID:
20351592
31.
Noakes T, Tucker R (2004) Inverse Relationship Between VO2Max and Economy in World-Class Cy-
clists. Med Sci Sport Exerc 36: 1083–1084. doi: 10.1249/01.MSS.0000128140.85727.2D PMID:
15179181
32.
Wagner P (2000) New Ideas on Limitations to Vo2max. Exerc Sport Sci Rev 28: 10–14. PMID: 11131681
33.
Blomqvist CG, Saltin B (1983) Cardiovascular adaptations to physical training. Annu Rev Physiol 45:
169–189. doi: 10.1146/annurev.ph.45.030183.001125 PMID: 6221687
34.
Schmidt W, Prommer N (2008) Effects of various training modalities on blood volume. Scand J Med Sci
Sports 18: 57–69. doi: 10.1111/j.1600-0838.2008.00833.x PMID: 18665953
35.
Hunter GR, Katsoulis K, McCarthy JP, Ogard WK, Bamman MM, Wood D, et al. (2011) Tendon length
and joint flexibility are related to running economy. Med Sci Sports Exerc 43: 1492–1499. doi: 10.1249/
MSS.0b013e318210464a PMID: 21266930
Correlation between Running Economy and V̇O2max
PLOS ONE | DOI:10.1371/journal.pone.0123101
April 7, 2015
9 / 10
36.
Scholz MN, Bobbert MF, van Soest AJ, Clark JR, van Heerden J (2008) Running biomechanics: shorter
heels, better economy. J Exp Biol 211: 3266–3271. doi: 10.1242/jeb.018812 PMID: 18840660
37.
Tartaruga MP, Brisswalter J, Peyré-Tartaruga LA, Avila AOV, Alberton CL, Coertjens M, et al. (2012)
The relationship between running economy and biomechanical variables in distance runners. Res
Q Exerc Sport 83: 367–375. PMID: 22978185
38.
Di Michele R, Merni F (2014) The concurrent effects of strike pattern and ground-contact time on run-
ning economy. J Sci Med Sport 17: 414–418. doi: 10.1016/j.jsams.2013.05.012 PMID: 23806876
Correlation between Running Economy and V̇O2max
PLOS ONE | DOI:10.1371/journal.pone.0123101
April 7, 2015
10 / 10
| The correlation between running economy and maximal oxygen uptake: cross-sectional and longitudinal relationships in highly trained distance runners. | 04-07-2015 | Shaw, Andrew J,Ingham, Stephen A,Atkinson, Greg,Folland, Jonathan P | eng |
PMC6040767 | RESEARCH ARTICLE
The effects of beetroot juice supplementation
on exercise economy, rating of perceived
exertion and running mechanics in elite
distance runners: A double-blinded,
randomized study
Carlos Balsalobre-Ferna´ndez1,2☯*, Blanca Romero-Moraleda2,3☯, Rocı´o Cupeiro2‡, Ana
Bele´n Peinado2‡, Javier Butragueño2‡, Pedro J. Benito2‡
1 Department of Physical Education, Sport and Human Movement, Universidad Auto´noma de Madrid,
Madrid, Spain, 2 Laboratory of Exercise Physiology Research Group, Department of Health and Human
Performance, School of Physical Activity and Sport Sciences-INEF, Universidad Politecnica de Madrid,
Madrid, Spain, 3 Faculty of Health, Camilo Jose´ Cela University, Madrid, Spain
☯ These authors contributed equally to this work.
‡ These authors also contributed equally to this work.
* [email protected]
Abstract
Purpose
Nitrate-rich beetroot juice supplementation has been extensively used to increase exercise
economy in different populations. However, its use in elite distance runners, and its potential
effects on biomechanical aspects of running have not been properly investigated. This study
aims to analyze the potential effects of 15 days of beetroot juice supplementation on physio-
logical, psychological and biomechanical variables in elite runners.
Methods
Twelve elite middle and long-distance runners (age = 26.3 ± 5.1yrs, VO2Max = 71.8±5.2
ml*kg-1*min-1) completed an incremental running test to exhaustion on a treadmill before
and after a 15-days supplementation period, in which half of the group (EG) consumed a
daily nitrate-rich beetroot juice and the other group (PG) consumed a placebo drink. Time to
exhaustion (TEx), running economy, vastus lateralis oxygen saturation (SmO2), leg stiffness
and rate of perceived exertion (RPE) were measured at 15, 17.1 and 20 km/h during the
incremental test.
Results
Likely to very likely improvements in EG were observed for the RPE (Standardized mean
difference (SMD) = -2.17, 90%CI = -3.23, -1.1), SmO2 (SMD = 0.72, 90%CI = 0.03, 1.41)
and TEx (SMD = 1.18, 90%CI = -0.14, 2.5) in comparison with PG. No other physiological
or biomechanical variable showed substantial improvements after the supplementation
period.
PLOS ONE | https://doi.org/10.1371/journal.pone.0200517
July 11, 2018
1 / 10
a1111111111
a1111111111
a1111111111
a1111111111
a1111111111
OPEN ACCESS
Citation: Balsalobre-Ferna´ndez C, Romero-
Moraleda B, Cupeiro R, Peinado AB, Butragueño J,
Benito PJ (2018) The effects of beetroot juice
supplementation on exercise economy, rating of
perceived exertion and running mechanics in elite
distance runners: A double-blinded, randomized
study. PLoS ONE 13(7): e0200517. https://doi.org/
10.1371/journal.pone.0200517
Editor: Gordon Fisher, University of Alabama at
Birmingham, UNITED STATES
Received: March 15, 2018
Accepted: June 26, 2018
Published: July 11, 2018
Copyright: © 2018 Balsalobre-Ferna´ndez et al. This
is an open access article distributed under the
terms of the Creative Commons Attribution
License, which permits unrestricted use,
distribution, and reproduction in any medium,
provided the original author and source are
credited.
Data Availability Statement: Data are available
from figshare (https://doi.org/10.6084/m9.figshare.
5873190.v1).
Funding: The authors received no specific funding
for this work.
Competing interests: The authors have declared
that no competing interests exist.
Conclusions
Fifteen days of nitrate-rich beetroot juice supplementation produced substantial improve-
ments in the time to exhaustion in elite runners; however, it didn’t produce meaningful
improvements in running economy, VO2Max or mechanical parameters.
Introduction
Performance at the highest level of competition in endurance events depends on several physi-
ological, psychological and biomechanical parameters [1–5]. Even marginal gains in some var-
iables might represent a competitive edge to win the race or break records; for instance,
wearing energy-recovering shoes have shown to increase running economy by about 4%,
which might have helped in the recent first sub-2h attempt on the marathon [6]. Maximal oxy-
gen consumption is one of the most relevant physiological variables in distance running per-
formance [7,8]; however, in elite athletes other variables seem to influence competitive success
to a greater extent [4,9]. Among them, running economy (i.e., the oxygen consumption at a
certain pace) has been proposed as one of the most important variables in distance running
performance. Theoretically, better running economy means either being able to run at the
same pace with a lower exertion or, for the same energy consumption, being able to run faster
[9,10]. Therefore, researchers and trainers are focused now on finding strategies to increase
economy in different kinds of endurance athletes such as runners, cyclists or swimmers. In
this sense, different studies have shown that heavy-loaded resistance training produces a
meaningful increase in running economy of athletes at different levels, including elite partici-
pants [11,12]. Besides training interventions, different nutritional strategies have also been
used to increase distance running performance, from high-carbohydrate to ketogenic diets
[13–15]. However, during the last decade one particular supplement has been investigated as a
potential beneficial aid for endurance events: beetroot juice [16,17]. Beetroot juice is a high
source of nitrate (NO3
-) that has been proposed to increase NO (nitric oxide) availability, a
potent signaling molecule that can increase vasodilatation, mitochondrial respiration or glu-
cose uptake, among other factors related to exercise performance [17]. Several studies have
shown that beetroot juice supplementation for about a week does have an ergogenic effect in
moderate trained runners, mainly via an increase in running economy [16,18,19]. However,
there is a lack of studies analyzing its potential benefits in elite subjects, who might need a
higher dose or exposure than recreational athletes to obtain the desired increases in perfor-
mance. A recent study [20] observed that 8 days of beetroot supplementation didn’t increase
performance variables in elite 1500m runners; however, whether larger interventions might
have benefits in elite populations is unknown. Moreover, to the best of our knowledge, no
studies have analyzed the potential effects of beetroot supplementation in mechanical aspects
of running such as leg stiffness, a variable that has been linked to running economy in different
studies [21–23]. For this, the purpose of this study is to analyze the physiological, psychological
and biomechanical effects of beetroot juice supplementation in elite distance runners.
Materials & methods
Participants
Twelve elite middle and long-distance runners, including European medalists and Olympians
were recruited from the High Performance Center of Madrid, Spain to join this investigation.
Inclusion criteria were as follows: male competitors in national and international events, with
Beetroot juice supplementation and exercise economy, rating of perceived exertion and running mechanics
PLOS ONE | https://doi.org/10.1371/journal.pone.0200517
July 11, 2018
2 / 10
personal bests in the urban 10k below 32 min, and below 3min50s. in the 1500m. See details in
Table 1. Prior to commencement, all participants provided written informed consent. Participants
were instructed to not consume beetroot of any form 2 months before the start of the intervention
and to maintain their normal diet throughout the testing period, to follow the same diet for 24
hours prior to each trial, to avoid food and drink in the hour before each trial, and to refrain from
strenuous exercise for 24 hours before each trial. None of the participants followed a plant-based
(vegetarian or vegan) diet, and they never consumed nitrate-rich beetroot juice in the past. The
study protocol complied with the Declaration of Helsinki for Human Experimentation and was
approved by the Institutional Review Board of the Universidad Camilo Jose´ Cela.
Procedures
Beetroot juice supplementation protocol. Participants were randomly assigned to either
a placebo (PG) or an experimental (EG) group in a double-blinded manner so neither the ath-
letes nor the main researchers could possibly identify who were consuming a nitrate-rich beet-
root juice supplementation. See Fig 1.
Participants in EG consumed a high-nitrate beetroot juice (6.5-mmol NO3-/70-mL, Beet It
Sport; James White Drinks, Ipswich, United Kingdom) for 15 consecutive days with their
breakfast meal, while participants in PG consumed a nitrate-free placebo from the same manu-
facturer (0.065-mmol NO3-/70-mL Beet It Sport; James White Drinks, Ipswich, United King-
dom) for the same amount of time. Placebo drinks had the exact same packaging, color, smell
and taste as its nitrate-rich counterpart, so neither the athletes or the researchers could know
who were consuming what. Participants were instructed to not use mouth rinse during the
supplementation period.
Physiological parameters
Incremental running test. The maximal graded test was performed with a computerized
treadmill (H/P/COSMOS 3PW 4.0, H/P/Cosmos Sports & Medical, Nussdorf-Traunstein,
Germany) to determine each subject’s maximal VO2, and to measure running economy at dif-
ferent speeds. Expired gases were measured breath-by-breath with the Jaeger Oxycon Pro gas
analyser (Erich Jaeger, Viasys Healthcare, Germany). Heart response was continuously moni-
tored with a 12-lead ECG. Participants started with 3 minutes warm-up at 10 km/h. Then the
protocol continued with 3 steady state steps of 3 min each, at 15, 17.1 and 20 km/h. Subse-
quently the speed was increased to 0.2 km/h every 12 seconds until exhaustion. A slope of 1%
was set throughout the test [24]. All the tests fulfilled at least 2 of the following criteria [2]: A
respiratory exchange ratio (RER) >1.10, a plateau in VO2 (corresponding to a variation of,
100 mlmin-1) despite an increase in exercise intensity and a peak HR>220-age. This test was
conducted at baseline and repeated 24-h after the 15th day of supplementation, so there were
at least 24h after the last beetroot juice ingestion.
O2 saturation measurement.
Throughout all the graded tests O2 saturation (SmO2, in %)
was registered using near-infrared spectroscopy by placing a wearable device on the belly of
the vastus laterallis (Moxy Monitor, Minnesota, USA). In this study, a wearable device was
used to measure the muscle oxygenation in the vastus lateralis of dominant leg (along the
Table 1. Characteristics of the participants. Personal bests are expressed as range.
Age (yrs.)
Height (cm)
Weight (kg)
VO2Max (mlkg-1min-1)
PB in urban 10 km (min:s)
PB in outdoor 1500m (min:s)
Experimental group
27.3±7.8
1.79±0.07
69.2±8.6
69.1±5.3
29:22–31:00
3:32–3:45
Control group
24.2±2.9
1.78±0.04
65.2±2.6
72.3±6.8
28:28–30:54
3:43–3:45
https://doi.org/10.1371/journal.pone.0200517.t001
Beetroot juice supplementation and exercise economy, rating of perceived exertion and running mechanics
PLOS ONE | https://doi.org/10.1371/journal.pone.0200517
July 11, 2018
3 / 10
vertical axis of the thigh, approximately 10–12 cm above the knee joint [25]. The device was
wrapped by an elastic tight without occluding the blood flow. Black bandages covered the
device to eliminate background light [26].
Biomechanical parameters
To measure leg stiffness the validated Runmatic app v2.8 installed on an iPhone 6 with iOS
10.3.3 was used following the procedures described elsewhere [27]. The Runmatic app was
used to record leg stiffness during the first minute of each steady state on the incremental test
(i.e., at 15, 17.1 and 20 km/h), and mean scores were registered.
Psychological parameters
At the completion of each steady state step of the incremental running test (i.e., during the last
10 seconds of the 3-min run at 15, 17.1 and 20 km/h) participants were asked to rate their RPE
Fig 1. CONSORT flowchart.
https://doi.org/10.1371/journal.pone.0200517.g001
Beetroot juice supplementation and exercise economy, rating of perceived exertion and running mechanics
PLOS ONE | https://doi.org/10.1371/journal.pone.0200517
July 11, 2018
4 / 10
in a 1–10 scale. Before starting the incremental test, participants were given instruction to
point to a number on a sheet held by a researcher corresponding to their RPE at each step dur-
ing the trial. Participants were told that 1 means no exertion at all, and 10 means maximal
exertion.
Statistical analyses
Data is presented as mean with standard deviations. The non-parametric Mann-Whitney’s U
statistical test was used to compare the pre-post differences between the experimental and pla-
cebo groups. Also, the standardized mean differences (SMD) with the corresponding 90% con-
fidence interval (CI) were calculated using a magnitude-based inference approach [28]. The
criteria for interpreting the magnitude of the SMD were: trivial (<0.2), small (0.2–0.6), moder-
ate (0.6–1.0) and large (>1.0). Probabilities were also determined to establish whether true dif-
ferences were lower, similar or higher than the smallest worthwhile change (0.2 x between–
participant SD). Quantitative chances of better or worse effects were assessed qualitatively as
follows: <1%, almost certainly not; 1–5%, very unlikely; 5–25%, unlikely; 25–75%, possible;
75–95%, likely; 95–99%, very likely; and >99%, almost certain. If the changes of better or
worse were both >5%, the true difference was assessed as unclear. Statistical analyses were per-
formed using a custom spreadsheet [28] and SPSS 24 for Mac. The level of asymptotic signifi-
cance was set as p< 0.05.
Results
Participants in EG showed unclear improvements in heart rate (SMD = 0.03, 90%CI = -0.36,
0.42) and running economy expressed in VO2 (SMD = 0.36, 90%CI = -0.5, 1.21) in compari-
son with PG. Also, changes in the VO2Max did not substantially differ in EG in comparison
with PG after the supplementation period (SMD = 0.04, 90%CI = -0.87, 0.95). However, likely
to very likely improvements in EG vs. PG were observed for the RPE (SMD = -2.17, 90%CI =
-3.23, -1.1), SmO2 (SMD = 0.72, 90%CI = 0.03, 1.41) (in the three steady stages) and time to
exhaustion (SMD = 1.18, 90%CI = -0.14, 2.5). Finally, no substantial improvements in leg stiff-
ness (SMD = -0.14, 90%CI = -0.44, 0.17) were observed during any of the steady stages of the
incremental test as revealed by the low to trivial SMD and the range of the 90%CI. See Fig 2,
Table 2 and Table 3 for more details.
Discussion
The results of our study showed low to trivial improvements in different physiological mea-
sures such as running economy at different paces after 15-days of nitrate-rich beetroot juice
supplementation in elite distance runners. These results are in line with other experiments
conducted with elite distance athletes, in which submaximal exercise economy was not altered
after the supplementation period either [20,29]. The supplementation period in our study was
about twice larger than other experiments with elite runners; however, more days of ingestion
of nitrate-rich beetroot juice supplementation doesn’t seem to boost the adaptations in sub-
maximal exercise economy that this supplement has extensively showed to produce in less
trained populations [17,18].
A novel contribution from our study was the analysis of the potential benefits of beetroot
juice ingestion on different biomechanical variables that has been linked to running perfor-
mance in the literature, such as leg stiffness [23,30]. For example, leg stiffness has showed to be
negatively associated with running economy, since, in theory, a stiffer lower limb would store
a higher amount of elastic energy that would assist in the force production during running
and, therefore, it would reduce the cost of the exercise [9,23]. Results in our study showed
Beetroot juice supplementation and exercise economy, rating of perceived exertion and running mechanics
PLOS ONE | https://doi.org/10.1371/journal.pone.0200517
July 11, 2018
5 / 10
trivial effects of beetroot juice supplementation in leg stiffness, similar to submaximal running
economy.
Contrary to the results discussed above, likely to very likely large improvements were
observed both in the rate of perceived exertion and the time to exhaustion in the experimental
group in comparison with the placebo group after the supplementation period, where athletes
who consumed nitrate-rich beetroot juice endured more time before voluntary stopping the
incremental test on the treadmill. One potential explanation for this observation is the moder-
ate increase in the SmO2 of the vastus lateralis of the athletes from the experimental group in
comparison with the control group after the supplementation period. Participants who con-
sumed nitrate-rich beetroot supplementation had larger percent of oxygen saturation in their
muscles than their counterparts during exercise. That could have limited the accumulation of
fatigue-related metabolites and reduce the depletion of PCr, which, in the end, could increase
the time to exhaustion. NIRS is used for providing information about performance and oxygen
muscle function and to evaluate response before, during and after exercise. In our study, the
initial-during-end-exercise SmO2 was greater in experimental group. That finding is related
Fig 2. Forest plot with standardized mean differences (SMD) and 90% confidence interval (CI) for the rate of perceived exertion
(RPE) and oxygen saturation (SmO2) at different running paces. Lower scores (i.e., to the left in the X-axis) means lower scores in the
experimental group.
https://doi.org/10.1371/journal.pone.0200517.g002
Beetroot juice supplementation and exercise economy, rating of perceived exertion and running mechanics
PLOS ONE | https://doi.org/10.1371/journal.pone.0200517
July 11, 2018
6 / 10
with recovery capacity and improvements in performance status delaying the fatigue [31].
Then, greater availability of O2 could have impacted RPE and, finally it could have led to the
observed increase in the time to exhaustion in the experimental group. For example, several
studies have observed that increased oxygen availability via hyperoxia enhances exercise per-
formance [32]. Nevertheless, further investigations are needed to better understand the mecha-
nisms by which beetroot juice supplementation could reduce RPE and increase time to
exhaustion in an incremental test on a treadmill in elite distance runners.
Summarizing, results in our study are in line with other investigations that observed no sig-
nificant increases in running economy after a supplementation period with nitrate-rich beet-
root juice in elite athletes [20,29]; however, the large increase in the time to exhaustion, the
reduction in RPE and the increase in the oxygen saturation of the vastus lateralis during exer-
cise, observed after 15 consecutive days of nitrate-rich beetroot supplementation provide
novel information about the effects of this nutritional aid. Further investigations are required
Table 2. Mean values and standard deviations for the experimental and placebo groups before (Pre) and after (Post) the supplementation period.
Experimental group
Placebo group
Pre
Post
Pre
Post
Biomechanical parameters
Leg stiffness (kN/m)
15 km/h
7.9 ± 0.9
8.6 ± 1.4
8.4 ± 1.2
8.5 ± 1.4
17.1 km/h
8.5 ± 1.2
8.7 ± 1.2
7.9 ± 1.1
8.3 ± 1.3
20 km/h
8.9 ± 0.8
9.2 ± 1.3
7.5 ± 1.0
8.1 ± 0.8
Physiological parameters
RE (mlkgmin-1)
15 km/h
50.8 ± 3.5
52.6 ± 2.8
52.3 ± 3.3
53.7 ± 3.8
17.1 km/h
58.3 ± 2.7
60.4 ± 2.3
60.1 ± 3.8
62.1 ± 3.3
20 km/h
67.3 ± 4.4
69.5 ± 2.9
70.5 ± 5.1
71.4 ± 4.8
RER
15 km/h
0.87 ± 0.02
0.83 ± 0.02
0.91 ± 0.02
0.87 ± 0.01
17.1 km/h
0.93 ± 0.02
0.90 ± 0.03
0.96 ± 0.02
0.94 ± 0.02
20 km/h
1.03 ± 0.01
0.97 ± 0.04
1.07 ± 0.03
1.03 ± 0.03
HR (bpm)
15 km/h
147.2 ± 13.7
146.9 ± 11.2
157.3 ± 10.0
151.4 ± 12.4
17.1 km/h
160.6 ± 12.2
157.3 ± 9.2
168.1 ± 8.1
165.9 ± 7.1
20 km/h
164.2 ± 19.1
168.4 ± 11.4
164.6 ± 21.1
177.1 ± 4.6
SmO2 (%)
15 km/h
41.7 ± 7.1
47.7 ± 3.9
46.6 ± 15.5
43.7 ± 5.9
17.1 km/h
34.4 ± 3.5
41.0 ± 2.0
38.2 ± 16.4
37.8 ± 5.2
20 km/h
24.4 ± 2.8
31.0 ± 6.9
26.8 ± 12.1
27.7 ± 4.8
TEx (s)
1173.0 ± 87.1
1269.0 ± 53.6
1251.0 ± 52.6
1230 ± 73.5
VO2Max (mlkgmin-1)
69.1 ± 5.3
70.1 ± 7.0
72.3 ± 6.8
74.9 ± 6.1
Psychological parameters
RPE
15 km/h
2.4 ± 0.5
2.3 ± 0.5
2.4 ± 0.5
3.5 ± 0.8
17.1 km/h
3.8 ± 0.8
4.0 ± 0.8
4.4 ± 0.5
5.6 ± 0.8
20 km/h
6.0 ± 0.7
6.0 ± 1.0
7.2 ± 0.4
7.7 ± 1.0
Abbreviations: RE = running economy; VO2Max = maximal oxygen consumption; RER = respiratory exchange ratio; TEx = time to exhaustion; HR = heart rate;
SmO2 = saturation of muscle O2; RPE = rate of perceived exertion
https://doi.org/10.1371/journal.pone.0200517.t002
Beetroot juice supplementation and exercise economy, rating of perceived exertion and running mechanics
PLOS ONE | https://doi.org/10.1371/journal.pone.0200517
July 11, 2018
7 / 10
to confirm the findings of the present study, as well as to better understand the mechanisms
behind these potential benefits for elite distance runners.
Practical application and conclusions
Fifteen consecutive days of nitrate-rich beetroot juice supplementation didn’t increase running
economy, locomotion mechanics or muscular power in elite distance runners. However, time to
exhaustion in an incremental test on a treadmill, as well as rate of perceived exertion at different
running paces were meaningfully higher and lower in the experimental group in comparison with
the placebo group, respectively. These results might be explained in part by a higher muscle oxy-
gen saturation observed in the vastus lateralis of the runners from the experimental group in com-
parison with the control group. Anyhow, one of the most performance-related variables in
distance running (i.e., time to exhaustion) showed large improvements after 15-days of nitrate-
rich beetroot juice supplementation. These results could have potential practical application for
elite distance runners seeking nutritional strategies to improve their running performance.
Supporting information
S1 File. CONSORT 2010 checklist of information to include in a randomized controlled
trial.
(DOC)
S2 File. Study protocol as approved by the ethics committee.
(PDF)
Author Contributions
Conceptualization: Carlos Balsalobre-Ferna´ndez, Blanca Romero-Moraleda.
Data curation: Carlos Balsalobre-Ferna´ndez, Blanca Romero-Moraleda.
Formal analysis: Carlos Balsalobre-Ferna´ndez.
Table 3. Pre-post differences on the studied variables in the placebo vs experimental conditions.
SMD (90%CI)
Chances of being beneficial/trivial/harmful
Qualitative inference
% of change (experimental vs placebo)
P
Biomechanical parameters
Leg stiffness
-0.14 (-0.44, 0.17)
4/61/36
Possibly
2.25 vs 4.64
0.310
Physiological parameters
RE
0.36 (-0.5, 1.21)
63/24/13
Unclear
7.07 vs 4.5
0.589
VO2Max
0.04 (-0.87, 0.95)
37/31/32
Unclear
4.41 vs. 3.66
1.000
Av_RER
-0.2 (-1.15, 0.74)
50/27/22
Unclear
-4.7 vs. -3.6
1.000
TEx (s)
1.18 (-0.14, 2.5)
90/6/4
Likely
8.18 vs 1.6
0.310
Av_HR
0.03 (-0.36, 0.42)
22/64/15
Unclear
-0.26 vs. 1.14
0.792
SmO2
0.72 (0.03, 1.41)
91/7/2
Likely
17.8 vs. -2.64
0.167
Psychological parameters
RPE
-2.17 (-3.23, -1.1)
100/0/0
Very likely
-9.0 vs 20.4
0.016
Standardized mean differences (SMD) with 90% Confidence Intervals (CI) express the magnitude of the difference on the pre-post changes between the experimental
and placebo groups. Positive values reflect higher increments on the placebo group after the supplementation period, while negative scores reflect lower increments on
the experimental group after the supplementation period. P is the asymptotic significance of the Mann-Whiteny’s U test.
Abbreviations: RE = running economy; VO2Max = maximal oxygen consumption; Av_RER = average respiratory exchange ratio (whole test); TEx = time to exhaustion;
Av_HR = average heart rate (whole test); SmO2 = saturation of muscle O2; RPE = rate of perceived exertion
https://doi.org/10.1371/journal.pone.0200517.t003
Beetroot juice supplementation and exercise economy, rating of perceived exertion and running mechanics
PLOS ONE | https://doi.org/10.1371/journal.pone.0200517
July 11, 2018
8 / 10
Investigation: Carlos Balsalobre-Ferna´ndez, Blanca Romero-Moraleda, Rocı´o Cupeiro, Ana
Bele´n Peinado, Javier Butragueño, Pedro J. Benito.
Methodology: Carlos Balsalobre-Ferna´ndez, Rocı´o Cupeiro, Ana Bele´n Peinado, Javier
Butragueño.
Supervision: Carlos Balsalobre-Ferna´ndez, Blanca Romero-Moraleda, Rocı´o Cupeiro, Ana
Bele´n Peinado, Javier Butragueño, Pedro J. Benito.
Writing – original draft: Carlos Balsalobre-Ferna´ndez, Blanca Romero-Moraleda.
Writing – review & editing: Carlos Balsalobre-Ferna´ndez, Blanca Romero-Moraleda, Rocı´o
Cupeiro, Ana Bele´n Peinado, Javier Butragueño, Pedro J. Benito.
References
1.
Alway SE, McCrory JL, Kearcher K, Vickers A, Frear B, Gilleland DL, et al. Resveratrol Enhances Exer-
cise-Induced Cellular and Functional Adaptations of Skeletal Muscle in Older Men and Women. Jour-
nals Gerontol Ser A. 2017; 72: 1595–1606. https://doi.org/10.1093/gerona/glx089 PMID: 28505227
2.
Rabada´n M, Dı´az V, Caldero´n FJ, Benito PJ, Peinado AB, Maffulli N. Physiological determinants of spe-
ciality of elite middle- and long-distance runners. J Sports Sci. 2011; 29: 975–982. https://doi.org/10.
1080/02640414.2011.571271 PMID: 21604227
3.
McCormick A, Meijen C, Marcora S. Psychological Determinants of Whole-Body Endurance Perfor-
mance. Sport Med. 2015; https://doi.org/10.1007/s40279-015-0319-6 PMID: 25771784
4.
Santos-Concejero J, Billaut F, Grobler L, Oliva´n J, Noakes TD, Tucker R. Maintained cerebral oxygen-
ation during maximal self-paced exercise in elite Kenyan runners. J Appl Physiol. 2015; 118: 156–162.
https://doi.org/10.1152/japplphysiol.00909.2014 PMID: 25414248
5.
Coleman DR, Cannavan D, Horne S, Blazevich AJ. Leg stiffness in human running: Comparison of esti-
mates derived from previously published models to direct kinematic-kinetic measures. J Biomech.
2012; 45: 1987–91. https://doi.org/10.1016/j.jbiomech.2012.05.010 PMID: 22682258
6.
Hoogkamer W, Kipp S, Frank JH, Farina EM, Luo G, Kram R. A Comparison of the Energetic Cost of
Running in Marathon Racing Shoes. Sport Med. 2017; https://doi.org/10.1007/s40279-017-0811-2
PMID: 29143929
7.
Bacon AP, Carter RE, Ogle EA, Joyner MJ. VO2max trainability and high intensity interval training in
humans: a meta-analysis. PLoS One. 2013/09/26. 2013; 8: e73182. https://doi.org/10.1371/journal.
pone.0073182 PMID: 24066036
8.
Vuorimaa T, Vasankari T, Rusko H. Comparison of physiological strain and muscular performance of
athletes during two intermittent running exercises at the velocity associated with VO2max. / Etude com-
parative de la contrainte physiologique et de la performance musculaire chez des athletes. Int J Sports
Med. Germany, Federal Republic of; 2000; 21: 96–101. Available: http://articles.sirc.ca/search.cfm?id=
S-165264 PMID: 10727068
9.
Moore IS. Is There an Economical Running Technique? A Review of Modifiable Biomechanical Factors
Affecting Running Economy. Sport Med. 2016; https://doi.org/10.1007/s40279-016-0474-4 PMID:
26816209
10.
Barnes KR, Kilding AE. Running economy: measurement, norms, and determining factors. Sport Med
—Open. 2015; 1: 8. https://doi.org/10.1186/s40798-015-0007-y PMID: 27747844
11.
Balsalobre-Ferna´ndez C, Santos-Concejero J, Grivas G V. Effects of Strength Training on Running
Economy in Highly Trained Runners. J Strength Cond Res. 2016; 30: 2361–2368. https://doi.org/10.
1519/JSC.0000000000001316 PMID: 26694507
12.
Beattie K, Carson BP, Lyons M, Rossiter A, Kenny IC. The Effect of Strength Training on Performance
Indicators in Distance Runners. J Strength Cond Res. 2016; https://doi.org/10.1519/JSC.
0000000000001464 PMID: 27135468
13.
Volek JS, Noakes T, Phinney SD. Rethinking fat as a fuel for endurance exercise. Eur J Sport Sci.
2015; 15: 13–20. https://doi.org/10.1080/17461391.2014.959564 PMID: 25275931
14.
Gastin PB. Energy system interaction and relative contribution during maximal exercise. Sports Med.
2001/09/08. 2001; 31: 725–741. PMID: 11547894
15.
Leckey JJ, Ross ML, Quod M, Hawley JA, Burke LM. Ketone Diester Ingestion Impairs Time-Trial Per-
formance in Professional Cyclists. Front Physiol. 2017; 8: 806. https://doi.org/10.3389/fphys.2017.
00806 PMID: 29109686
Beetroot juice supplementation and exercise economy, rating of perceived exertion and running mechanics
PLOS ONE | https://doi.org/10.1371/journal.pone.0200517
July 11, 2018
9 / 10
16.
Mcmahon NF. The Effect of Dietary Nitrate Supplementation on Endurance Exercise Performance in
Healthy Adults: A Systematic Review and Meta-Analysis. Sport Med. Springer International Publishing;
2016; https://doi.org/10.1007/s40279-016-0617-7 PMID: 27600147
17.
Jones AM. Dietary nitrate supplementation and exercise performance. Sport Med. 2014; https://doi.org/
10.1007/s40279-014-0149-y PMID: 24791915
18.
Lansley KE, Winyard PG, Fulford J, Vanhatalo A, Bailey SJ, Blackwell JR, et al. Dietary nitrate supple-
mentation reduces the O2 cost of walking and running: a placebo-controlled study. J Appl Physiol.
2011; 110: 591–600. https://doi.org/10.1152/japplphysiol.01070.2010 PMID: 21071588
19.
Maughan RJ, Greenhaff PL, Hespel P. Dietary supplements for athletes: Emerging trends and recurring
themes. J Sports Sci. 2011; 29: S57–S66. Available: http://search.ebscohost.com/login.aspx?direct=
true&AuthType=ip,cookie,url,uid&db=sph&AN=69537745&lang=es&site=ehost-live https://doi.org/10.
1080/02640414.2011.587446 PMID: 22150428
20.
Boorsma RK, Whitfield J, Spriet LL. Beetroot Juice Supplementation Does Not Improve Performance of
Elite 1500-m Runners. Med Sci Sport Exerc. 2014; 46: 2326–2334. https://doi.org/10.1249/MSS.
0000000000000364 PMID: 24781895
21.
Santos-Concejero J, Olivan J, Mate-Munoz JL, Muniesa C, Montil M, Tucker R, et al. Gait Cycle Char-
acteristics and Running Economy in Elite Eritrean and European Runners. Int J Sports Physiol Perform.
2014/10/14. 2014; https://doi.org/10.1123/ijspp.2014-0179 PMID: 25310728
22.
Saunders PU, Pyne DB, Telford RD, Hawley JA. Factors affecting running economy in trained distance
runners. Sport Med. New Zealand; 2004; 34: 465–485. Available: http://articles.sirc.ca/search.cfm?id=
S-969152
23.
Rogers SA, Whatman CS, Pearson SN, Kilding AE. Assessments of Mechanical Stiffness and Relation-
ships to Performance Determinants in Middle-Distance Runners. Int J Sports Physiol Perform. 2017; 1–
23. https://doi.org/10.1123/ijspp.2016-0594 PMID: 28290718
24.
Jones AM, Doust JH. A 1% treadmill grade most accurately reflects the energetic cost of outdoor run-
ning. J Sports Sci. 1996; 14: 321–327. https://doi.org/10.1080/02640419608727717 PMID: 8887211
25.
Wang B, Xu G, Tian Q, Sun J, Sun B, Zhang L, et al. Differences between the Vastus Lateralis and Gas-
trocnemius Lateralis in the Assessment Ability of Breakpoints of Muscle Oxygenation for Aerobic
Capacity Indices During an Incremental Cycling Exercise. J Sports Sci Med. 2012; 11: 606–13. PMID:
24150069
26.
Ferrari M, Mottola L, Quaresima V. Principles, techniques, and limitations of near infrared spectroscopy.
Can J Appl Physiol. 2004; 29: 463–87. Available: http://www.ncbi.nlm.nih.gov/pubmed/15328595
PMID: 15328595
27.
Balsalobre-Ferna´ndez C, Agopyan H, Morin J-B. The Validity and Reliability of an iPhone App for Mea-
suring Running Mechanics. J Appl Biomech. 2017; 33: 222–226. https://doi.org/10.1123/jab.2016-0104
PMID: 27918692
28.
Hopkins WG, Marshall SW, Batterham AM, Hanin J. Progressive Statistics for Studies in Sports Medi-
cine and Exercise Science. Med Sci Sport Exerc. 2009; 41: 3–12.
29.
McQuillan JA, Dulson DK, Laursen PB, Kilding AE. Dietary Nitrate Fails to Improve 1 and 4km Cycling
Performance in Highly-Trained Cyclists. Int J Sport Nutr Exerc Metab. 2016; 1–26.
30.
Sano K, Ishikawa M, Nobue A, Danno Y, Akiyama M, Oda T, et al. Muscle-tendon interaction and EMG
profiles of world class endurance runners during hopping. Eur J Appl Physiol. 2012/12/12. 2013; 113:
1395–1403. https://doi.org/10.1007/s00421-012-2559-6 PMID: 23229882
31.
Brizendine JT, Ryan TE, Larson RD, McCully KK. Skeletal Muscle Metabolism in Endurance Athletes
with Near-Infrared Spectroscopy. Med Sci Sport Exerc. 2013; 45: 869–875. https://doi.org/10.1249/
MSS.0b013e31827e0eb6 PMID: 23247709
32.
Mallette MM, Stewart DG, Cheung SS. The Effects of Hyperoxia on Sea-Level Exercise Performance,
Training, and Recovery: A Meta-Analysis. Sport Med. 2018; 48: 153–175. https://doi.org/10.1007/
s40279-017-0791-2 PMID: 28975517
Beetroot juice supplementation and exercise economy, rating of perceived exertion and running mechanics
PLOS ONE | https://doi.org/10.1371/journal.pone.0200517
July 11, 2018
10 / 10
| The effects of beetroot juice supplementation on exercise economy, rating of perceived exertion and running mechanics in elite distance runners: A double-blinded, randomized study. | 07-11-2018 | Balsalobre-Fernández, Carlos,Romero-Moraleda, Blanca,Cupeiro, Rocío,Peinado, Ana Belén,Butragueño, Javier,Benito, Pedro J | eng |
PMC10030110 | Dhawale and Venkadesan. eLife 2023;12:e67177. DOI: https:// doi. org/ 10. 7554/ eLife. 67177
1 of 20
How human runners regulate footsteps
on uneven terrain
Nihav Dhawale1,2, Madhusudhan Venkadesan1*
1Department of Mechanical Engineering and Materials Science, Yale University, New
Haven, United States; 2National Centre for Biological Sciences, Tata Institute of
Fundamental Research, Mumbai, India
Abstract Running stably on uneven natural terrain takes skillful control and was critical for
human evolution. Even as runners circumnavigate hazardous obstacles such as steep drops, they
must contend with uneven ground that is gentler but still destabilizing. We do not know how foot-
steps are guided based on the uneven topography of the ground and how those choices influence
stability. Therefore, we studied human runners on trail- like undulating uneven terrain and measured
their energetics, kinematics, ground forces, and stepping patterns. We find that runners do not
selectively step on more level ground areas. Instead, the body’s mechanical response, mediated by
the control of leg compliance, helps maintain stability without requiring precise regulation of foot-
steps. Furthermore, their overall kinematics and energy consumption on uneven terrain showed little
change from flat ground. These findings may explain how runners remain stable on natural terrain
while devoting attention to tasks besides guiding footsteps.
Editor's evaluation
This paper presents fundamental evidence for the control mechanisms used by running humans to
maintain stability while running on naturalistically uneven terrain. The authors use a creative and
compelling combination of experiments and modeling to analyze running on terrain with mildly
stochastic undulating roughness, a condition that resembles natural terrain conditions, such as trail
running. The findings suggest that humans use open- loop, intrinsically stable strategies to run on
this terrain, and not visually guided foot placement, making an important contribution to under-
standing the context- dependent role of vision in human locomotion.
Introduction
Running on natural terrain is an evolutionarily important human ability (Carrier et al., 1984; Bramble
and Lieberman, 2004), which requires the skillful negotiation of uneven ground (Lee and Lishman,
1977; Warren et al., 1986). Part of the challenge is planning a path in real- time that navigates around
obstacles or sudden steep drops. Even after finding a path around such hazards, the ground would
be uneven. Planning the stepping pattern using detailed information of every bump and dip of the
ground is typically infeasible on natural trails because the ground is often covered by foliage or grass.
But the seemingly slight unevenness, albeit gentler than large obstacles or drops, could have signif-
icant consequences to stability. Mathematical modeling predicts that even slightly uneven ground,
with peak- to- valley height variations less than the dorso- plantar foot height, could be severely desta-
bilizing unless some form of mitigation strategy is employed to deal with them (Dhawale et al., 2019).
In this paper, we investigate how human runners deal with these types of undulating uneven ground.
Studies on human walking find that footsteps are visually guided to plan a path through complex,
uneven terrain (Matthis et al., 2018; Thomas et al., 2020; Bonnen et al., 2021). Although there are
RESEARCH ARTICLE
*For correspondence:
[email protected]
Competing interest: The authors
declare that no competing
interests exist.
Funding: See page 17
Received: 04 February 2021
Preprinted: 22 February 2021
Accepted: 21 February 2023
Published: 22 February 2023
Reviewing Editor: Monica A
Daley, University of California,
Irvine, United States
Copyright Dhawale and
Venkadesan. This article is
distributed under the terms
of the Creative Commons
Attribution License, which
permits unrestricted use and
redistribution provided that the
original author and source are
credited.
Research article
Physics of Living Systems | Neuroscience
Dhawale and Venkadesan. eLife 2023;12:e67177. DOI: https:// doi. org/ 10. 7554/ eLife. 67177
2 of 20
no similar studies of running on naturalistic uneven terrain, we may expect that vision’s role is multi-
fold. For example, in the evolutionary context of persistence hunting (Carrier et al., 1984; Bramble
and Lieberman, 2004), vision is needed to track footprints and continuously survey the landscape
for prey in addition to dealing with the terrain’s unevenness. The potentially competing demands on
visual attention—for stability versus other functional goals—is probably more exacting in running than
in walking because of the greater speeds involved and the shorter time available to sense and act.
Additional important factors to consider on uneven terrain include dynamic stability (Holmes et al.,
2006; Dhawale et al., 2019; Daley and Biewener, 2006; Voloshina and Ferris, 2015), leg safety
(Birn- Jeffery et al., 2014), peak force mitigation (Blum et al., 2014), and anticipatory leg adjustments
(Birn- Jeffery and Daley, 2012; Müller et al., 2015). However, we presently lack studies of human
runners on naturalistic uneven terrain to investigate the role of vision- guided footstep regulation and
the subtle regulation of body mechanics for maintaining stability, which motivates the overground
running experiments presented in this paper.
In addition to vision, the body’s mechanical responses aid stability and are neurally modulated
through muscle contractions. These mechanical properties have been studied theoretically, and
experimental data have been interpreted, through the lens of models that approximate the runner as
a point- like mass on a massless leg, commonly referred to as the spring- legged inverted pendulum
(SLIP) model (Seyfarth et al., 2002; Daley et al., 2006; Geyer et al., 2006; Birn- Jeffery et al., 2014;
Müller et al., 2016; Seethapathi and Srinivasan, 2019). SLIP models have hypothesized multiple
stabilization strategies for terrain with random height variations, several of which have found experi-
mental support: higher leg retraction rates (Karssen et al., 2015), wider lateral foot placement (Volos-
hina and Ferris, 2015; Mahaki et al., 2019), and the possible use of vision to guide foot placement
(Birn- Jeffery and Daley, 2012). But SLIP models do not help understand the effect of slope variations
because the ground force is constrained to always point to the center of mass irrespective of whether
the foot contacts the ground on a level or sloping region. That is a consequence of the zero moment
of inertia about the center of mass for SLIP models. Analyses of models with non- zero moment of
inertia show that both height and slope variations are detrimental to stability, with slope being more
destabilizing (Dhawale et al., 2019), reminiscent of common experience among runners.
Understanding why slope variations degrade stability could generate hypotheses and testable predic-
tions for how human runners deal with stability on naturalistic uneven terrain. The mathematical analyses
of Dhawale et al., 2019, find that random variations in slope lead to step- to- step fluctuations in the
fore- aft ground impulse. For steady forward running, the net forward impulse should be zero for every
step. But small step- to- step random variation of the fore- aft ground impulse leads to a gradual accumu-
lation of sagittal plane angular momentum, which ultimately destabilizes the runner. However, the rate
at which the destabilizing angular momentum builds up depends on where on the terrain the foot lands
and how the body responds to landing on the ground, thus suggesting two mitigating strategies. One
strategy is to minimize the fore- aft impulse that is experienced at touch down, which has the effect of
significantly slowing down the fluctuation- induced build- up of destabilizing angular momentum. This can
be achieved by reducing the forward speed of the foot at touchdown via leg retraction and by reducing
limb compliance so that the momentum of the rest of the body contributes lesser to the fore- aft impulse.
Another strategy is to try and land primarily on local maxima or other flat regions of the terrain so that
the destabilizing influence of random slope variations is reduced. The experimental assessment of these
two strategies is the topic of this paper.
Most past experimental studies of uneven terrain running have used step- like blocks to show how
humans and animals deal with height variations on the ground (Daley et al., 2006; Müller et al., 2015).
Later work modified the terrain design to use blocks that were narrow enough so that the foot had to
span more than one fore- aft block, leading the foot to be randomly tilted during foot flat (Voloshina
and Ferris, 2015). Specifically, the blocks were of three different heights (labeled A, B, and C), which
leads to six possible height difference pairings (AB, BA, AC, CA, BC, CB). In natural terrain, the variation
in slope is continuously graded, which would allow for more variation in the foot flat angle. Moreover,
as hypothesized by theoretical analysis (Dhawale et al., 2019), it is not only the foot angle that affects
whole body dynamics, but the force direction from the ground also matters. In this regard, the natural
terrain may differ from the block design, particularly during initial contact and push- off when only a small
region of the foot makes contact with the ground. During that time, the block design would not influence
the ground forces like the sloped ground of natural undulating terrain would. Moreover, complex terrain
Research article
Physics of Living Systems | Neuroscience
Dhawale and Venkadesan. eLife 2023;12:e67177. DOI: https:// doi. org/ 10. 7554/ eLife. 67177
3 of 20
types may be required to capture the range of strategies used to run on naturalistic uneven terrain. This is
suggested by studies that examine walking on a variety of outdoor terrain and show that stride variability
and energetics significantly depend on terrain complexity (Kowalsky et al., 2021). Undulating uneven
terrain have been studied in the context of walking (Kent et al., 2019; Kowalsky et al., 2021), but not
running. So there is a need for experiments to study running on undulating terrain with continuously
varying slopes to expand the current understanding of how uneven terrain affects stability. In this paper
we experimentally assess foot placement patterns, fore- aft ground impulses, stepping kinematics, and
metabolic power consumption on undulating uneven terrain whose unevenness is akin to running trails
(Figure 1).
Methods
Protocol and experimental measurements
We conducted overground running experiments with nine subjects (eight men, one woman; age
23–45 years, body mass 66.1 ± 8.5 kg, leg length 0.89 ± 0.04 m, reported as mean ± SD). All subjects
were able- bodied, ran approximately 30 km per week, and had run at least one half- marathon or
marathon within the previous year. Experiments were conducted at the National Centre for Biolog-
ical Sciences, Bangalore, India, with informed consent from the volunteers, and IRB approval
(TFR:NCB:15_IBSC/2012).
Subjects ran back- and- forth on three 24 m long and 0.6 m wide tracks (Figure 2a). In addition to
a flat track, we used two custom- made uneven tracks, uneven I and uneven II, which had increasing
unevenness. Uneven I and uneven II had peak- to- valley height differences (amplitude) of 18 ± 6 and
28 ± 11 mm (mean ± SD), respectively, and peak- to- peak horizontal separation (wavelength) of 102 ± 45
and 108 ± 52 mm, respectively (Figure 2b, c and d). We recorded kinematics using an 8- camera
motion capture system (Vicon Inc., Oxford, UK) at 300 frames per second and measured the ground
reaction forces at 600 Hz using two force plates (AMTI Inc., model BP600900) embedded beneath
the center of the track. The cameras recorded an approximately 10 m long segment of the center of
the track. Breath- by- breath respirometry was also recorded by a mobile gas analyzer (Oxycon Mobile,
CareFusion Inc.).
A single trial consisted of a 3 min period of standing when the resting metabolic rate was recorded
followed by subjects running back- and- forth on the track for at least 8 min and up to 10 min, dictated
by VO2 reading equilibration time and the subject’s ability to maintain speed over the course of the
Figure 1. Uneven terrain experiments. (a) We conducted human- subject experiments on flat and uneven terrain
while recording biomechanical and metabolic data. The reflective markers and the outline of the force plate are
digitally exaggerated for clarity. (b) Footsteps were recorded to determine whether terrain geometry influences
stepping location, illustrated here by a mean- subtracted contour plot of terrain height for an approximately 6 foot
segment of uneven II overlaid with footsteps (location of the heel marker). Blue and red circles represent opposite
directions of travel and transparency level differentiates trials.
The online version of this article includes the following source data for figure 1:
Source data 1. Dimensional mass and leg length of every subject.
Research article
Physics of Living Systems | Neuroscience
Dhawale and Venkadesan. eLife 2023;12:e67177. DOI: https:// doi. org/ 10. 7554/ eLife. 67177
4 of 20
trial. Each subject ran on all three terrains, with the order randomized. We controlled the running
speed using a moving light array in 24 m long LED strips laid on either side of the track (Figure 2a).
Subjects were instructed to stay within the bounds of a 3 m illuminated segment of the LED strip
that traveled at 3 m/s. This speed was chosen as it was comfortable for all subjects and lies within
the endurance running speed range for humans (Bramble and Lieberman, 2004). Running speed
fluctuated within a trial, however mean speed as well as speed variability were consistent across
terrain types (see Results for details). Subjects were provided with standardized, commercially avail-
able running shoes.
Uneven terrain
Terrain unevenness was heuristically specified so that peak- to- valley height variations were approx-
imately equal to the height of the malleolus while standing barefoot on level ground, and peak- to-
peak horizontal distances were similar to foot length (Figure 2b). Large terrain height variations may
elicit obstacle avoidance strategies, which is not the subject of this paper, and peak- to- peak hori-
zontal separation longer than the step length may make the slope variation too gentle. Conversely,
small height variations that are similar to the heel pad thickness, and peak- to- peak horizontal sepa-
ration that is smaller than the foot length, will likely be smoothed out by foot and sole compliance
(Venkadesan et al., 2017).
The uneven terrains were constructed by Mars Adventures Inc. (Bangalore, India) by laying fiber
glass over heuristically created contours. Epoxy was used to harden the fiber glass sheets into a stiff
shell which was coated with a slurry of sand and epoxy to create a surface that texturally resembles
weathered rock. The width at the ends of the uneven track were broadened to approximately 1 m to
allow for runners to change direction while remaining on the terrain. The terrain was then digitized
using a dense arrangement of reflective markers that were recorded by the motion capture system.
Kinematics
Foot kinematics were recorded using fiducial markers that were fixed to the shoes over the calcaneus,
second distal metatarsal head, and below the lateral malleolus. Markers were attached to the hip, over
Figure 2. Details of the experiment design. (a) Schematic of the running track, camera placement, force plate
positions and the LED strip with a 3 m illuminated section. (b) The terrain was designed so that the range of its
height distribution h was comparable to ankle height hpp ∼ hf and peak- to- peak distances λ (along the length of
the track) were comparable to foot length λ ∼ lf . (c) Histograms of the mean subtracted heights h of the uneven
terrain. (d) Histograms of the peak- to- peak separation λ of the uneven terrain.
Research article
Physics of Living Systems | Neuroscience
Dhawale and Venkadesan. eLife 2023;12:e67177. DOI: https:// doi. org/ 10. 7554/ eLife. 67177
5 of 20
the left and right lateral superior iliac spine, and the left and right posterior superior iliac spine. The
mean position of the hip markers was used to estimate the center of mass location.
Stance is defined as when the heel marker’s forward velocity was minimized and its height was
within 15 mm of the marker’s height during standing. The threshold of 15 mm was chosen to account
for terrain height variations so that stance may be detected even when the heel lands on a local peak
of the uneven terrain.
The center of mass forward speed v = dstep/tstep is found from the distance dstep covered by the
center of mass in the time duration tstep between consecutive touchdown events. Leg angle at touch-
down is defined as the angle between the vertical and the line formed by joining the heel marker
to the center of mass. Virtual leg length at touchdown is defined as the distance between the heel
marker and the center of mass. Foot length lf is defined as the average distance in the horizontal plane
between the toe and heel marker, across all subjects. The center of mass trajectory during stance was
fitted with a regression line in the horizontal plane. The step width is twice the distance of nearest
approach of the stance foot from the regression line. This definition allows for the runner’s center of
mass trajectory to deviate while preserving a definition of step width that is consistent with those
previously used (Donelan et al., 2001; Arellano and Kram, 2011). We estimated meander, i.e., the
deviation of the center of mass from a straight trajectory, using (d − d0)/d0 , where d is the distance
covered by the center of mass in the horizontal plane during a single run across the length of the track
and d0 is the length of the straight- line fit to the center of mass trajectory. Foot velocity or center of
mass velocity at landing were calculated by fitting a cubic polynomial to the heel marker trajectory
or center of mass trajectory, respectively, in a 100 ms window before touchdown, and calculating the
time derivative of the fitted polynomial at the moment just prior to touchdown. Leg retraction rate ω
is determined using ω = vf/||l|| , where vf is the component of the foot’s relative velocity with respect
to the center of mass that is perpendicular to the virtual leg vector l (vector joining heel to center of
mass).
Step width, step length, and virtual leg length at touchdown are normalized by the subject’s leg
length, defined as the distance between the greater trochanter and lateral malleolus.
To correct for slight angular misalignments between the motion capture reference frame and the
long axis of the running track, we align the average CoM trajectory over the entire track length to be
parallel to the y- axis of the motion capture reference frame. This correction reflects the experimental
observation that the subjects run along the center of the track.
Kinetics
Force plate data were low- pass filtered using an eighth order, zero- phase, Butterworth filter with a
cut- off frequency of 270 Hz. Touchdown on the force plates was defined by a threshold for the vertical
force of four standard deviations above the mean unloaded baseline reading.
The forward collision impulse, defined as the maximal decelerating fore- aft impulse J∗y , was found
by integrating the fore- aft component Fy of the ground reaction force during the deceleration phase
as
J∗
y = max
t
t
ˆ
0
Fy(τ) dτ
.
(1)
We normalized J∗y by the aerial phase forward momentum mvy , where vy is the forward speed of the
center of mass during the aerial phase.
Energetics
Net metabolic rate is defined as the resting metabolic power consumption subtracted from the power
consumption during running and normalized by the runner’s mass. Metabolic power consumption is
determined using measurements of the rate of O2 consumption and CO2 consumption using formulae
from Brockway, 1987. For running, this is calculated after discarding the first 3 min of the run to elim-
inate the effect of transients. The resting metabolic power consumption is calculated after discarding
the first minute of the standing period of the trial. Data from each trial were visually inspected to
ensure that the rates of O2 consumption and CO2 production had reached a steady state, seen as a
plateau in the data trace.
Research article
Physics of Living Systems | Neuroscience
Dhawale and Venkadesan. eLife 2023;12:e67177. DOI: https:// doi. org/ 10. 7554/ eLife. 67177
6 of 20
Shuttle running
Of the total track length of 24 m, a 1.2 m turnaround segment was designed at each end to facilitate
the subjects to reverse their running direction without stepping off the track. These end segments
were 1 m wide, which was broader than the rest of the track that was only 0.6 m wide. The runners
would reach the end of the track and turn around promptly. Guiding light bars that controlled the
running speed would be half ‘absorbed’ into the end before reversing direction, which allowed for
sufficient time for the subjects to turn around while still maintaining the same average speed. The
subjects were given, and took, around 0.5 s to turn around. The subjects ran at a steady speed within
the capture volume that covers the middle 10 m of the track (see Results for details). The cameras
could not capture the ends of the track but the experimenters observed that the subjects stayed
within the moving light bar through the 21.6 m long straight portion of the track. The experimental
protocol used in this study was tuned through pilot trials involving the authors of this manuscript and
two initial subjects. The data from these pilot trial subjects are not part of the reported results in this
manuscript.
Foot stepping analysis
Directed foot placement scheme
The runners’ foot landing locations were compared to a Markov chain Monte Carlo (MCMC) model
which finds stepping locations with the lowest terrain unevenness subject to constraints of matching
experimentally measured stepping kinematics. All participants were heel- strike runners on all terrain
types, as judged from the double peak in the vertical ground reaction force profile. Therefore, the
stepping model sampled the terrain in rear- foot sized patches, which we define to be 95 mm × 95
mm (dimensions are chosen to be half the size of the foot length, 190 mm). The interquartile range of
heights ( hIQR ) in each patch was used as a measure of its unevenness.
Starting from an initial position (xi, yi) , the model takes the next step to (xi+1, yi+1) in the following
stages: open- loop stage, minimization stage, and a noise process given by,
open- loop stage:
ˆxi+1 = xi + (−1)isw, ˆyi+1 = yi + (−1)jsl.
(2)
Minimization stage:
(x′
i+1, y′
i+1) = arg min(x,y) t(x, y),
x ∈ [ˆxi+1 − σsw,ˆxi+1 + σsw],
y ∈ [ˆyi+1 − σsl,ˆyi+1 + σsl].
(3)
Noise process:
xi+1 = x′
i+1 + ηx, yi+1 = y′
i+1 + ηy,
where ηx ∼ vM(1, 0, σsw), ηy ∼ vM(1, 0, σsl).
(4)
In the open- loop stage, the model takes a step forward and sideways dictated by the experimentally
measured mean step length sl and mean step width sw, respectively. The exponent j is either 0 or
1 and keeps track of the direction of travel. The function t(x, y) evaluates the interquartile range of
heights of a rear- foot sized patch centered around position (x, y) . In the minimization step, the model
conducts a bounded search about (ˆxi+1,ˆyi+1) for the location that minimizes t(x, y) . The search region is
defined by the standard deviations of the measured step width σsw and step length σsl . To perform the
minimization, a moving rear- foot sized window with step- sizes of σsw/10 along the width of the track
and σsl/10 along its length are used to evaluate t(x, y) at various candidate stepping locations within
the search region. The step- sizes for translating the moving window were chosen because they were
much smaller than typical terrain features and thus the landing location with the lowest unevenness
(x′
i+1, y′
i+1) was determined by the terrain properties, not model parameters. To simulate sensorimotor
noise, the location of this minimum (x′
i+1, y′
i+1) is perturbed by random variables ηx, ηy . The random
variables are drawn from von Mises distributions with κ = 1 , centered about zero, and scaled so that
the base of support for the distributions are σsw and σsl , respectively.
Research article
Physics of Living Systems | Neuroscience
Dhawale and Venkadesan. eLife 2023;12:e67177. DOI: https:// doi. org/ 10. 7554/ eLife. 67177
7 of 20
At the ends of the track, the x position of the runner is reset so that the runner is at the center of
the track, and the direction of travel is reversed ( j value is toggled). We simulate for 100,000 steps
to ensure that reported terrain statistics at footstep locations as well as step length and step width
converge, i.e., errors between simulations in these parameters are less than 1% of their mean value.
Quantifying foot placement patterns
We used a second analysis of footstep patterns that correlated the foot landing probability with
terrain unevenness. To perform this analysis, we define a foot placement index to estimate the prob-
ability that the runner’s foot lands within a foot- sized patch of the track. To calculate this index, we
first divide the terrain into a grid of 0.5 foot lengths × 1.0 foot lengths cells, with the longer side of
the cell along the length of the track (Figure 3a). We count the number of footsteps fi,j in each cell
ci,j , where i indexes the position of the cell along the length of the track and j indexes the position
of cell transverse to the track. The point of landing is determined by the location of the heel marker.
Even if the fore- foot crosses over the adjacent cell boundary, the location of the heel marker uniquely
specifies the landing cell identity. We also define step length- sized neighborhoods that contain cell ci,j
which are one step- length long and as wide as the track. Each such neighborhood has a cumulative
footstep count Si that depends on the longitudinal location i of the cell. The average across all such
step length- sized neighborhoods that contain cell ci,j is S . This average S is used to normalize each fi,j
to yield the foot placement index pi,j according to,
pi,j = fi,j
S .
(5)
The index pi,j measures the fraction of times a foot lands in cell ci,j compared to all other cells that are
within a step length distance of it (Figure 3b). If runners were perfectly periodic with no variation in
footstep location from one run over the terrain to the next, pi,j = 1 for cells on which subjects stepped
and pi,j = 0 otherwise. If, however, stepping location was the result of a uniform random process, pi,j
would be a constant for every cell of the terrain and equal to the reciprocal of the number of cells in
a step- length sized box. Heat maps of the foot placement index pi,j are shown in Figure 3—figure
supplement 1. We report the total number of footsteps recorded for each trial in Figure 3—source
data 1.
To probe foot placement strategies we determine whether the foot placement index pi,j correlates
with the median height or the interquartile range of heights within the cell ci,j . Positive correlation with
the median height would indicate stepping on local maxima that are flatter than the surrounding, and
Figure 3. Foot placement analysis. (a) Red circles denote footstep locations (392 footsteps) in the ‘ x − y ’ plane for
a representative trial on uneven II. The grid spacing is 190 mm along the length of the track and 95 mm along its
width. Step length s0 is shown for reference. T is the length of the capture volume and W is the width of the track.
Note that the x and y axes of this figure are not to the same scale. (b) The probability of landing on a foot- sized
region of the track is quantified by the foot placement index Equation 5 shown as a heatmap with the color bar at
the top left.
The online version of this article includes the following source data and figure supplement(s) for figure 3:
Source data 1. Footstep counts for each subject on all terrain.
Figure supplement 1. Subject- wise foot placement patterns.
Research article
Physics of Living Systems | Neuroscience
Dhawale and Venkadesan. eLife 2023;12:e67177. DOI: https:// doi. org/ 10. 7554/ eLife. 67177
8 of 20
negative correlation with the interquartile range would indicate stepping on flatter regions with more
uniform height. We test this hypothesis through the use of a statistical model described in ‘Statistical
analysis and reporting’.
Collision model
To delineate the relative contributions of joint stiffness and forward foot speed to the fore- aft impulse,
we model the impulse due to the foot- ground interaction. In the model, a planar three- link chain
represents the foot, shank, and thigh, and a fourth link represents the torso. Following Dempster,
1955, all masses and lengths are expressed as fractions of the body mass and leg length, respec-
tively. This model builds upon the leg collision model of Lieberman et al., 2010, by including addi-
tional segments representing the thigh and torso and calculating the fore- aft collisional impulse. The
collision is assumed to be instantaneous and inelastic, with a point- contact between the leg and the
ground. Such collision models are widely used to capture the stance impulse due to ground forces in
walking (Donelan et al., 2002; Ruina et al., 2005) and running (Srinivasan and Ruina, 2006; Dhawale
et al., 2019). Because the collision is assumed to be instantaneous, only infinite forces contribute to
the impulse (Chatterjee and Ruina, 1998; Lieberman et al., 2010). Therefore, to investigate the
effect of joint compliance, we model the hinge joints connecting the links as either infinitely compliant
or perfectly rigid. The advantage of these contact models is their ability to accurately capture the
impulse without the numerous additional parameters needed to represent the complete force- time
history when contact occurs between two bodies (Chatterjee and Ruina, 1998).
We use experimental data on center of mass velocity and leg retraction rate just prior to landing,
along with the leg angle at touchdown, to compute a predicted collisional impulse. Because all our
runner’s were heel- strikers, we use foot- strike index s = 0.15 for the collision calculations (Lieberman
et al., 2010). The foot- strike index ranges from 0 for heel strikes to 1 for forefoot strikes and encodes
the runner’s foot strike pattern. The ratio of the collisional impulse to the measured whole body
momentum just prior to landing is calculated for the model at the two joint stiffness extremes and
compared with experimental measurements of the normalized fore- aft impulse. By analyzing the colli-
sional impulse for these two extremes of joint stiffness, we isolate the contributions to the fore- aft
impulse arising from varying the joint stiffness versus varying the forward foot speed at landing.
Notation
Notation used in this section is as follows. Scalars are denoted by italic symbols (e.g. I for the moment
of inertia), vectors by bold, italic symbols ( v for velocity), and points or landmarks in capitalized non-
italic symbols (such as center of mass G in Figure 4a). Vectors associated with a point, such as the
velocity of center of mass G are written as vG , with the upper- case alphabet in the subscript specifying
O
E
B
F
C
K
D
mf
Ms
Mt
lf
slf
Ls
Lt
O
B
R1
R2
C
a
b
c
d
G
y
z
+
A
θ
M
H
e
N
R3
Lto
D
J
Figure 4. Model for estimating fore- aft collision impulses from kinematic data. (a) A four- link model of the foot
(A–B), shank (B–C), thigh (C–D), and torso (D–N) moving with center of mass velocity v−
G and angular velocity Ω−
collides with the ground at angle θ . (G) represents the center of mass. Leg length and body mass are obtained
from data and scaled according to Dempster, 1955, to obtain segment lengths and masses. Free- body diagrams
show all non- zero external impulses: (b) collisional impulse J acting at O, and panels (c, d, e) show reaction
impulses R1 , R2 , and R3 acting at B, C, and D, respectively.
Research article
Physics of Living Systems | Neuroscience
Dhawale and Venkadesan. eLife 2023;12:e67177. DOI: https:// doi. org/ 10. 7554/ eLife. 67177
9 of 20
the point in the plane. Moment of inertia variables are subscripted with ‘/A’ representing the moment
of inertia computed about point A, such as I/G representing the moment of inertia about the center
of mass G. Position vectors are denoted by rA/B which denotes the position of point A with respect
to point B. Variables just before the collision with the terrain are denoted by the superscript ‘–’, and
just after the collision by the superscript ‘+’. Equations with variables that have no superscript apply
throughout stance.
Rigid joints
Consider the L- shaped bar (Figure 4a) falling with velocity v−
G = v−
y ˆȷ + v−
z ˆk and angular velocity
Ω− = ω−ˆı . Upon contact with the ground, the point O on the foot instantly comes to rest and the
center of mass translational and angular velocities change to v+
G = v+yˆȷ + v+z ˆk, Ω+ = ω+ˆı . Due to the
instantaneous collision assumption, finite forces like the gravitational force do not contribute to the
collisional impulse, and the ground reaction force at point O leads to the impulse J (Figure 4b).
Angular momentum balance about the contact point O yields the relationship between pre and post
collision velocities,
MbrG/O × v−
G + I/GΩ− = MbrG/O × v+
G + I/GΩ+,
(6a)
vG = vO + Ω × rG/O,
(6b)
where v+
O = 0.
(6c)
The total mass Mb is the sum of the masses of the torso M , thigh Mt , shank Ms , and foot mf. We solve
for ω+ in Equation 6a and obtain the post- collision center of mass velocity v+
G using Equation 6b.
From this, the collision impulse J and the normalized fore- aft collisional impulse |J∗y |/Jb are calculated
using,
J = Mb(v+
G − v−
G ),
(7a)
J∗y = J .ˆj,
(7b)
and Jb = Mb(v−
G ·ˆȷ).
(7c)
Compliant joints
If the L- bar has compliant joints, then the post- collision velocities for each segment may vary. There-
fore, we write additional angular momentum balance equations for each segment to solve for the
post- collision state. Since the only non- zero external impulse acting on the shank, thigh, and torso
segments is the reaction impulse R1 acting at B (Figure 4c), the only non- zero external impulse on
the thigh and torso portion of the leg is the reaction impulse R2 acting at C (Figure 4d), and the only
non- zero external impulse acting on the torso portion of the leg is the reaction impulse R3 acting at
D (Figure 4e), we write angular momentum balance equations for the entire body and these three
segments as
MbrG/O × v−
G + I/GΩ− = mfrE/O × v+
E + I/EΩ+
E+
MsrF/O × v+
F + I/FΩ+
F +
MtrK/O × v+
K + I/KΩ+
K+
MrH/O × v+
H + I/HΩ+
H,
(8a)
MsrF/B × v−
F + MtrK/B × v−
K +
MrH/B × v−
H + (I/F + I/K + IH)Ω− =
MsrF/B × v+
F + I/FΩ+
F +
MtrK/B × v+
K + I/KΩ+
K+
MrH/B × v+
H + I/HΩ+
H,
(8b)
MtrK/C × v−
K + MrH/C × v−
H +
(I/K + I/H)Ω− =
MtrK/C × v+
K + I/KΩ+
K+
MrH/C × v+
H + I/HΩ+
H,
(8c)
MrH/D × v−
H + I/HΩ− = MrH/D × v+
H + I/HΩ+
H
(8d)
Research article
Physics of Living Systems | Neuroscience
Dhawale and Venkadesan. eLife 2023;12:e67177. DOI: https:// doi. org/ 10. 7554/ eLife. 67177
10 of 20
where I/E, I/F, I/K, I/H are moments of inertia of the foot, shank, thigh, and torso segments, respec-
tively, about their centers. The linear and angular velocities of the foot ( vE, ΩE ), shank ( vF, ΩF ), thigh
( vK, ΩK ), and torso ( vH, ΩH ) are related to the velocity of the contact point O as
vE = vO + ΩE × rE/O,
(9a)
vF = vO + ΩE × rB/O + ΩF × rF/B,
(9b)
vK = vO + ΩE × rB/O + ΩF × rC/B + ΩK × rK/C,
(9c)
vH = vO + ΩE × rB/O + ΩF × rC/B + ΩK × rD/C + ΩH × rH/D,
(9d)
where v−
O = v−
G + Ω− × rO/G,
(9e)
and v+
O = 0.
(9f)
Simultaneously solving Equations 8a, b, c, d–9a, b, c, d, e, f yields the post- collision velocities for
each segment of the L- bar. From these, we calculate the normalized fore- aft collision impulse for the
compliant model using Equation 7a, b, c.
Statistical methods
Sample size
Sample size could refer to the number of subjects or the number of footsteps that were used in the
analyses. The number of subjects recruited was informed by typical participant numbers that were
used in similar past studies (Donelan et al., 2004; Voloshina and Ferris, 2015; Seethapathi and
Srinivasan, 2019). There is an additional consideration for sufficiency of sample numbers for the foot
placement analysis. The steps should densely sample the approximately 10 m long central region of
the track, where the motion capture system was recording from. The 5262 recorded steps (2526 on
uneven I, 2736 on uneven II) are sufficient to densely sample the measurement region assuming a
rear- foot sized patch for each step.
Statistical analysis and reporting
Measures of central tendency (mean or median) and variability (standard deviation or interquartile
range) of the distributions of step width, step length, center of mass speed, forward foot speed at
landing, fore- aft impulse, virtual leg length at touchdown, leg angle at touchdown, net metabolic rate,
and meander are reported for each trial.
We use three different linear mixed models to determine (a) whether gait variables vary with terrain
type, (b) whether leg angle at touchdown and decelerating fore- aft impulses covary with forward foot
speed at touchdown, and (c) whether the foot placement index pi,j (Equation 5) correlates with the
median height or the interquartile range of heights within the terrain region at landing. The statistical
models are run using the lmerTest package in R (Kuznetsova et al., 2017). We use a linear mixed-
model fit by restricted maximum likelihood t- tests with Satterthwaite approximations to degrees of
freedom. An ANOVA on the first model tests for the effect of the terrain factor, an ANCOVA on
the second model tests for the effect of the terrain factor and the covariate forward foot speed,
and an ANCOVA on the third model tests whether the probability of landing on a terrain patch pi,j
significantly covaries with the height or unevenness of that terrain patch. Post- hoc pairwise compar-
isons, where relevant, are performed using the emmeans package in RStudio with p- values adjusted
according to Tukey’s method.
A measure of central tendency or variability within a trial is the dependent variable y for the first
linear mixed model. There are 27 observations for the dependent variable y corresponding to each
trial (nine subjects running on three terrain). Terrain is the fixed factor and subjects are random factors
in the model given by
yij = (β0 + µj) + βiterraini + ϵij,
(10)
where i = 1, 2 and j = 1 . . . 9 . The intercept β0 (value of y on flat terrain) and parameters βi for uneven I
and uneven II are estimated for this model. The random factor variables µj are assumed to be normally
distributed about zero and account for inter- subject variability of the intercept. The model residuals
are ϵij which are also assumed to be normally distributed about zero.
Research article
Physics of Living Systems | Neuroscience
Dhawale and Venkadesan. eLife 2023;12:e67177. DOI: https:// doi. org/ 10. 7554/ eLife. 67177
11 of 20
The second linear mixed model uses stepwise data where each step is grouped by subject and
terrain type. Each of the 1086 steps in this dataset contains a value for subject number, terrain type,
touchdown leg angle, decelerating fore- aft impulse, and forward foot speed at touchdown. The linear
model for the dependent variable y (touchdown leg angle or fore- aft impulse) is
yij = (β0 + µ1j) + βiterraini + (βf + µ2j + νi)footspeed + ϵij
(11)
where i = 1, 2 and j = 1 . . . 9 . Like in Equation 10, the model estimates the intercept β0 , i.e., the value
of y on flat terrain when foot speed = 0 , βi for terrain factor, and the slope βf for the dependence
of y on forward foot speed at touchdown. The variable µ1j account for inter- subject variability of the
intercept, and the variables µ2j and νi account for inter- subject and terrain- specific variability of the
slope βf , respectively. The residuals ϵij are assumed to be normally distributed.
Using a dataset of 5262 steps from all subjects on uneven I and uneven II, we extract 1515 landing
probabilities (as detailed in ‘Quantifying foot placement patterns’). To test whether runners aimed for
terrain regions with low unevenness, we use a linear mixed model of the form,
ykl = (µ1l + ν1k) + (µ2l + ν1k)terr + ϵkl
(12)
where k = 1, 2 for the two uneven terrain and l = 1 → 9 for the nine subjects. The dependent variable
y is the probability of landing in a foot- sized cell pi,j and the independent variable ‘terr’ refers to the
median terrain height of the cell or the interquartile range of heights within the cell. The variables µ1l
accounts for subject- specific variability in the terrain- specific intercept ν1k . The variables µ2l accounts
for subject- specific variability in the terrain- specific slope ν2k .
Nondimensionalization
Following Alexander and Jayes, 1983, we express lengths in units of leg length ℓ and speed in units
of
√gℓ , where g is acceleration due to gravity. Statistically significant post- hoc comparisons are addi-
tionally reported in dimensional units using g = 9.81 m/s2, and the mean of the measurements across
subjects, namely, ℓ = 0.89 m and m = 66.1 kg.
Figure 5. Foot placement on uneven terrain. Histogram of the interquartile range of heights ( hIQR ) at footstep
locations for the directed sampling scheme (red), experiments (yellow), and the blind sampling scheme (blue)
on (a) uneven I (2526 footsteps) and (b) uneven II (2736 footsteps). Note that hIQR varies over a greater range on
uneven II.
The online version of this article includes the following source data and figure supplement(s) for figure 5:
Source data 1. Output of the Markov chain sampling (directed scheme) of the Uneven I terrain.
Source data 2. Output of the Markov chain sampling (directed scheme) of the Uneven II terrain.
Source data 3. Output of the uniform random sampling (blind scheme) of the Uneven I terrain.
Source data 4. Output of the uniform random sampling (blind scheme) of the Uneven II terrain.
Source data 5. Subject- wise, per- step data of the terrain height at foot landing locations on the Uneven I terrain.
Source data 6. Subject- wise, per- step data of the terrain height at foot landing locations on the Uneven II terrain.
Figure supplement 1. Subject- wise foot placement analysis on uneven I.
Figure supplement 2. Subject- wise foot placement analysis on uneven II.
Figure supplement 3. Subject- wise foot placement analysis.
Research article
Physics of Living Systems | Neuroscience
Dhawale and Venkadesan. eLife 2023;12:e67177. DOI: https:// doi. org/ 10. 7554/ eLife. 67177
12 of 20
Results
Foot placement on uneven terrain
To test whether real runners prefer to land on flatter patches, the measured footsteps were compared
against two extreme models, a null hypothesis of a blind runner and an alternative hypothesis of a
directed runner whose footsteps are selectively aimed at level parts of the terrain. The blind scheme
uses a uniform random sample of rear- foot sized patches of the terrain to obtain statistics of the
terrain at landing locations. The directed scheme preferentially samples more level patches using an
MCMC model (‘Directed foot placement scheme’ in Methods).
The experimentally measured stepping patterns are the same as the blind scheme on both uneven
I and II in terms of the terrain unevenness as quantified by hIQR (human subjects versus blind scheme
in Figure 5). However, the directed scheme finds substantially more level landing patches, showing
that it was possible for the runners to land on more level ground (directed scheme in Figure 5). These
trends are also borne out in a subject- wise analysis (Figure 5—figure supplements 1 and 2).
The directed scheme found more level patches and exhibited decreased variability in step length
and step width compared with the experimental data. The mean step length and width of the directed
scheme are the same as the experimental data on both uneven I and uneven II. However, the stan-
dard deviation of step length decreased by 80% on both uneven I and uneven II compared to exper-
imental measurements. This corresponds to a change of 0.013 and 0.011 m for the mean subject on
uneven I and uneven II, respectively. The standard deviation of step width for the directed scheme
decreased by 80% (0.0006 m) on uneven I and by 84% (0.005 m) on uneven II compared to experi-
mental measurements.
The overall statistics of the terrain location at foot landing may obscure step- to- step dependence
of the foot landing on terrain features. A second analysis of correlating foot landing probability pi,j
Figure 6. Regulation of fore- aft impulses. (a) The fore- aft impulse J∗y (gray shaded area) is found by integrating
the measured fore- aft ground reaction force Fy (black curve) during the deceleration phase. (b) Mean
J∗
y
mvy for 9
subjects on 3 terrain types. Central red lines denote the median, boxes represent the interquartile range, whiskers
extend to 1.5 times the quartile range, and open circles denote outliers. (c) Measured
J∗
y
mvy (green circles) versus
relative forward foot speed at landing (forward foot speed/center of mass speed) for each step recorded on all
terrain types (total 1081 steps). The green line is the regression fit for the data. The dark and light gray lines are
the predicted fore- aft impulse for the mean stiff and compliant jointed models, respectively. Per step model
predictions in Figure 6—figure supplement 1. (d) Measured versus predicted fore- aft impulses for every step.
The dotted line represents perfect prediction.
The online version of this article includes the following source data and figure supplement(s) for figure 6:
Source data 1. Subject- wise, per- step data of fore- aft impulse, foot speed, and touchdown angle.
Figure 6 continued on next page
Research article
Physics of Living Systems | Neuroscience
Dhawale and Venkadesan. eLife 2023;12:e67177. DOI: https:// doi. org/ 10. 7554/ eLife. 67177
13 of 20
with the interquartile range of the terrain heights in the foot- sized cell was consistent with results
described above and showed no significance (Table 1). Taken together, these results indicate that the
runners did not guide their footsteps toward flatter areas of the terrain.
Fore-aft impulses
The fore- aft ground reaction force in stance initially decelerates the center of mass before accelerating
it forward (Figure 6a). We find that less than 6 ± 1 % (mean ± SD) of the forward momentum is lost
during the deceleration phase of stance and there is no dependence on terrain or subject (Figure 6b).
The low variability of the fore- aft impulse, just 1% of the forward momentum, suggests that it is tightly
regulated across runners, terrain, and steps.
The regulation of foot speed is unlikely to be the primary determinant of the low variability in the
collision impulse. This is because the dimensionless forward foot speed at touchdown across all terrain
varied by nearly 50% of its mean ( 0.4 ± 0.2 , Table 2), whereas fore- aft collision impulses varied only by
17% of its mean. A statistical analysis lends further support and shows that the dimensionless fore- aft
impulse depends significantly, but only weakly, on the dimensionless forward foot speed at landing
(Table 3, p = 0.001 , slope = 0.01 ± 0.003 ).
To further investigate this weak dependence of the retarding impulse on foot speed, we analyzed
the mechanics of foot landing and the resultant impulse using a four- link chain model of the leg
and torso. The joints are either completely rigid or infinitely compliant when the foot undergoes
a rigid, inelastic collision with the ground (‘Collision model’ in Methods). The models at the two
extremes of joint stiffness bound the experimental data, with the compliant model underestimating
the measured fore- aft impulse while the stiff model overestimates it (Figure 6c, d, and Figure 6—
figure supplement 1). This is expected because the muscle contraction needed for weight support
and propulsion would induce non- zero but non- infinite stiffness at the joints. Although both models
overestimate the dependence of the fore- aft impulse on foot speed, the slope of the compliant model
is closest to the measurements (Figure 6c, Figure 6—figure supplement 1). The slope of measured
speed- impulse data is 0.01 ± 0.003 ( p = 0.001 , Table 3), closer to compliant model than the stiff model,
whose slopes are 0.0203 ± 0.010 ( p < 0.0001 ) and 0.056 ± 0.005 ( p < 0.0001 ), respectively. The measured
fore- aft impulse for most steps was below 0.07 (whiskers extend to 1.5 times the interquartile range
in Figure 6). The compliant model’s predicted fore- aft impulses show good agreement with measure-
ments when the impulse is below 0.07 (measured versus predicted in Figure 6d), and disagree only
for the occasional steps when runners experience more severe fore- aft impulses. Unlike the compliant
model, the stiff model consistently over- estimates the measured fore- aft impulse over its entire range.
Thus, we propose that maintaining low joint stiffness at landing helps maintain low fore- aft impulses
despite variations in touchdown foot speed.
Source data 2. Per- step data of the measured and predicted fore- aft impulse for the compliant and stiff- leg
collision models.
Figure supplement 1. Detailed results of the collision analysis.
Figure 6 continued
Table 1. Correlation between landing probability and terrain unevenness.
Details of the ANCOVAs on the linear mixed models from Equation 12 showing denominator
degrees of freedom, F- values, and p- values from the dataset of stepping probabilities and terrain
height statistics of 1515 recorded pi,j values for all subjects on uneven I and uneven II. Since the foot
placement index pi,j values show very little variability (Figure 5—figure supplement 3), the model
with the median terrain height was singular.
Independent variable
DenDF
F- value
p- Value
IQR terrain height
20.6
3.03
0.10
The online version of this article includes the following source data for table 1:
Source data 1. Subject- wise statistics of the terrain’s height in heel- sized patches and the probability of stepping
in that patch.
Research article
Physics of Living Systems | Neuroscience
Dhawale and Venkadesan. eLife 2023;12:e67177. DOI: https:// doi. org/ 10. 7554/ eLife. 67177
14 of 20
Leg retraction
Increased leg retraction rate results in reduced forward foot speed at touchdown, thereby altering the
fore- aft impulse (Karssen et al., 2015; Dhawale et al., 2019). The mean non- dimensional forward foot
speed at landing is terrain- dependent and lower by 0.17 ± 0.04 ( p = 0.001 ) on uneven I compared to
flat ground, and by 0.15 ± 0.04 ( p = 0.002 ) on uneven II compared to flat ground (Figure 7a, Table 2).
For the mean subject, these correspond to reductions in forward foot speed of 0.48 ± 0.11 m/s on
uneven I and 0.42 ± 0.11 m/s on uneven II compared to flat ground.
We find that touchdown angle depends significantly but only weakly on forward foot speed at
landing ( p ≈ 0 , slope = 0.07 ± 0.01 rad, Table 3). If the dimensionless forward foot speed at landing
Table 2. Kinematic variables on different terrain types reported as mean ± SD, except for meander values which are reported as
median ± interquartile range.
For each variable, we show details of the ANOVAs performed on the linear model in Equation 10, i.e., the F- value and p- value
for the terrain factor. The denominator degrees of freedom for all ANOVAs was 16. Post- hoc comparisons are reported when the
ANOVAs reached the significance bound of α = 0.05 .
Variable
Flat
Uneven I
Uneven II
F- value
p- Value
Net metabolic rate (W/kg)
13.1 ± 0.5
13.7 ± 0.9
13.7 ± 0.8
2.97
0.08
Median step width (%LL)
3.9 ± 1.9
4.1 ± 1.5
4.7 ± 2.0
4.53
0.03
IQR step width (% LL)
3.9 ± 1.4
4.3 ± 0.9
5.0 ± 1.2
3.65
0.05
Mean step width (%LL)
4.2 ± 1.7
4.7 ± 1.6
5.2 ± 1.7
8.69
0.003
SD step width (% LL)
2.8 ± 0.8
3.4 ± 0.6
3.6 ± 0.6
5.54
0.01
Mean step length (%LL)
128 ± 6
126 ± 9
125 ± 9
1.07
0.37
SD step length (%LL)
6 ± 1
7 ± 4
6 ± 1
0.64
0.54
Mean meander ( ×10−4 )
3.21 ± 2.59
3.97 ± 1.65
4.88 ± 4.62
1.48
0.25
SD meander ( ×10−4 )
0.67 ± 0.53
1.33 ± 1.40
1.27 ± 2.78
1.58
0.23
Mean fwd. foot speed (froude num.)
0.53 ± 0.17
0.36 ± 0.10
0.37 ± 0.12
13.08
0.0004
SD fwd. foot speed (froude num.)
0.17 ± 0.05
0.14 ± 0.05
0.18 ± 0.07
1.48
0.26
Mean CoM speed (m/s)
3.24 ± 0.07
3.21 ± 0.07
3.18 ± 0.09
2.32
0.13
SD CoM speed (m/s)
0.11 ± 0.03
0.13 ± 0.04
0.12 ± 0.03
2.00
0.17
Mean touchdown leg length (%LL)
120 ± 5
119 ± 4
119 ± 4
4.28
0.03
SD touchdown leg length (%LL)
1.1 ± 0.7
0.9 ± 0.3
1.3 ± 1.2
1.32
0.29
Mean touchdown leg angle (rad)
0.20 ± 0.02
0.20 ± 0.02
0.21 ± 0.02
3.90
0.04
SD touchdown leg angle (rad)
0.03 ± 0.02
0.02 ± 0.003
0.03 ± 0.02
2.10
0.15
The online version of this article includes the following source data for table 2:
Source data 1. Subject- wise, per- step data on foot and leg kinetics and kinematics.
Table 3. Details of the ANCOVAs performed on the linear model described in Equation 11 showing
the denominator degrees of freedom, F- value and p- value for the fixed terrain factor, and the
estimated slopes βf for the fixed forward foot speed effect.
Dependent variable
Factor
DenDF
F- value
p- Value
βf
Touchdown leg angle
Terrain
193
1.48
0.23
-
Fwd. foot speed
38
115.83
<0.0001
0.07±0.01 rad
Fore- aft impulse
Terrain
79
1.45
0.24
-
Fwd. foot speed
78
12.83
0.001
0.01±0.003
Research article
Physics of Living Systems | Neuroscience
Dhawale and Venkadesan. eLife 2023;12:e67177. DOI: https:// doi. org/ 10. 7554/ eLife. 67177
15 of 20
varied through its entire observed range from −0.2 to 1.1, it would result in a change in landing angle
of 0.08 rad or 5°.
Stepping kinematics
We find that the median non- dimensional step width is terrain dependent (Figure 7b, Table 2) and
increased on uneven II versus flat ground by 0.004 ± 0.001 ( p = 0.03 ). Step width variability, i.e., the
interquartile range of step widths within a trial, is also terrain dependent ( p = 0.05 , Figure 7c, Table 2)
and greater on uneven II versus level ground by 0.005 ± 0.002 ( p = 0.04 ). For the mean subject, median
step width increased by 4 ± 1 mm and the step width variability (IQR) increased by 6 ± 2 mm.
Energetics
The approximately 5% increase in metabolic power consumption on the uneven terrain compared to
flat we measured was not statistically significant ( p = 0.08 , Figure 7d, Table 2).
Discussion
Our primary finding is that runners do not use visual information about terrain unevenness to guide
their footsteps. In addition, the fore- aft collisions that they experience seem almost decoupled from
the forward speed with which their foot lands on the ground. Based on the modeling estimate of colli-
sional impulses and comparison with measurements, we propose that low joint stiffness underlie the
regulation of fore- aft impulses, likely contributing to stability (Dhawale et al., 2019). Taken together,
these results suggest that runners rely not on vision- based path planning, but on their body’s passive
mechanical response for remaining stable on undulating uneven terrain. Additionally, the changes
in step- width kinematics on the uneven versus flat terrain may reflect sensory feedback mediated
stepping strategies similar to those reported previously (Seipel and Holmes, 2005; Seethapathi and
Srinivasan, 2019), but more work is needed to investigate whether the differences were the result of
feedback control or simply the result of variability injected by the terrain’s unevenness.
Figure 7. Energetics and stepping kinematics. (a) Box plot of the mean forward foot speed at landing (units
of froude number). (b) Box plot of the median step width (normalized to leg length). (c) Box plot of the step
width variability. Central red lines denote the median, boxes represent the interquartile range, whiskers extend
to 1.5 times the quartile range, and open circles denote outliers. The distribution of step widths within a trial
deviated from normality and hence we report the median and the interquartile range of the distribution for each
trial (Figure 7—figure supplement 1), instead of the mean and standard deviation as is reported for all other
variables. (d) Net metabolic rate normalized to subject mass. Whiskers represent standard deviation across the
nine subjects. An ANOVA on the linear mixed model described in Equation 10 was used to determine whether
gait measures described above differed between terrain conditions with a significance threshold of 0.05.
The online version of this article includes the following source data and figure supplement(s) for figure 7:
Source data 1. Subject- wise, per- step data on step width.
Figure supplement 1. Subject- wise step width statistics.
Figure supplement 2. Representative respirometry data.
Research article
Physics of Living Systems | Neuroscience
Dhawale and Venkadesan. eLife 2023;12:e67177. DOI: https:// doi. org/ 10. 7554/ eLife. 67177
16 of 20
Measurements of fore- aft impulses have not been previously examined in the context of stability.
A previous theoretical analysis hypothesized that reducing tangential collisions and maintaining low
fore- aft impulses reduces the risk of falling by tumbling in the sagittal- plane (Dhawale et al., 2019).
Our data are consistent with this model. We find that only 6 ± 1% of the forward momentum was lost
in stance although the forward foot speed at landing varied by nearly 50%. This reduction in variability
is surprising because, all else held the same, speed and impulse are expected to be linearly related.
This suggests that the fore- aft impulse is tightly regulated by other means. By examining the role of
leg joint compliance using model- based analyses of the data, we found that the measured fore- aft
impulses were partly consistent with an idealized extreme of zero stiffness in the joints at the point of
landing. However, joint stiffness in a real runner cannot be too small because it is needed to withstand
the torques for weight support and propulsion. Thus, we propose that the low variability in fore- aft
impulses arises from active regulation of joint stiffness.
Past studies on running birds (Blum et al., 2014; Birn- Jeffery et al., 2014) provide some hints on
why leg compliance, and not foot speed, might be the preferred means to regulate fore- aft impulses.
To deal with abrupt changes in terrain height, running birds regulate foot speed and leg retraction
rates to maintain consistent leg forces and reduce discomfort or injury risk. Although our terrain has
smoothly varying terrain and not the step- like blocks used in the bird studies, our runners may still
have encountered sudden height changes because they did not precisely regulate their stepping
pattern to avoid uneven terrain areas. Like the running birds, they may have regulated foot speed
to mitigate discomfort and high forces. Thus, by employing leg compliance to reduce the fore- aft
impulse, the runners could deal with stability independent of foot speed regulation for safety and
comfort. However, caution is warranted when comparing our results with these past studies. The bird
studies used SLIP models to interpret their findings, but such models are energy conserving and unaf-
fected by slope variations that were part of our terrain design. Furthermore, the peak- to- peak height
variation of our terrain was less than 6% of the leg length, unlike Blum et al., 2014 and Birn- Jeffery
et al., 2014, who used larger step- like obstacles of 10% leg length or more. For example, we see
no change in the variability of the leg landing angle between flat and uneven terrain trials (Table 2),
which is expected if leg landing angle responded to variations in terrain height (Blum et al., 2014;
Birn- Jeffery et al., 2014). So large step- like obstacles probably induce different swing- leg control
strategies compared with undulating terrain with smaller height variations.
We found variability in step- to- step kinematics that are largely consistent with previous studies
on step- like terrain, but with some notable differences. Studies of running birds hypothesize that
crouched postures could aid stability on uneven terrain (Blum et al., 2011; Birn- Jeffery and Daley,
2012), as do human- subject data from treadmill running (Voloshina and Ferris, 2015). We find a
slight decrease in the virtual leg length at touchdown on the most uneven terrain compared to flat,
but the difference was only around 1% of the leg length (Table 2), whose effect on stability would
be negligible. We find higher leg retraction rates on uneven terrain, as also reported in running birds
(Birn- Jeffery and Daley, 2012; Blum et al., 2014). Leg retraction has been hypothesized to improve
running stability in the context of point- mass models by altering leg touchdown angle to aid stability
(Seyfarth et al., 2003; Blum et al., 2010). However, we find only a weak dependence between
leg retraction rate and leg touchdown angles. Human- subject treadmill experiments report that step
width and step length variability increased by 27% and 26%, respectively, and mean step length or
step width were the same for flat and uneven terrain (Voloshina and Ferris, 2015). Like those studies,
we find 24% greater step width variability on uneven terrain compared to flat, but no significant
changes in step length variability (Figure 7b, Table 2). We additionally find that the median step width
increased on uneven terrain by 13%. The increase in median step width that we measure could be
due to lateral stability challenges of running on relatively more complex terrain with smoothly varying
slope and height variations in all directions.
Unlike treadmill running studies, we do not find a statistically significant increase in metabolic
power consumption on uneven terrain versus flat ground, but the mean increase of around 5% is
similar to Voloshina and Ferris, 2015. The acceleration and deceleration when subjects turn around
during our overground trials could affect the metabolic energy expenditure. Therefore, caution
is warranted in comparing the absolute value of our reported energetics data with other studies
on treadmills or unidirectional running. But several aspects of the experimental design allow us to
compare the respirometry data between the different terrain types. For every subject, we ensured that
Research article
Physics of Living Systems | Neuroscience
Dhawale and Venkadesan. eLife 2023;12:e67177. DOI: https:// doi. org/ 10. 7554/ eLife. 67177
17 of 20
the breath- by- breath respirometry data stabilized within the first 3 min and only used the stabilized
value for further analyses (‘Energetics’ in Methods). If the transients had dominated the respirometry
measurements, the measurements would not have stabilized (Figure 7—figure supplement 2). The
use of the moving light bar on either side of the track ensured that the subjects maintained the same
speed on all the terrain types. Moreover, the turnaround patches were designed to have the same
terrain statistics (flat, uneven I, uneven II) as the rest of the track, thus ensuring that there were no
abrupt terrain transitions. This allowed us to control for and mitigate the effects of the turnaround
phases when comparing the results between the different terrain types.
We find no evidence that subjects used visual information from the terrain geometry to plan
footsteps despite predicted advantages to stability (Dhawale et al., 2019). This finding differs from
walking studies that highlight the role of vision in guiding step placement on natural, uneven terrain
(Matthis et al., 2018; Bonnen et al., 2021). The stochastic stepping model was able to consistently
find landing locations with lower unevenness than the human subjects, while matching the measured
mean stepping statistics and even reducing step- to- step variability, thus showing that the absence of
a foot placement strategy was not due to a lack of feasible landing locations. We speculate that foot
placement strategies are used for obstacle avoidance (Matthis and Fajen, 2014) on more complex
terrain while our terrains were designed to be continuously undulating and not have large, singular
obstacles. While our data suggest that terrain- guided foot placement strategies are not required
for stability on gently undulating terrain, it leaves open the possibility that there is a skill- learning
component to such foot placement strategies which we could not measure since our volunteers were
not experienced trail runners. Further experiments with runners of varying skill levels could test such
a hypothesis.
Conclusions
Footsteps were not directed toward flatter regions of the terrain despite predicted benefits to
stability. Instead, we found evidence for a previously uncharacterized control strategy, namely that
the body’s stabilizing mechanical response due to low fore- aft impulses was used to mitigate the
destabilizing effects of stepping on uneven areas. The limited need for visual attention may explain
how runners could employ vision for other functional goals, such as planning a path around large
obstacles, or in an evolutionary context, tracking footprints to hunt prey on uneven terrain without
falling. Whether other animals employ similar strategies on uneven terrain is presently unknown but
data from galloping dogs show that they do not alter their gait on uneven terrain (Wilshin et al.,
2020), thus suggesting that other adept runners potentially employ similar principles for stability. We
propose that our results could translate to new strategies for reducing the real- time image processing
burden in robotic systems, and could also help in training trail runners by emphasizing limber joints
when dealing with uneven terrain.
Acknowledgements
Human Frontier Science Program and Wellcome Trust- DBT Alliance for funding.
Additional information
Funding
Funder
Grant reference number
Author
Human Frontier Science
Program
RGY0091/2013
Madhusudhan Venkadesan
The Wellcome Trust DBT
India Alliance
Madhusudhan Venkadesan
The funders had no role in study design, data collection and interpretation, or the
decision to submit the work for publication. For the purpose of Open Access, the
authors have applied a CC BY public copyright license to any Author Accepted
Manuscript version arising from this submission.
Research article
Physics of Living Systems | Neuroscience
Dhawale and Venkadesan. eLife 2023;12:e67177. DOI: https:// doi. org/ 10. 7554/ eLife. 67177
18 of 20
Author contributions
Nihav Dhawale, Data curation, Software, Formal analysis, Validation, Investigation, Visualization, Meth-
odology, Writing – original draft, Writing – review and editing; Madhusudhan Venkadesan, Concep-
tualization, Resources, Data curation, Formal analysis, Supervision, Funding acquisition, Validation,
Investigation, Visualization, Methodology, Writing – original draft, Project administration, Writing –
review and editing
Author ORCIDs
Nihav Dhawale http://orcid.org/0000-0002-5193-9064
Madhusudhan Venkadesan http://orcid.org/0000-0001-5754-7478
Ethics
Human subjects: The study was approved by the Institute Ethics Committee (Human Studies) of the
National Centre for Biological Sciences, Bengaluru, India (TFR:NCB:15\_IBSC/2012), where the exper-
iments were conducted. Informed consent was obtained by the experimenter N. Dhawale and M.
Venkadesan, who are the authors of this manuscript. The procedure followed for seeking informed
consent followed the steps that were approved by the Ethics Committee mentioned above.
Decision letter and Author response
Decision letter https://doi.org/10.7554/eLife.67177.sa1
Author response https://doi.org/10.7554/eLife.67177.sa2
Additional files
Supplementary files
• MDAR checklist
Data availability
All data points plotted are in either the main text, the figure supplements, or source data attached to
figures and tables.
References
Alexander Rm, Jayes AS. 1983. A dynamic similarity hypothesis for the gaits of quadrupedal mammals. Journal
of Zoology 201:135–152. DOI: https://doi.org/10.1111/j.1469-7998.1983.tb04266.x
Arellano CJ, Kram R. 2011. The effects of step width and arm swing on energetic cost and lateral balance during
running. Journal of Biomechanics 44:1291–1295. DOI: https://doi.org/10.1016/j.jbiomech.2011.01.002, PMID:
21316058
Birn- Jeffery AV, Daley MA. 2012. Birds achieve high robustness in uneven terrain through active control of
landing conditions. The Journal of Experimental Biology 215:2117–2127. DOI: https://doi.org/10.1242/jeb.
065557, PMID: 22623200
Birn- Jeffery AV, Hubicki CM, Blum Y, Renjewski D, Hurst JW, Daley MA. 2014. Do ’'t break a leg: Running birds
from quail to ostrich prioritise leg safety and economy on uneven terrain. The Journal of Experimental Biology
217:3786–3796. DOI: https://doi.org/10.1242/jeb.102640, PMID: 25355848
Blum Y, Lipfert SW, Rummel J, Seyfarth A. 2010. Swing leg control in human running. Bioinspiration &
Biomimetics 5:026006. DOI: https://doi.org/10.1088/1748-3182/5/2/026006, PMID: 20498515
Blum Y, Birn- Jeffery A, Daley MA, Seyfarth A. 2011. Does a crouched leg posture enhance running stability and
robustness? Journal of Theoretical Biology 281:97–106. DOI: https://doi.org/10.1016/j.jtbi.2011.04.029, PMID:
21569779
Blum Y, Vejdani HR, Birn- Jeffery AV, Hubicki CM, Hurst JW, Daley MA. 2014. Swing- leg trajectory of running
guinea fowl suggests task- level priority of force regulation rather than disturbance rejection. PLOS ONE
9:e100399. DOI: https://doi.org/10.1371/journal.pone.0100399, PMID: 24979750
Bonnen K, Matthis JS, Gibaldi A, Banks MS, Levi DM, Hayhoe M. 2021. Binocular vision and the control of foot
placement during walking in natural terrain. Scientific Reports 11:20881. DOI: https://doi.org/10.1038/
s41598- 021-99846-0, PMID: 34686759
Bramble DM, Lieberman DE. 2004. Endurance running and the evolution of Homo. Nature 432:345–352. DOI:
https://doi.org/10.1038/nature03052, PMID: 15549097
Brockway JM. 1987. Derivation of formulae used to calculate energy expenditure in man. Human Nutrition
Clinical Nutrition 41:463–471 PMID: 3429265.
Carrier DR, Kapoor AK, Kimura T, Nickels MK, Scott EC, So JK, Trinkaus E. 1984. The energetic paradox of
human running and hominid evolution [ and comments and reply ]. Current Anthropology 25:483–495. DOI:
https://doi.org/10.1086/203165
Research article
Physics of Living Systems | Neuroscience
Dhawale and Venkadesan. eLife 2023;12:e67177. DOI: https:// doi. org/ 10. 7554/ eLife. 67177
19 of 20
Chatterjee A, Ruina A. 1998. Two interpretations of rigidity in rigid- body collisions. Journal of Applied
Mechanics 65:894–900. DOI: https://doi.org/10.1115/1.2791929
Daley MA, Biewener AA. 2006. Running over rough terrain reveals limb control for intrinsic stability. PNAS
103:15681–15686. DOI: https://doi.org/10.1073/pnas.0601473103, PMID: 17032779
Daley MA, Usherwood JR, Felix G, Biewener AA. 2006. Running over rough terrain: guinea fowl maintain
dynamic stability despite a large unexpected change in substrate height. The Journal of Experimental Biology
209:171–187. DOI: https://doi.org/10.1242/jeb.01986, PMID: 16354788
Dempster WT. 1955. Space Requirements of the Seated Operator, Geometrical, Kinematic, and Mechanical
Aspects of the Body with Special Reference to the Limbs. Michigan State Univ East Lansing.
Dhawale N, Mandre S, Venkadesan M. 2019. Dynamics and stability of running on rough terrains. Royal Society
Open Science 6:181729. DOI: https://doi.org/10.1098/rsos.181729, PMID: 31032027
Donelan JM, Kram R, Kuo AD. 2001. Mechanical and metabolic determinants of the preferred step width in
human walking. Proceedings. Biological Sciences 268:1985–1992. DOI: https://doi.org/10.1098/rspb.2001.
1761, PMID: 11571044
Donelan JM, Kram R, Kuo AD. 2002. Mechanical work for step- to- step transitions is a major determinant of the
metabolic cost of human walking. The Journal of Experimental Biology 205:3717–3727. DOI: https://doi.org/
10.1242/jeb.205.23.3717, PMID: 12409498
Donelan J, Shipman D, Kram R, Kuo A. 2004. Mechanical and metabolic requirements for active lateral
stabilization in human walking. Journal of Biomechanics 37:827–835. DOI: https://doi.org/10.1016/j.jbiomech.
2003.06.002, PMID: 15111070
Geyer H, Seyfarth A, Blickhan R. 2006. Compliant leg behaviour explains basic dynamics of walking and running.
Proceedings. Biological Sciences 273:2861–2867. DOI: https://doi.org/10.1098/rspb.2006.3637, PMID:
17015312
Holmes P, Full RJ, Koditschek D, Guckenheimer J. 2006. The dynamics of legged locomotion: Models, analyses,
and challenges. SIAM Review 48:207–304. DOI: https://doi.org/10.1137/S0036144504445133
Karssen JGD, Haberland M, Wisse M, Kim S. 2015. The effects of swing- leg retraction on running performance:
Analysis, simulation, and experiment. Robotica 33:2137–2155. DOI: https://doi.org/10.1017/
S0263574714001167
Kent JA, Sommerfeld JH, Mukherjee M, Takahashi KZ, Stergiou N. 2019. Locomotor patterns change over time
during walking on an uneven surface. The Journal of Experimental Biology 222:jeb202093. DOI: https://doi.
org/10.1242/jeb.202093, PMID: 31253712
Kowalsky DB, Rebula JR, Ojeda LV, Adamczyk PG, Kuo AD. 2021. Human walking in the real world: Interactions
between terrain type, gait parameters, and energy expenditure. PLOS ONE 16:e0228682. DOI: https://doi.
org/10.1371/journal.pone.0228682, PMID: 33439858
Kuznetsova A, Brockhoff PB, Christensen RHB. 2017. lmertest package: Tests in linear mixed effects models.
Journal of Statistical Software 82:1–26. DOI: https://doi.org/10.18637/jss.v082.i13
Lee DN, Lishman R. 1977. Visual control of locomotion. Scandinavian Journal of Psychology 18:224–230. DOI:
https://doi.org/10.1111/j.1467-9450.1977.tb00281.x, PMID: 897600
Lieberman DE, Venkadesan M, Werbel WA, Daoud AI, D’Andrea S, Davis IS, Mang’eni RO, Pitsiladis Y. 2010.
Foot strike patterns and collision forces in habitually barefoot versus shod runners. Nature 463:531–535. DOI:
https://doi.org/10.1038/nature08723, PMID: 20111000
Mahaki M, Bruijn SM, van Dieën JH. 2019. The effect of external lateral stabilization on the use of foot
placement to control mediolateral stability in walking and running. PeerJ 7:e7939. DOI: https://doi.org/10.
7717/peerj.7939, PMID: 31681515
Matthis JS, Fajen BR. 2014. Visual control of foot placement when walking over complex terrain. Journal of
Experimental Psychology. Human Perception and Performance 40:106–115. DOI: https://doi.org/10.1037/
a0033101, PMID: 23750964
Matthis JS, Yates JL, Hayhoe MM. 2018. Gaze and the control of foot placement when walking in natural terrain.
Current Biology 28:1224–1233.. DOI: https://doi.org/10.1016/j.cub.2018.03.008, PMID: 29657116
Müller R, Häufle DFB, Blickhan R. 2015. Preparing the leg for ground contact in running: The contribution of
feed- forward and visual feedback. The Journal of Experimental Biology 218:451–457. DOI: https://doi.org/10.
1242/jeb.113688, PMID: 25524978
Müller R, Birn- Jeffery AV, Blum Y. 2016. Human and avian running on uneven ground: A model- based
comparison. Journal of the Royal Society, Interface 13:122. DOI: https://doi.org/10.1098/rsif.2016.0529, PMID:
27655670
Ruina A, Bertram JEA, Srinivasan M. 2005. A collisional model of the energetic cost of support work qualitatively
explains leg sequencing in walking and galloping, pseudo- elastic leg behavior in running and the walk- to- run
transition. Journal of Theoretical Biology 237:170–192. DOI: https://doi.org/10.1016/j.jtbi.2005.04.004, PMID:
15961114
Seethapathi N, Srinivasan M. 2019. Step- to- step variations in human running reveal how humans run without
falling. eLife 8:e38371. DOI: https://doi.org/10.7554/eLife.38371, PMID: 30888320
Seipel JE, Holmes P. 2005. Running in three dimensions: Analysis of a point- mass sprung- leg model. The
International Journal of Robotics Research 24:657–674. DOI: https://doi.org/10.1177/0278364905056194
Seyfarth A, Geyer H, Günther M, Blickhan R. 2002. A movement criterion for running. Journal of Biomechanics
35:649–655. DOI: https://doi.org/10.1016/s0021-9290(01)00245-7, PMID: 11955504
Seyfarth A, Geyer H, Herr H. 2003. Swing- leg retraction: A simple control model for stable running. The Journal
of Experimental Biology 206:2547–2555. DOI: https://doi.org/10.1242/jeb.00463, PMID: 12819262
Research article
Physics of Living Systems | Neuroscience
Dhawale and Venkadesan. eLife 2023;12:e67177. DOI: https:// doi. org/ 10. 7554/ eLife. 67177
20 of 20
Srinivasan M, Ruina A. 2006. Computer optimization of a minimal biped model discovers walking and running.
Nature 439:72–75. DOI: https://doi.org/10.1038/nature04113, PMID: 16155564
Thomas NDA, Gardiner JD, Crompton RH, Lawson R. 2020. Keep your head down: Maintaining gait stability in
challenging conditions. Human Movement Science 73:102676. DOI: https://doi.org/10.1016/j.humov.2020.
102676, PMID: 32956985
Venkadesan M, Mandre S, Bandi MM. 2017. Biological feet: evolution, mechanics and applications. Sharbafi MA,
Seyfarth A (Eds). Bioinspired Legged Locomotion Butterworth- Heinemann. Yale University. p. 461–486. DOI:
https://doi.org/10.1016/B978-0-12-803766-9.00010-5
Voloshina AS, Ferris DP. 2015. Biomechanics and energetics of running on uneven terrain. The Journal of
Experimental Biology 218:711–719. DOI: https://doi.org/10.1242/jeb.106518, PMID: 25617451
Warren WH, Young DS, Lee DN. 1986. Visual control of step length during running over irregular terrain. Journal
of Experimental Psychology 12:259–266. DOI: https://doi.org/10.1037/0096-1523.12.3.259
Wilshin S, Reeve MA, Spence AJ. 2020. Dog galloping on rough terrain exhibits similar limb co- ordination
patterns and gait variability to that on flat terrain. Bioinspiration & Biomimetics 16:015001. DOI: https://doi.
org/10.1088/1748-3190/abb17a, PMID: 33684074
| How human runners regulate footsteps on uneven terrain. | 02-22-2023 | Dhawale, Nihav,Venkadesan, Madhusudhan | eng |
PMC5100986 | RESEARCH ARTICLE
Movement Demands of Elite Under-20s and
Senior International Rugby Union Players
Daniel J. Cunningham1☯, David A. Shearer2,3☯, Scott Drawer4‡, Ben Pollard4‡,
Robin Eager4‡, Neil Taylor4, Christian J. Cook1, Liam P. Kilduff1,3☯*
1 Applied Sport Technology Exercise and Medicine Research Centre (A-STEM), College of Engineering,
Swansea University, Swansea, Wales, 2 School of Psychology and Therapeutic Studies, University of South
Wales, Rhondda Cynon Taff, Wales, 3 Welsh Institute of Performance Science, College of Engineering,
Swansea University, Swansea, Wales, 4 The Rugby Football Union, Greater London, England
☯ These authors contributed equally to this work.
‡ These authors also contributed equally to this work.
* [email protected]
Abstract
This study compared the movement demands of elite international Under-20 age grade
(U20s) and senior international rugby union players during competitive tournament match
play. Forty elite professional players from an U20 and 27 elite professional senior players
from international performance squads were monitored using 10Hz global positioning sys-
tems (GPS) during 15 (U20s) and 8 (senior) international tournament matches during the
2014 and 2015 seasons. Data on distances, velocities, accelerations, decelerations, high
metabolic load (HML) distance and efforts, and number of sprints were derived. Data files
from players who played over 60 min (n = 258) were separated firstly into Forwards and
Backs, and more specifically into six positional groups; FR–Front Row (prop & hooker),
SR–Second Row, BR–Back Row (Flankers & No.8), HB–Half Backs (scrum half & outside
half), MF–Midfield (centres), B3 –Back Three (wings & full back) for match analysis. Linear
mixed models revealed significant differences between U20 and senior teams in both the
forwards and backs. In the forwards the seniors covered greater HML distance (736.4 ±
280.3 vs 701.3 ± 198.7m, p = 0.01) and severe decelerations (2.38 ± 2.2 vs 2.28 ± 1.65, p =
0.05) compared to the U20s, but performed less relative HSR (3.1 ± 1.6 vs 3.2 ± 1.5, p <
0.01), moderate (19.4 ± 10.5 vs 23.6 ± 10.5, p = 0.01) and high accelerations (2.2 ± 1.9 vs
4.3 ± 2.7, p < 0.01) and sprint•min-1 (0.11 ± 0.06 vs 0.11 ± 0.05, p < 0.01). Senior backs cov-
ered a greater relative distance (73.3 ± 8.1 vs 69.1 ± 7.6 m•min-1, p < 0.01), greater High
Metabolic Load (HML) distance (1138.0 ± 233.5 vs 1060.4 ± 218.1m, p < 0.01), HML efforts
(112.7 ± 22.2 vs 98.8 ± 21.7, p < 0.01) and heavy decelerations (9.9 ± 4.3 vs 9.5 ± 4.4, p =
0.04) than the U20s backs. However, the U20s backs performed more relative HSR (7.3 ±
2.1 vs 7.2 ± 2.1, p <0.01) and sprint•min-1 (0.26 ± 0.07 vs 0.25 ± 0.07, p < 0.01). Further
investigation highlighted differences between the 6 positional groups of the teams. The
positional groups that differed the most on the variables measured were the FR and MF
groups, with the U20s FR having higher outputs on HSR, moderate & high accelerations,
moderate, high & severe decelerations, HML distance, HML efforts, and sprints•min-1. For
the MF group the senior players produced greater values for relative distance covered,
PLOS ONE | DOI:10.1371/journal.pone.0164990
November 8, 2016
1 / 13
a11111
OPEN ACCESS
Citation: Cunningham DJ, Shearer DA, Drawer S,
Pollard B, Eager R, Taylor N, et al. (2016)
Movement Demands of Elite Under-20s and Senior
International Rugby Union Players. PLoS ONE 11
(11): e0164990. doi:10.1371/journal.
pone.0164990
Editor: Karen Hind, Leeds Beckett University,
UNITED KINGDOM
Received: March 1, 2016
Accepted: October 4, 2016
Published: November 8, 2016
Copyright: © 2016 Cunningham et al. This is an
open access article distributed under the terms of
the Creative Commons Attribution License, which
permits unrestricted use, distribution, and
reproduction in any medium, provided the original
author and source are credited.
Data Availability Statement: Data are available
from the Swansea University Ethics Committee for
researchers who meet the criteria for access to
confidential data.
Funding: The author(s) received no specific
funding for this work and no authors have any
financial or other interest in the products or
distributor of the products named in the study. The
Rugby Football Union provided support in the form
of salaries for authors Scott Drawer, Robin Eager,
Ben Pollard and Neil Taylor, but did not have any
additional role in the study design, data collection
HSR, moderate decelerations, HML distance and sprint•min-1. The BR position group was
most similar with the only differences seen on heavy accelerations (U20s higher) and mod-
erate decelerations (seniors higher). Findings demonstrate that U20s internationals appear
to be an adequate ‘stepping stone’ for preparing players for movement characteristics
found senior International rugby, however, the current study highlight for the first time that
certain positional groups may require more time to be able to match the movement
demands required at a higher playing level than others. Conditioning staff must also bear in
mind that the U20s players whilst maintaining or improving match movement capabilities
may require to gain substantial mass in some positions to match their senior counterparts.
Introduction
Rugby union is a high intensity intermittent sport, where periods of intense static exertions,
collisions and running at various intensities are interspersed with random periods of lower
intensity work and rest [1, 2]. Recent work has characterised the movement demands of senior
professional rugby union players [1, 3–7]. There is, however, a lack of literature on movement
demands, physical characteristics and match analysis at the highest level of rugby union (inter-
national), with only a few studies published on work:rest ratios [8, 9], endocrine response [10,
11], time motion analysis [12] and a recent publication on movement demands [13]. A study
by Quarrie et al., [12] reported rugby union players covered on average between 5.5 and 6.3 km
per game during 27 international matches observedin their study. Backs generally covered
greater distances compared to the forwards, conversely forwards sustained greater contact
loads from scrums, rucks and mauls. These researchers utilised video and player tracking soft-
ware to quantify distances and contact elements of the game. Unfortunately the current micro-
sensor technology appears inadequate in quantifying the collision based elements of the game
[14]. A cluster analysis revealed 5 distinct groups of players (e.g. props, second rows, back row,
wings & fullback (back 3), centres & fly half)with the authors suggesting that hookers should
be grouped with props or second rows and not back row players, which has previously been the
case in older time-motion studies [15–17]. Although positional groups covered similar dis-
tances during matches, the distances they covered at various speed zones varied considerably
and the amount of game time also varied significantly across positions due to tactical substitu-
tions [12]. They also suggested that there was a difference in the amount of high speed running
(>5m•s-1) performed at international level compared to lower levels of the game and therefore,
players hoping to compete at international level need to be conditioned for the increased inten-
sity of match play [12].
In addition, there is little in the literature about the elite development pathways (e.g. U20s
internationals). With the exception of the work by Lombard et al., [18] on the 10 year physical
evolution of South African U20s players, the research of Barr and colleagues [19] reporting
speed characteristics of U20s players from a nation outside of the top 10 (i.e. tier 2), and move-
ment characteristics in U20 players However, currently no literature exists on how these
demands map with those of their senior counterparts. This information would be useful in
order to prepare players for the movement demands of senior rugby, whilst minimising the
risk of injuries by monitoring playing/training ensuring acute:chronic workloads are appropri-
ate [20].
There has been a historical evolution of both senior and junior rugby players from a physical
perspective [18, 21] with the rate of increase in body mass particularly large in the last 30–40
Movement Demands of Elite Under-20s and Senior International Rugby Union Players
PLOS ONE | DOI:10.1371/journal.pone.0164990
November 8, 2016
2 / 13
and analysis, decision to publish, or preparation of
the manuscript. The specific roles of these authors
are articulated in the ‘author contributions’ section.
Competing Interests: We have the following
interests. Scott Drawer, Robin Eager, Ben Pollard
and Neil Taylor are employed by The Rugby
Football Union. There are no patents, products in
development or marketed products to declare. This
does not alter our adherence to all the PLOS ONE
policies on sharing data and materials, as detailed
online in the guide for authors.
years [21]. For example, South African U20s players have increased in height (~2.8%), weight
(~14%), strength (~51%), muscular endurance (~50%), and improved speed times over 10m
(~7%) and 40m (~4%) but have not improved aerobic performance over a 13 year period [18].
This change is most likely due to the amount of training time available since the advent of pro-
fessionalism and the training method advances made within the domain of strength and condi-
tioning coupled with the desire for larger rugby players in order to gain the upper hand in the
collision/contact area of the game. Despite seniors and juniors showing rapid developments in
physical characteristics there still appears to be differences between these groups. For example,
at the 2015 6 Nations tournament (the highest level of international competition in the north-
ern hemisphere) the average weight of a senior English forward was 115.8 ± 8.12 kg compared
to 110.3 ± 8.14 kg for an U20s forward. Similarly for the backs 93.8 ± 6.9 vs 89.9 ± 5.7 kg for
seniors and U20s (unpublished data). Argus and co-workers [22] also found moderate to very
large differences in mass, upper and lower body strength and upper body and lower body
power between academy and senior professional southern hemisphere rugby union players.
Interestingly Barr et al., [19] found no significant differences between senior and U20s interna-
tionals for initial and maximal sprint velocity, however, when initial and maximal sprint
momentum was calculated there were significant differences between the groups. These find-
ings are further corroborated by the work of Hansen and colleagues [23] who reported signifi-
cant differences between elite senior and junior players for mass and measures of strength and
power, but not for speed times over 5, 10 and 30m.
In rugby league, significant differences in distance travelled during match play between pro-
fessional senior and elite junior players have been reported by McClellan & Lovell [24]. Specifi-
cally, the mean total distance travelled during professional games (8371 ± 897 m) was
significantly greater than that travelled during elite junior (4646 ± 978 m) match-play.
Research indicates a progression of physical characteristics between playing levels in rugby lea-
gue players, Gabbett [25] outlined the progressive improvement in the physiological capacities
(mass, speed, agility and aerobic endurance) of rugby league players as the playing level
increases from U13 right the way through to professional level. A similar progression in physi-
ological characteristics was reported in an elite English rugby union academy, when Darrall-
Jones et al., [26] undertook a comprehensive testing battery with the U16, U18 and U21 acad-
emy squads. They reported a progressive increase between the groups on mass, strength, power
and momentum, however no differences were found on aerobic endurance or speed times.
Recent work by Gannon et al [27] showed improvements in strength and power measures over
the course of a season in a professional club environment. The largest improvements were seen
in the early to mid-season period with a drop off towards the end of the season but still achiev-
ing an overall gain from the start of the season. Neither of these studies made any comparison
between age grade players and seniors.
A greater understanding of player movement patterns in senior and U20s international
rugby union, may give an indication of the positional requirements of performance (as shown
in senior professional rugby by Lindsay et al., [3]). Which may also aid in targeting qualities/
physical outputs players need to work on. This will facilitate the planning and implementation
of training programmes and development pathways that elicit the required physiological adap-
tations specific to individual player needs, whilst ensuring increases in training/playing load
are applied appropriately to minimise the risk of injury [20]. Conversely, it could help identify
outstanding performers who could be fast tracked into the senior set up. Therefore, assessing
how U20s compare to seniors and ascertaining where this development tool sits in the player
pathway to senior international honours needs to be addressed. Therefore, the aim of this
study was to compare U20s international competitions with senior international competitions
based on movement demands recorded using GPS devices.
Movement Demands of Elite Under-20s and Senior International Rugby Union Players
PLOS ONE | DOI:10.1371/journal.pone.0164990
November 8, 2016
3 / 13
Methods
Elite professional junior players from an U20s international performance squad (n = 43), and
elite professional senior players from an international performance squad (n = 27) participated
in the study. Prior to providing written informed consent, participants were given information
outlining the rationale, potential applications and procedures associated with the study. Ethics
approval was granted by the Swansea University Ethics Committee. All players were consid-
ered healthy and injury-free at the time of the study and were in full-time training. Players
were grouped broadly as forwards and backs and more specificallyin sub units within those
groups. With front row (FR), second row (SR) and back row (BR) making up the forwards.
Half backs (HB), midfield/centres (MF) and back three (B3) players making up the backs. The
U20s players (Table 1) provided a total of 161 GPS files from 15 games from two 6 Nations
tournaments (2014 and 2015) and the 2015 Junior World Cup. The senior players (Table 1)
provided a total of 97 GPS files from 8 games from the 2014 and 2015 6 Nations tournaments.
Previous studies have shown that substitute players display greater work-rates compared to
players who start the match, suggesting that these players do not pace their involvement [28].
Therefore, to be included in the analysis players had to complete 60 mins match time [5, 29].
The seniors won all 8 games (5 home, 3 away), the U20 won 8/10 (5 home, 5 away) in the 6
Nations and 4/5 at the Junior World Cup (neutral venues). The average points scored per game
was 31.3 and 36, and points conceded 15.4 and 11.4 for senior and U20 respectively. Each
player provided at least 1 GPS file with the largest number of files provided by any one player
being 11 and 8 in the U20s and seniors, respectively. A total of 79 GPS units were used during
the study, units were returned to the manufacturer at the end of each competition for mainte-
nance/repair.
Procedures
All Matches took place between January 2014 and June 2015, each player wore a GPS unit
(Viper Pod, STATSport, Belfast, UK) in a bespoke pocket incorporated into their playing jersey
on the upper thoracic spine between the scapulae to reduce movement artefacts [30]. The GPS
units captured data at a sampling frequency of 10Hz utilising the 4 best available satellites.
Recent advancements in GPS technology have made 10 Hz units commercially available,
Table 1. Anthropometric Characteristics of Position Groups.
Positional Group
Team
Age (years) M ± SD
Height (cm) M ± SD
Mass (kg) M ± SD
FR
Senior
26.1 ± 2.3
185.7 ± 4.2
119.1 ± 5.0
U20
19.5 ± 0.7
184.4 ± 3.0
111.8 ± 5.6
SR
Senior
26.4 ± 3.3
199.2 ± 1.6
116.8 ± 4.8
U20
19.7 ± 0.5
199.7 ± 2.3
115.2 ± 4.1
BR
Senior
26.0 ± 3.3
190.0 ± 2.6
117.7 ± 10.4
U20
19.9 ± 0.3
187.7 ± 2.7
101.6 ± 3.9
HB
Senior
24.2 ± 2.5
179.5 ± 6.0
88.7 ± 4.6
U20
19.6 ± 0.4
176.0 ± 2.1
84.2 ± 4.1
MF
Senior
25.7 ± 1.3
190.2 ± 4.1
102.3 ± 6.9
U20
19.5 ± 0.6
183.0 ± 4.9
96.1 ± 6.6
B3
Senior
24.6 ± 3.4
182.6 ± 4.1
91.7 ± 2.1
U20
19.6 ± 0.5
183.7 ± 4.3
89.6 ± 4.9
FR = Front Row (Prop & Hooker), SR = Second Row, BR = Back Row, HB = Half Backs, MF = Midfield/Centres, B3 = Back Three (Wing & Full Back).
doi:10.1371/journal.pone.0164990.t001
Movement Demands of Elite Under-20s and Senior International Rugby Union Players
PLOS ONE | DOI:10.1371/journal.pone.0164990
November 8, 2016
4 / 13
which are more accurate for quantifying movement patterns in team sports [31, 32]. For exam-
ple, Varley et al., [32] reported that a 10 Hz GPS unit was two to three times more accurate for
instantaneous velocity during tasks completed at a range of velocities compared to a criterion
measure, 6 times more reliable for measuring maximum instantaneous velocity and had a coef-
ficient of variation less than or similar to the calculated smallest worthwhile change [33] during
all phases of acceleration/deceleration.More specificallythis brand of GPS devices has been
used in team sports to assess movement demands during training and competitive matches
[13, 34–39]. In our study, all participants were already familiarized with the devices as part of
their day-to-day training and playing practices. Units were activated according to the manufac-
turer’s guidelines immediately prior to the pre-match warm-up (~30–60 minutes before kick-
off),and to avoid inter-unit variation players wore the same GPS device for each match. Post
match, timings from the game were (e.g. kick off, half time, sin binning etc) were entered into
the software the raw data files were then processed and data for distance covered, and accelera-
tion/deceleration events in pre-set zones were derived automatically by the software (Viper
PSA software, STATSports, Belfast, UK).
Locomotor Variables
The distance relative to playing time (m•min-1), high speed running (HSR) relative to playing
time >18.1km•h-1 (the threshold used in numerous rugby GPS studies in both codes; e.g. Aus-
tin & Kelly [40] & Jones et al., [4]), number of sprints relative to playing time (sprints•min-1),
moderate, high and severe intensity accelerations and decelerations (±2-3m•s-2, ±3-4m•s-2,
±>4m•s-2), high metabolic load distance (HML; defined as distance covered accelerating and
decelerating over 2 m•s-2 and/or distance covered >5 m•s-1), and high metabolic load efforts
(the number of separate movements/efforts undertaken in producing HML distance). Total
time was calculated for ‘playing time’ only, that is, the time the player was on the playing field
only, with time off the field (e.g., half time, periods on the bench/sin bin) removed from the
data analysis. Time off during match play, such as injury time or video referee, was included in
the study, because this was part of the game duration; hence ‘playing time’ may exceed the stan-
dard 80 minutes of match play.
Data Analysis
Linear mixed models were used to examine each dependent variable for the interaction
between teams in respect to positional types (forwards and backs) and groups (e.g., front rows,
second rows etc.). To allow for the nested design of the data, random intercepts were modelled
for participants (individual GPS measures), teams (i.e., U20 vs Senior) and competition (U20 6
Nations 2014, U20 6 Nations 2016, U20 Junior World Cup 2015, Senior 6 Nations 2014, Senior
6 Nations 2015). Attempts were made to also model for random slopes for the same variables
but this resulted in over-specified models. Where significant interactions were identified, dif-
ferences were interpreted using a combination of estimates of fixed effects, examination of
means and 95% confidence intervals (Tables C and D in S1 File).
Results
Examination of means and standard deviations indicated visible difference between teams as a
function of position type (Table 2). Linear mixed models indicated a significant interaction
between Team (U20 v Senior) and Positional Type (Forwards and Backs) for M•min-1 (p <
.001), HSR m•min-1 (p < .001), Accelerations 2-3m•s-2 (p < .01), Accelerations 3-4m•s-2 (p <
.001), Decelerations 3-4m•s-2 (p < .001), Decelerations >4m•s-2 (p < .01), HML Distance (m)
(p < .001), Sprint•min-1 (p < .001), and HML Efforts (p < .001). Further examination of fixed
Movement Demands of Elite Under-20s and Senior International Rugby Union Players
PLOS ONE | DOI:10.1371/journal.pone.0164990
November 8, 2016
5 / 13
effects for these significant interactions revealed that U20 forwards had significantly higher
HSR m•min-1 (p < .001, CI: -3.33 to– 1.08), Accelerations 2-3m•s-2 (p < .01, CI: -9.99 to–
1.46), Accelerations 3-4m•s-2 (p < .001, CI: -4.45 to– 1.72), Decelerations 3-4m•s-2 (p < .05,
CI: -3.87 to– 0.44), Sprint•min-1 (p < .001, CI: -.12 to– 0.04), and significantly lower Accelera-
tions >4m•s-2 (p < .05, CI: 0.13 to 3.66) and HML Distance (p < .05, CI: -292.12 to– 41.40).
For backs, fixed effect revealed U20 players had significantly lower values for M•min-1 (p <
.001, CI: 4.52 to 11.34), HML Distance (p < .001, CI: 140.65 to 402.96) HML efforts (p < .001,
CI: 10.74 to 34.20) and Decelerations 3-4m•s-2 (p < .05, CI: 0.13 to 3.66), while significantly
higher values for HSR m•min-1 (p < .001, CI: 1.15 to 3.51) and Sprint•min-1 (p < .001, CI: 0.04
to 0.13). However, some other variables were close to significance also (Table C in S1 File).
Examination of means and standard deviations indicated visible difference between teams
as a function of positional groups (Table 3). Linear mixed models indicated a significant inter-
action between team (U20 v Senior) and positional groups (e.g., half-back, second rows etc.)
for M•min-1 (p < .05), Accelerations 2-3m•s-2 (p < .05), Decelerations 2-3m•s-2 (p < .001),
Decelerations 3-4m•s-2 (p> .05), HML Distance (p < .01), and HML efforts (p > .001). Esti-
mates of fixed effects were used to indicate differences between specific playing group between
teams where interactions occurred.U20 Front Rows scored significantly higher than seniors
for HSR m•min-1 (p < .001, CI: -5.15 to -1.61), Accelerations 2-3m•s-2 (p < .001, CI: -21.08 to
-8.09), Accelerations 3-4m•s-2 (p < .001, CI: -6.35 to -1.84), Decelerations 2-3m•s-2 (p < .001,
CI: -17.78 to -6.79), Decelerations 3-4m•s-2 (p < .001, CI: -7.55 to -2.20), Decelerations
>4m•s-2 (p < .01, CI: -4.04 to -0.59), HML Distance (p < .001, CI: -618.18 to -240.92), HML
efforts (p < .001, CI: -51.90 to -20.07) and Sprint•min-1 (p < .001, CI: -0.20 to -0.07). U20 Sec-
ond rows scored significantly higher for HSR m•min-1 (p < .05, CI: -3.92 to -0.13), Accelera-
tions 3-4m•s-2 (p < .01, CI: -5.72 to -0.91), however seniors performed more Sprint•min-1
(p < .001, CI: -0.20 to -0.07). U20 Back rows had significantly less Decelerations 2-3m•s-2 (p <
.01, CI: 1.74 to 11.46) but more Accelerations 3-4m•s-2 (p < .05, CI: -4.26 to -0.14). U20 Half
back had significantly lower scores for for M•min-1 (p < .001 CI: 6.48 to 17.80), Decelerations
2-3m•s-2 (p < .05, CI: 0.20 to 11.86), HML Distance (p < .05, CI: 56.47 to 503.78) and HML
efforts (p < .001, CI: 19.72 to 55.84). U20 Midfield players had significantly lower values for
Table 2. Movement Characteristics for Senior and U20s, Forwards and Backs Groups.
Position Group
Forwards
Backs
Seniors (n = 15)
U20s (n = 21)
Seniors (n = 12)
U20s (n = 22)
GPS Variable
M ± SD
M ± SD
M ± SD
M ± SD
M•min-1
66.8 ± 7.0
61.5 ± 8.0
73.3 ± 8.1*
69.1 ± 7.6
HSR m•min-1
3.1 ± 1.6
3.2 ± 1.5^
7.2 ± 2.1
7.3 ± 2.1*
HML Distance (m)
736.4 ± 280.3^
701.3 ± 198.7
1138.0 ± 233.5*
1060.4 ± 218.1
HML Efforts
84.8 ± 30.4
78.8 ± 21.5
112.7 ± 22.2*
98.8 ± 21.7
Accelerations 2-3m•s-2
19.42 ± 10.5*
23.6 ± 8.9^
26.4 ± 8.4
26.1 ± 10.1
Accelerations 3-4m•s-2
2.2 ± 1.9*
4.3 ± 2.7^
4.9 ± 3.0*
6.4 ± 4.5
Accelerations >4m•s-2
0.69 ± 0.95
0.47 ± 0.84
1.04 ± 1.22
0.89 ± 1.37
Decelerations 2-3m•s-2
24.56 ± 11.5
25.2 ± 9.3
28.4 ± 7.7*
25.3 ± 9.3
Decelerations 3-4m•s-2
6.4 ± 4.0
7.5 ± 3.5^
9.9 ± 4.3*
9.5 ± 4.4
Decelerations >4m•s-2
2.38 ± 2.2^
2.28 ± 1.65
4.39 ± 2.77
4.95 ± 3.0
Sprint•min-1
0.11 ± 0.06
0.11 ± 0.05^
0.25 ± 0.07
0.26 ± 0.07*
^ = Significantly higher than either Senior or U20 forwards counterpart.
* = Significantly higher than either Senior or U20 backs counterpart.
doi:10.1371/journal.pone.0164990.t002
Movement Demands of Elite Under-20s and Senior International Rugby Union Players
PLOS ONE | DOI:10.1371/journal.pone.0164990
November 8, 2016
6 / 13
Table 3. Movement Characteristics Presented by Playing Groups.
Position Group
FR
SR
BR
HB
MF
B3
Team
M ± SD
Team
M ± SD
Team
M ± SD
Team
M ± SD
Team
M ± SD
Team
M ± SD
M•min-1
U20s
60.1 ± 7.2
U20s
60.8 ± 5.9
U20s
63.2 ± 9.7
U20s
67.5 ± 9.1*
U20s
70.5 ± 6.8*
U20s
68.7 ± 7.6*
Seniors
61.1 ± 7.9
Seniors
67.6 ± 6.5
Seniors
69.9 ± 4.3
Seniors
77.4 ± 5.6
Seniors
71.9 ± 10.0
Seniors
70.8 ± 7.1
HSR m•min-1
U20s
2.5 ± 1.3*
U20s
3.0 ± 1.1*
U20s
4.0 ± 1.6
U20s
5.5 ± 2.4
U20s
7.2 ± 1.7*
U20s
8.1 ± 1.7*
Seniors
1.8 ± 1.1
Seniors
2.9 ± 1.2
Seniors
4.0 ± 1.4
Seniors
6.3 ± 1.6
Seniors
8.0 ± 2.3
Seniors
7.4 ± 2.2
HML Distance (m)
U20s
584.9 ± 199.1*
U20s
673.3 ± 124.1
U20s
820.6 ± 182.5
U20s
954.8 ± 304.0*
U20s
1103.4 ± 168.6*
U20s
1069.7 ± 203.0
Seniors
452.3 ± 172.9
Seniors
747.1 ± 227.9
Seniors
911.4 ± 210.2
Seniors
1144.5 ± 175.9
Seniors
1205.4 ± 265.6
Seniors
1076.1 ± 246.0
HML Efforts
U20s
66.0 ± 22.6*
U20s
77.2 ± 14.5
U20s
90.7 ± 19.0
U20s
99.9 ± 27.2*
U20s
105.0 ± 15.9
U20s
93.4 ± 22.4
Seniors
54.6 ± 19.3
Seniors
85.8 ± 26.7
Seniors
103.4 ± 22.4
Seniors
126.3 ± 14.2
Seniors
113.9 ± 23.1
Seniors
99.7 ± 20.5
Accelerations 2-3m•s-2
U20s
17.8 ± 6.6*
U20s
22.9 ± 7.4
U20s
29.0 ± 8.5
U20s
23.5 ± 13.6
U20s
27.4 ± 10.3
U20s
26.1 ± 8.1
Seniors
10.0 ± 6.0
Seniors
21.3 ± 9.4
Seniors
24.4 ± 9.5
Seniors
26.8 ± 7.9
Seniors
28.5 ± 9.4
Seniors
24.4 ± 7.9
Accelerations 3-4m•s-2
U20s
3.5 ± 2.4*
U20s
3.8 ± 2.1*
U20s
5.5 ± 3.1*
U20s
4.3 ± 5.4
U20s
5.9 ± 2.8
U20s
7.6 ± 4.9
Seniors
1.1 ± 1.3
Seniors
1.8 ± 1.9
Seniors
3.1 ± 2.0
Seniors
4.8 ±2.9
Seniors
4.0 ± 3.0
Seniors
5.7 ± 3.0
Accelerations >4m•s-2
U20s
0.39 ± 0.75
U20s
0.25 ± 0.53*
U20s
0.71 ± 1.04
U20s
0.33 ± 0.49
U20s
0.45 ± 0.78
U20s
1.47 ± 1.73
Seniors
0.50 ± 0.65
Seniors
0.83 ± 1.19
Seniors
0.73 ± 0.98
Seniors
1.19 ± 1.22
Seniors
1.07 ± 1.44
Seniors
0.89 ± 1.08
Decelerations 2-3m•s-2
U20s
21.1 ± 9.0*
U20s
24.8 ± 9.8
U20s
28.8 ± 7.9*
U20s
24.7 ± 9.9*
U20s
28.1 ± 8.9*
U20s
23.3 ± 9.1
Seniors
13.0 ± 6.1
Seniors
24.5 ± 10.4
Seniors
32.0 ± 8.4
Seniors
31.1 ± 5.6
Seniors
31.1 ± 8.6
Seniors
23.9 ± 6.7
Decelerations 3-4m•s-2
U20s
6.2 ± 3.7*
U20s
8.0 ± 3.0
U20s
8.2 ± 3.4
U20s
6.5 ± 3.6
U20s
11.5 ± 4.1*
U20s
9.2 ± 4.3
Seniors
3.5 ± 2.4
Seniors
5.7 ± 2.4
Seniors
8.6 ± 4.3
Seniors
9.4 ± 4.8
Seniors
11.3 ± 4.2
Seniors
9.2 ± 3.9
Decelerations >4m•s-2
U20s
2.2 ± 1.9*
U20s
1.6 ± 1.3
U20s
2.9 ± 1.5
U20s
3.0 ± 2.0
U20s
5.2 ± 3.5
U20s
5.5 ± 2.6*
Seniors
1.0 ± 1.5
Seniors
2.7 ± 2.2
Seniors
3.1 ± 2.3
Seniors
3.7 ± 2.3
Seniors
4.3 ± 2.7
Seniors
5.1 ± 3.1
Sprint•min-1
U20s
0.09 ± 0.04*
U20s
0.10 ± 0.03*
U20s
0.14 ± 0.05
U20s
0.18 ± 0.06
U20s
0.27 ± 0.06*
U20s
0.29 ± 0.06*
Seniors
0.06 ± 0.04
Seniors
0.11 ± 0.04
Seniors
0.14 ± 0.05
Seniors
0.21 ± 0.06
Seniors
0.28 ± 0.07
Seniors
0.27 ± 0.08
FR = Front Row (Prop & Hooker), SR = Second Row, BR = Back Row, HB = Half Backs, MF = Midfield/Centres, B3 = Back Three (Wing & Fullback).
* = significant difference to Senior counterpart
doi:10.1371/journal.pone.0164990.t003
Movement Demands of Elite Under-20s and Senior International Rugby Union Players
PLOS ONE | DOI:10.1371/journal.pone.0164990
November 8, 2016
7 / 13
M•min-1 (p < .05, CI: 1.27 to 12.34), HSR m•min-1 (p < .001, CI: 1.05 to 5.05), Decelerations
2-3m•s-2 (p < .05, CI: 0.21 to 11.73), HML Distance (p < .01, CI: 122.37 to 543.28) and HML
efforts (p < .01, CI: 7.07 to 41.74), Sprint•min-1 (p < .001, CI: 0.04 to 0.18), but higher values
for Decelerations 3-4m•s-2 (p < .05, CI: 0.24 to 5.99). Finally, U20 Back Three players had sig-
nificantly lower values for M•min-1 (p < .05, CI: 1.12 to 10.58), and HML Distance (p < .01,
CI: 51.45 to 399.61), but higher values for HSR m•min-1 (p < .001, CI: 0.81 to 4.09), Decelera-
tions >4m•s-2 (p < .05, CI: 0.20 to 3.39) and Sprint•min-1 (p < .001, CI: 0.04 to 0.15).
Discussion
The aim of this study was to compare the locomotor demands of senior international and age
grade international (U20) rugby union matches using GPS devices. The current study is the first
to present an analysis of movement demands of senior international competition in comparison
to the elite junior international competition. The results of the present study increase our under-
standing of the movement demands of competition experiencedby players in existing interna-
tional rugby union development pathways and determine whether the U20s competition reflects
the movement demands of senior match-play. Therefore, the results of the current study may
have implications for the design and implementation of physical conditioning programmes in
order to prepare players for the movement demands of senior international rugby.
In general, the seniors covered greater relative distance for both forwards (66.8 ± 7.1 vs
61.5 ± 8.0m•min-1) and backs (73.3 ± 8.1 vs 69.1 ± 7.6m•min-1), however this was only statisti-
cally significant for the backs. The U20s forwards performed more HSR m•min-1 accelerations
in zones 2–3 & 3-4m•s-2, decelerations 3-4m•s-2 and sprint•min-1 than the seniors, but less
HML distance. In the Backs, the senior group covered more relative distance (m•min-1) per-
formed more decelerations 3-4m•s-2, more HML distance & efforts, but the U20s performed
more HSR m•min-1 and sprint•min-1. The U20s also had significantly longer match time,
which could be due to different substitution strategies or a number of other factors (e.g. more
injury stoppages, discipline issues, third match official use). The relative distance values pre-
sented in the current study are lower than a recent publication from a southern hemisphere
club team [3] (Forwards: 77.3 ± 20.5, Backs: 84.7 ± 10.4m•min-1), and in between values pro-
duced by 2 different Pro 12 clubs [5, 7] (Forwards: 60.4 ± 7.8 & 71.6 ± 10.1, Backs: 67.8 ± 8.2 &
81.0 ± 10.2m•min-1). However, when comparing to data published from the Premiership (For-
wards: 64.6 IQR 6.3, Backs: 71.1 IQR 11.7 m•min-1) from which the current players are drawn,
it appears that U20s international competition is marginally below the movement demands of
the Premiership, while senior international competition is higher, in terms of relative distance
covered. This may indicate that U20s rugby is preparing players for movement demands in
Premiership rugby, which in turn will help prepare for full international matches. However,
given the likely variation in tactics/playing styles between the teams and the opposition faced,
care must taken when making comparisons [41, 42].
Although generally there are differences between senior and U20s backs and forwards, the
number of variables that were significantly different varied across each positional group. There
were no significant differences in relative distance covered between U20s and seniors in any
forward positional group (front row, second row, back row). There were significant differences
in relative distance covered for the half backs (77.37 ± 5.62 vs 67.47 ± 9.10m•min-1), midfield
(71.9 ± 10.0 vs 70.5 ± 6.8m•min-1) and back three (70.8 ± 7.1 vs 68.7 ± 7.6m•min-1) with
seniors covering greater distances in all cases. The U20s covered greater relative HSR distance
in the front & second rows and back three position groups, with the opposite being the case for
the midfield group. No differences between seniors and U20s were seen for back row or half
backs for HSR.
Movement Demands of Elite Under-20s and Senior International Rugby Union Players
PLOS ONE | DOI:10.1371/journal.pone.0164990
November 8, 2016
8 / 13
The front row and midfield groups had the most differences between seniors and U20s. Sig-
nificant differences were found between front row groups on HSR (relative), accelerations (2–3
and 3-4m•s-2), decelerations (2–3, 3–4 and >4m•s-2), HML distance, HML efforts, and
sprints•min-1 with the U20s having higher outputs on each variable. Conversely for the mid-
field group seniors had significantly greater values for relative distance covered, HSR (relative),
decelerations (2–3•s-2), HML distance and sprints•min-1. The back row group was most similar
with only accelerations (2–3•s-2), decelerations (2–3•s-2) being significantly different. Compar-
ing acceleration and deceleration data with previous literature is somewhat problematic as
Cunniffe et al. [6] reported no differences between backs and forwards groups, however, a very
low sample size was utilised (1 back, 1 forward during 1 game) together with different accelera-
tion zones. Jones et al., [5] reported distance covered while in various acceleration and deceler-
ation zones for both backs and forwards combined when investigating temporal fatigue in their
study, which makes comparison impossible, as the current study used number of acceleration
and deceleration events. Owen and co-workers [43] utilised comparable zones and reported
number of acceleration/deceleration events, their work supports the current finding that backs
are involved in more frequent acceleration and deceleration events. However, the current study
appears to have a slightly higher frequency for both backs and forwards, potentially as a result
of the level of competition (Super vs International rugby).
The number of sprints•min-1 performed was significantly different between U20s and
senior forwards (0.11 ± 0.05 vs 0.11 ± 0.06) and U20s and senior backs (0.26 ± 0.07 vs
0.25 ± 0.07). However, this is unlikely to be of practical significance given the low frequency.
Backs performed more sprints than the forwards (~x2.5) in both groups. The greatest differ-
ence between positional groups was the U20s front row group who performed almost double
the amount of their senior counterparts (8.73 ± 4.52 vs 4.71 ± 3.45). This could be due to U20s
players being lighter (119.1 ± 5.0 vs 111.8 ± 5.6 kg) and potentially more mobile, or simply a
reflection of their physical capabilities. The number of sprints reported in the current study is
higher than those reported by Jones and co-workers [5] most likely again to the difference in
playing standard (club vs international) or potentially style of play.
Overall there were a number of differences between the forwards and backs of the U20s and
senior teams. However, when broken down further into positional groups, variations in differ-
ences between the two teams in certain positions emerged. The positional groups that appeared
most different between the teams on the metrics measured were the front row and midfield
groups, with the U20s front row performing more than their senior counterparts on HSR,
moderate and heavy accelerations, all decelerations, HML distance HML efforts. As the senior
players tend to be heavier and stronger, the static exertions (not measureable by GPS) from
scrums have been shown to be greater in the senior international game [44]. This may result in
transient fatigue, whereby there is a reduction in high-intensity activity performed immediately
following an intense bout, with a subsequent recovery later in performance [45], and account
for their lower movement scores. The opposite was true for the midfield group with the seniors
producing higher scores for relative distance covered, HSR, moderate decelerations, HML dis-
tance and sprints•min-1. Perhaps indicating these players are used more frequently in a more
direct, gain-line based game plan.
High speed running (HSR) has previously been shown to distinguish between playing levels
in a number of sports (e.g. [12, 46, 47]), with the more elite levels covering greater distances in
this speed zone. However, in the current study overall the U20s forwards and backs groups,
performed more relative HSR than their senior counterparts. This wasn’t the case for each posi-
tional group however, both senior and U20 back row and half backs had no differences between
them for HSR. Whilst the senior midfield group outperformed the U20s. One potential reason
for these discrepancies is that the two groups used here (senior and U20) are not elite and non-
Movement Demands of Elite Under-20s and Senior International Rugby Union Players
PLOS ONE | DOI:10.1371/journal.pone.0164990
November 8, 2016
9 / 13
elite as used in studies where HSR has been a distinguishing factor. Both groups could be
viewed as ‘elite’, in support of this 8 players from the U20s cohort have already progressed to
the senior squads. It is also worth noting U20s players generally weigh less than their senior
counterparts so will need to be able to maintain the same movement work load (e.g. distance
covered, HSR distance, accelerations, decelerations) whilst increasing in mass to prepare for
senior internationals. To our knowledge this is the first study comparing movement demands
of U20 and senior International rugby union matches. The data suggests that the movement
demands in Under 20s internationals are adequate for preparing players for movement
demands reported in International rugby. However, certain positional groups might require
more work and/or time to match their senior counterparts than others. Conditioning staff
must also bear in mind that the U20s players whilst maintaining or improving match move-
ment capabilities may require to gain substantial mass in some positions to match their senior
counterparts.
Supporting Information
S1 File. Table A. Supplementary Movement Characteristicsfor Senior and U20s, Forwards
and Backs Groups. Data presented as Mean ± S.D Table B. SupplementaryMovement Char-
acteristics Presented by Playing Groups. FR = Front Row (Prop & Hooker), SR = Second
Row, BR = Back Row, HB = Half Backs, MF = Midfield/Centres,B3 = Back Three (Wing &
Fullback). Data presented as Mean ± S.D. Table C. Estimates of fixed effects for all GPS vari-
ables displaying difference for position type between U20s and Seniors Table D. Estimates
of fixed effects for all GPS variables displaying difference for position group between U20s
and Seniors
(DOCX)
Author Contributions
Conceptualization: DC LK SD RE CC DS BP NT.
Data curation: DC LK SD RE CC DS BP NT.
Formal analysis: DC LK SD RE BP.
Funding acquisition: SD LK.
Methodology:DC LK SD RE CC DS BP NT.
Project administration: DC LK SD RE CC DS BP NT.
Supervision:LK.
Writing – original draft: DC LK SD RE CC DS BP NT.
Writing – review& editing: DC LK DS.
References
1.
Cahill N, Lamb K, Worsfold P, Headey R, Murray S. The movement characteristics of English Premier-
ship rugby union players. J Sports Sci. 2013; 31(3):229–37. doi: 10.1080/02640414.2012.727456
PMID: 23009129.
2.
Roberts SP, Trewartha G, Higgitt RJ, El-Abd J, Stokes KA. The physical demands of elite English
rugby union. J Sports Sci. 2008; 26(8):825–33. doi: 10.1080/02640410801942122 PMID: 18569548.
3.
Lindsay A, Draper N, Lewis J, Gieseg SP, Gill N. Positional demands of professional rugby. Eur J
Sport Sci. 2015:1–8. doi: 10.1080/17461391.2015.1025858 PMID: 25830235.
Movement Demands of Elite Under-20s and Senior International Rugby Union Players
PLOS ONE | DOI:10.1371/journal.pone.0164990
November 8, 2016
10 / 13
4.
Jones MR, West DJ, Crewther BT, Cook CJ, Kilduff LP. Quantifying positional and temporal movement
patterns in professional rugby union using global positioning system. Eur J Sport Sci. 2015:1–9. doi:
10.1080/17461391.2015.1010106 PMID: 25675258.
5.
Jones MR, West DJ, Harrington BJ, Cook CJ, Bracken RM, Shearer DA, et al. Match play performance
characteristics that predict post-match creatine kinase responses in professional rugby union players.
BMC Sports Sci Med Rehabil. 2014; 6(1):38. doi: 10.1186/2052-1847-6-38 PMID: 25419462; PubMed
Central PMCID: PMCPMC4240886.
6.
Cunniffe B, Proctor W, Baker JS, Davies B. An evaluation of the physiological demands of elite rugby
union using Global Positioning System tracking software. J Strength Cond Res. 2009; 23(4):1195–
203. doi: 10.1519/JSC.0b013e3181a3928b PMID: 19528840.
7.
Reardon C, Tobin DP, Delahunt E. Application of Individualized Speed Thresholds to Interpret Position
Specific Running Demands in Elite Professional Rugby Union: A GPS Study. PLOS ONE. 2015; 10(7):
e0133410. doi: 10.1371/journal.pone.0133410 PMID: 26208315; PubMed Central PMCID:
PMCPMC4514747.
8.
Lacome M, Piscione J, Hager JP, Bourdin M. A new approach to quantifying physical demand in rugby
union. J Sports Sci. 2014; 32(3):290–300. doi: 10.1080/02640414.2013.823225 PMID: 24016296.
9.
McLean DA. Analysis of the physical demands of international rugby union. J Sports Sci. 1992; 10
(3):285–96. doi: 10.1080/02640419208729927 PMID: 1602530.
10.
Cunniffe B, Hore AJ, Whitcombe DM, Jones KP, Baker JS, Davies B. Time course of changes in immu-
neoendocrine markers following an international rugby game. Eur J Appl Physiol. 2010; 108(1):113–
22. doi: 10.1007/s00421-009-1200-9 PMID: 19756700.
11.
Cunniffe B, Hore AJ, Whitcombe DM, Jones KP, Davies B, Baker JS. Immunoendocrine responses
over a three week international rugby union series. J Sports Med Phys Fitness. 2011; 51(2):329–38.
PMID: 21681170.
12.
Quarrie KL, Hopkins WG, Anthony MJ, Gill ND. Positional demands of international rugby union: evalu-
ation of player actions and movements. J Sci Med Sport. 2013; 16(4):353–9. doi: 10.1016/j.jsams.
2012.08.005 PMID: 22975233.
13.
Cunningham D, Shearer DA, Drawer S, Eager R, Taylor N, Cook C, et al. Movement Demands of Elite
U20 International Rugby Union Players. PLOS ONE. 2016; 11(4):e0153275. doi: 10.1371/journal.
pone.0153275 PMID: 27055230; PubMed Central PMCID: PMCPMC4824470.
14.
Clarke AC, Anson JM, Pyne DB. Proof of concept of automated collision detection technology in rugby
sevens. J Strength Cond Res. 2016. doi: 10.1519/JSC.0000000000001576 PMID: 27467515.
15.
Deutsch MU, Kearney GA, Rehrer NJ. Time—motion analysis of professional rugby union players dur-
ing match-play. J Sports Sci. 2007; 25(4):461–72. doi: 10.1080/02640410600631298 PMID:
17365533.
16.
Eaton C, George K. Position specific rehabilitation for rugby union players. Part I: Empirical movement
analysis data. Physical Therapy in Sport. 2006; 7(1):22–9. doi: 10.1016/j.ptsp.2005.08.006 PMID:
WOS:000236248400004.
17.
Duthie G, Pyne D, Hooper S. Time motion analysis of 2001 and 2002 super 12 rugby. J Sports Sci.
2005; 23(5):523–30. doi: 10.1080/02640410410001730188 PMID: 16195000.
18.
Lombard WP, Durandt JJ, Masimla H, Green M, Lambert MI. Changes in body size and physical char-
acteristics of South African under-20 rugby union players over a 13-year period. J Strength Cond Res.
2015; 29(4):980–8. doi: 10.1519/JSC.0000000000000724 PMID: 25387267.
19.
Barr MJ, Sheppard JM, Gabbett TJ, Newton RU. Long-term training-induced changes in sprinting
speed and sprint momentum in elite rugby union players. J Strength Cond Res. 2014; 28(10):2724–31.
doi: 10.1519/JSC.0000000000000364 PMID: 24402451.
20.
Gabbett TJ. The training-injury prevention paradox: should athletes be training smarter and harder? Br
J Sports Med. 2016; 50(5):273–80. doi: 10.1136/bjsports-2015-095788 PMID: 26758673; PubMed
Central PMCID: PMCPMC4789704.
21.
Olds T. The evolution of physique in male rugby union players in the twentieth century. J Sports Sci.
2001; 19(4):253–62. doi: 10.1080/026404101750158312 PMID: 11311023.
22.
Argus CK, Gill ND, Keogh JW. Characterization of the differences in strength and power between dif-
ferent levels of competition in rugby union athletes. J Strength Cond Res. 2012; 26(10):2698–704. doi:
10.1519/JSC.0b013e318241382a PMID: 22105055.
23.
Hansen KT, Cronin JB, Pickering SL, Douglas L. Do force-time and power-time measures in a loaded
jump squat differentiate between speed performance and playing level in elite and elite junior rugby
union players? J Strength Cond Res. 2011; 25(9):2382–91. doi: 10.1519/JSC.0b013e318201bf48
PMID: 21804430.
Movement Demands of Elite Under-20s and Senior International Rugby Union Players
PLOS ONE | DOI:10.1371/journal.pone.0164990
November 8, 2016
11 / 13
24.
McLellan CP, Lovell DI. Performance analysis of professional, semiprofessional, and junior elite rugby
league match-play using global positioning systems. J Strength Cond Res. 2013; 27(12):3266–74. doi:
10.1519/JSC.0b013e31828f1d74 PMID: 23478474.
25.
Gabbett TJ. Physiological characteristics of junior and senior rugby league players. Br J Sports Med.
2002; 36(5):334–9. PMID: 12351330; PubMed Central PMCID: PMCPMC1724544. doi: 10.1136/
bjsm.36.5.334
26.
Darrall-Jones JD, Jones B, Till K. Anthropometric and Physical Profiles of English Academy Rugby
Union Players. J Strength Cond Res. 2015; 29(8):2086–96. doi: 10.1519/JSC.0000000000000872
PMID: 25647656.
27.
Gannon EA, Stokes KA, Trewartha G. Strength and Power Development in Professional Rugby Union
Players Over a Training and Playing Season. Int J Sports Physiol Perform. 2016; 11(3):381–7. doi: 10.
1123/ijspp.2015-0337 PMID: 26307851.
28.
Higham DG, Pyne DB, Anson JM, Eddy A. Movement patterns in rugby sevens: effects of tournament
level, fatigue and substitute players. J Sci Med Sport. 2012; 15(3):277–82. doi: 10.1016/j.jsams.2011.
11.256 PMID: 22188846.
29.
McLellan CP, Lovell DI, Gass GC. Performance analysis of elite Rugby League match play using
global positioning systems. J Strength Cond Res. 2011; 25(6):1703–10. doi: 10.1519/JSC.
0b013e3181ddf678 PMID: 21358424.
30.
Harley JA, Barnes CA, Portas M, Lovell R, Barrett S, Paul D, et al. Motion analysis of match-play in
elite U12 to U16 age-group soccer players. J Sports Sci. 2010; 28(13):1391–7. doi: 10.1080/
02640414.2010.510142 PMID: 20967674.
31.
Castellano J, Casamichana D, Calleja-Gonza´lez J, Roma´n JS, Ostojic SM. Reliability and Accuracy of
10 Hz GPS Devices for Short-Distance Exercise. J Sports Sci Med. 2011; 10(1):233–4. PMID:
24137056; PubMed Central PMCID: PMCPMC3737891.
32.
Varley MC, Fairweather IH, Aughey RJ. Validity and reliability of GPS for measuring instantaneous
velocity during acceleration, deceleration, and constant motion. J Sports Sci. 2012; 30(2):121–7. doi:
10.1080/02640414.2011.627941 PMID: 22122431.
33.
Batterham AM, Hopkins WG. Making meaningful inferences about magnitudes. Int J Sports Physiol
Perform. 2006; 1(1):50–7. PMID: 19114737.
34.
Anderson L, Orme P, Di Michele R, Close GL, Milsom J, Morgans R, et al. Quantification of Seasonal
Long Physical Load in Soccer Players With Different Starting Status From the English Premier League:
Implications for Maintaining Squad Physical Fitness. Int J Sports Physiol Perform. 2016. doi: 10.1123/
ijspp.2015-0672 PMID: 26915393.
35.
Anderson L, Orme P, Di Michele R, Close GL, Morgans R, Drust B, et al. Quantification of training load
during one-, two- and three-game week schedules in professional soccer players from the English Pre-
mier League: implications for carbohydrate periodisation. J Sports Sci. 2016; 34(13):1250–9. doi: 10.
1080/02640414.2015.1106574 PMID: 26536538.
36.
Tierney P, Young A, Clarke N, Duncan M. Match play demands of 11 versus 11 professional football
using Global Positioning System tracking: Variations across common playing formations. Human
Movement Science: Elsevier; 2016. p. 1–8.
37.
Russell M, Sparkes W, Northeast J, Kilduff LP. Responses to a 120 min reserve team soccer match: a
case study focusing on the demands of extra time. J Sports Sci. 2015; 33(20):2133–9. doi: 10.1080/
02640414.2015.1064153 PMID: 26148212.
38.
Russell M, Sparkes W, Northeast J, Cook CJ, Love TD, Bracken RM, et al. Changes in Acceleration
and Deceleration Capacity Throughout Professional Soccer Match-Play. J Strength Cond Res. 2016;
30(10):2839–44. doi: 10.1519/JSC.0000000000000805 PMID: 25474342.
39.
Russell M, Sparkes W, Northeast J, Cook CJ, Bracken RM, Kilduff LP. Relationships between match
activities and peak power output and Creatine Kinase responses to professional reserve team soccer
match-play. Hum Mov Sci. 2016; 45:96–101. doi: 10.1016/j.humov.2015.11.011 PMID: 26615476.
40.
Austin DJ, Kelly SJ. Positional differences in professional rugby league match play through the use of
global positioning systems. J Strength Cond Res. 2013; 27(1):14–9. doi: 10.1519/JSC.
0b013e31824e108c PMID: 22344046.
41.
Gabbett TJ. Influence of the opposing team on the physical demands of elite rugby league match play.
J Strength Cond Res. 2013; 27(6):1629–35. doi: 10.1519/JSC.0b013e318274f30e PMID: 23037616.
42.
Jones NM, James N, Mellalieu SD. An objective method for depicting team performance in elite profes-
sional rugby union. J Sports Sci. 2008; 26(7):691–700. doi: 10.1080/02640410701815170 PMID:
18409100.
Movement Demands of Elite Under-20s and Senior International Rugby Union Players
PLOS ONE | DOI:10.1371/journal.pone.0164990
November 8, 2016
12 / 13
43.
Owen SM, Venter RE, du Toit S, Kraak WJ. Acceleratory match-play demands of a Super Rugby team
over a competitive season. J Sports Sci. 2015; 33(19):2061–9. doi: 10.1080/02640414.2015.1028086
PMID: 25846204.
44.
Preatoni E, Stokes KA, England ME, Trewartha G. The influence of playing level on the biomechanical
demands experienced by rugby union forwards during machine scrummaging. Scand J Med Sci
Sports. 2013; 23(3):e178–84. doi: 10.1111/sms.12048 PMID: 23362799.
45.
Mohr M, Krustrup P, Bangsbo J. Match performance of high-standard soccer players with special refer-
ence to development of fatigue. J Sports Sci. 2003; 21(7):519–28. doi: 10.1080/
0264041031000071182 PMID: 12848386.
46.
Sirotic AC, Coutts AJ, Knowles H, Catterick C. A comparison of match demands between elite and
semi-elite rugby league competition. J Sports Sci. 2009; 27(3):203–11. doi: 10.1080/
02640410802520802 PMID: 19153858.
47.
Jennings DH, Cormack SJ, Coutts AJ, Aughey RJ. International field hockey players perform more
high-speed running than national-level counterparts. J Strength Cond Res. 2012; 26(4):947–52. doi:
10.1519/JSC.0b013e31822e5913 PMID: 22446668.
Movement Demands of Elite Under-20s and Senior International Rugby Union Players
PLOS ONE | DOI:10.1371/journal.pone.0164990
November 8, 2016
13 / 13
| Movement Demands of Elite Under-20s and Senior International Rugby Union Players. | 11-08-2016 | Cunningham, Daniel J,Shearer, David A,Drawer, Scott,Pollard, Ben,Eager, Robin,Taylor, Neil,Cook, Christian J,Kilduff, Liam P | eng |
PMC4919094 | Supplement: Prediction and Quantification of Individual Athletic
Performance of Runners
Duncan A.J. Blythe ∗ 1,2 and Franz J. Király † 3
1African Institute for Mathematical Sciences, Bagamoyo, Tanzania
2 Bernstein Center for Computational Neuroscience, Berlin, Germany
3 Department of Statistical Science, University College London, United Kingdom
Methods
The following provides a guideline for reproducing the results. Raw and pre-processed data in MATLAB and
CSV formats is available upon request, subject to approval by British Athletics. Complete and documented
source code of algorithms and analyses as well as data can be obtained from [1, 2].
Data Source
The basis for our analyses is the online database www.thepowerof10.info, which catalogues British individ-
uals’ performances achieved in officially ratified athletics competitions since 1954, including Olympic athletic
events (field and non-field events), non-Olympic athletic events, cross country events and road races of all
distances.
With the permission of British Athletics, we obtained an excerpt of the database by automated querying
of the freely accessible parts of www.thepowerof10.info, restricted to ten types of running events: 100m,
200m, 400m, 800m, 1500m, the Mile, 5000m (track and road races), 10000m (track and road races), Half-
Marathon and Marathon.
Other types of running events were available but excluded from the present
analyses; the reasons for exclusion were a smaller total of attempts (e.g. 3000m), a different population of
runners (e.g. 3000m is mainly attempted by younger runners), and varying conditions (steeplechase/ hurdles
and cross-country races).
The data set consists of two tables: athletes.csv, containing records of individual runners, with fields:
runner ID, gender, date of birth; and events.csv, containing records of individual attempts on running
events until August 3, 2013, with fields: runner ID, event type, date of the attempt, and performance in
seconds.
Data Cleaning
Our excerpt of the database contains (after error and duplication removal) records of 164,746 individuals
of both genders, ranging from the amateur to the elite, young to old, and a total of 1,410,789 individual
performances for 10 different types of events (see previous section).
Gender is available for all runners in the database (101,775 male, 62,971 female). The dates of birth
of 114,168 runners are missing (recorded as January 1, 1900 in athletes.csv due to particulars of the
automated querying); the date of birth of six runners is set to missing due to a recorded age at recorded
attempts of eight years or less.
For the above runners, a total of 1,417,476 attempts was recorded, out of which 1,410,789 remained in
the data set after cleaning: 192,947 over 100m, 194,107 over 200m, 109,430 over 400m, 239,666 over 800m,
∗[email protected]
†[email protected]
1
176,284 over 1500m, 6,590 at the Mile distance, 96,793 over 5000m (the track and road races), 161,504 over
10000m (on the track and road races), 140,446 for the Half-Marathon and 93,033 for the Marathon. 6,643
duplicate events were removed, and a number of 44 events whose reported performances are better than the
official world records of their time, or extremely slow. Dates of the attempt were set to missing for 225 of
the attempts that recorded January 1, 1901, and one of the attempts that recorded August 20, 2038.
Data Preprocessing
The events and athletes data sets are collated into (10×164, 746)-tables/matrices of performances, where the
10 columns correspond to events and the 164, 746 rows to individual runners. Rows are indexed increasingly
by runner ID, columns by the type of event. Each entry of the table/matrix contains one performance (in
seconds) of the runner by which the row is indexed, at the event by which the column is indexed, or a missing
value. If the entry contains a performance, the date of that performance is stored as meta-information.
We consider two different modes of collation, yielding one table/matrix of performances of size (10 ×
164, 746) each.
In the first mode, which in Tables 1 ff. is referenced as “best”, one proceeds as follows. First, for each
individual runner, one finds the best event of each individual, measured by population percentile. Then, for
each type of event which was attempted by that runner within a year before or after that best event, the
best performance for that type of event is entered into the table. If a certain event was not attempted in
this period, it is recorded as missing.
For the second mode of collation, which in Tables 1 ff. is referenced as “random”, one proceeds as follows.
First, for each individual runner, a calendar year is uniformly randomly selected among the calendar years
in which that runner has attempted at least one event. Then, for each type of event which was attempted
by that runner within the selected calendar year, the best performance for that type of event is entered into
the table. If a certain event was not attempted in the selected calendar year, it is recorded as missing.
The first collation mode ensures that the data is of high quality: runners are close to optimal fitness,
since their best performance was achieved in this time period. Moreover, since fitness was at a high level, it
is plausible that the number of injuries incurred was low leading to multiple attempts being made at events
– this will lead to higher data quality; indeed it can be observed that the number of attempts per event is
higher in this period, effectively decreasing the influence of noise and the chance that outliers are present
after collation.
The second collation mode is used to check whether and, if so how strongly, the results depend on the
runners being close to optimal fitness.
In both cases choosing a narrow time frame ensures that performances are relevant to one another for
prediction.
Runner-Specific Summary Statistics
For each given runner, several summaries are computed based on the collated matrix.
Performance percentiles are computed for each event which a runner attempts in relation to the other
runners’ performances on the same event. These column-wise event-specific percentiles, yield a percentile
matrix with the same filling pattern (pattern of missing entries) as the collated matrix.
The preferred distance for a given runner is the geometric mean of the attempted events’ distances.
That is, if s1, . . . , sm are the distances for the events which the runner has attempted, then ˜s = (s1 · s2 · . . . ·
sm)1/m is the preferred distance.
The training standard for a given runner is the mean of all performance percentiles in the corresponding
row.
The no. events for a given runner is the number of events attempted by a runner in the time period of
the data considered (best or random).
Note that the percentiles yield a mostly physiological description; the preferred distance is a behavioural
summary since it describes the type of events the runner attempts. The training standard combines both
physiological and behavioural characteristics.
Percentiles and training standard depend on the collated matrix. When we consider genders, age or
runners who have attempted more than a certain no.event these summary statistics are calculated separately
2
for the subgroup. However, within subgroup these values depend on the entire collated submatrix for that
subgroup.
Outlier Removal
Outliers are removed from the data in both collated matrices.
An outlier score for each runner/row is
obtained as the difference of maximum and minimum of all performance percentile of the runner. The five
percent rows/runners with the highest outlier score are removed from the matrix.
Prediction: Evaluation and Validation
Prediction accuracy is evaluated on row-sub-samples of the collated matrices, defined by (a) a potential
subgroup, e.g., given by age or gender, (b) degrees-of-freedom constraints in the prediction methods that
require a certain number of entries per row, and (c) a certain range of performance percentiles of runners.
The row-sub-samples referred to in the main text and in Tables 1 ff. are obtained by (a) retaining all
rows/runners in the subgroup specified by gender, or age in the best event, (b) retaining all rows/runners
with at least no. events or more entries non-missing, and discarding all rows/runners with strictly less
than no. events entries non-missing, then (c) retaining all runners in a certain percentile range.
The
percentiles referred to in (c) are computed as follows: first, for each column, in the data retained after step
(b), percentiles are computed. Then, for each row/runner, the best of these percentiles is selected as the
score over which the overall percentiles are taken.
The accuracy of prediction is principally measured empirically in terms of out-of-sample root mean
squared error (RMSE) and mean absolute error (MAE), with RMSE, MAE, and standard deviations es-
timated from the empirical sample of residuals obtained in 1000 iterations of leave-one-out validation. In
selected analyses we measure error in terms of relative RMSE (rRMSE),
1
N
P
i
predictor(i)−predicted(i)
predicted(i)
2
and
relative MAE (defined analogously) (rMAE).
Given the row-sub-sample matrix obtained from (a), (b), (c), prediction and thus we perform leave-one-
out validation in two ways: (i) predicting the left-out entry from potentially all remaining entries. In this
scenario, the prediction method may have access to the performances of the runner in question which lie in
the future of the event to be predicted, though only performances of other events are available; (ii) predicting
the left-out entry from all remaining entries of other runners, but only from those events of the runner in
question that lie in the past of the event to be predicted. In this task, temporal causality is preserved on
the level of the single runner for whom prediction is done; though information about other runners’ results
that lie in the future of the event to be predicted may be used.
The third option (iii) where predictions are made only from past events has not been studied due to
the size of the data set which makes collation of the data set for every single prediction per method and
group a computationally extensive task, and due to the potential group-wise sampling bias which would be
introduced, skewing the measures of prediction-quality—the population of runners on the older attempts is
different in many respects from the more recent attempts. We further argue that in the absence of such
technical issues, evaluation as in (ii) would be equivalent to (iii); since the performances of two randomly
picked runners, no matter how they are related temporally, may be reasonably modelled as statistically
independent; positing the contrary would be equivalent to postulating that any given runner’s performance
is very likely to be directly influenced by a large number of other runners’ performance history, which is an
assumption which is scientifically implausible. Given the above, due to equivalence of (ii) and (iii), and the
issues occurring in (iii) exclusively, we can conclude that (ii) is preferrable over (iii) from a scientific and
statistical viewpoint.
Prediction: Target Outcomes
The principal target outcome for the prediction is “performance”, which we present to the prediction methods
in three distinct parameterisations. This corresponds to passing not the raw performance matrices obtained
in the section “Data Pre-processing” to the prediction methods, but re-parameterized variants where the
non-missing entries undergo a univariate variable transform. The three parameterizations of performance
considered in our experiments are the following:
3
(a) normalized: performance as the time in which the given runner (row) completes the event in question
(column), divided by the average time in which the event in question (column) is completed in the sub-
sample;
(b) log-time: performance as the natural logarithm of time in seconds in which the given runner (row)
completes the event in question (column);
(c) speed: performance as the average speed in meters per second, with which the given runner (row)
completes the event in question (column).
The words in italics indicate which parameterisation is referred to in Table 1. The error measures, RMSE
and MAE, are evaluated in the same parameterisation in which prediction is performed. We do not evaluate
performance directly in un-normalized time units, as in this representation performances between 100m and
the Marathon span 4 orders of magnitude (base-10), which would skew the measures of goodness heavily
towards accuracy over the Marathon.
Unless stated otherwise, predictions are made in the same parameterisation on which the models are
learnt.
Prediction: Models and Algorithms
In the experiments, a variety of prediction methods are used to perform prediction from the performance
data, given as described in “Prediction: Target Outcomes”, evaluated by the measures as described in the
section “Prediction: Evaluation and Validation”.
In the code available for download, each method is encapsulated as a routine which predicts a missing entry
when given the (training entries in the) performance matrix. The methods can be roughly divided in four
classes: (1) naive baselines, (2) representatives of the state-of-the-art in prediction of running performance,
(3) representatives of the state-of-the-art in matrix completion, and (4) our proposed method and its variants.
The naive baselines are:
(1.a) mean: predicting the the mean over all performances for the same event within the subgroup considered.
(1.b) k-NN: k-nearest neighbours prediction. The parameter k is obtained as the minimizer of out-of-sample
RMSE on five groups of 50 randomly chosen validation data points from the training set, from among
k = 1, k = 5, and k = 20.
The representatives of the state-of-the-art in predicting running performance are:
(2.a) Riegel: The Riegel power law formula with exponent 1.06.
(2.b) power law: A power law
predictor, as per the Riegel formula, but with the exponent estimated from the data. The exponent is the
same for all runners and estimated as the minimizer of the residual sum of squares. (2.c) ind.power law:
A power law predictor, as per the Riegel formula, but with the exponent estimated from the data. The
exponent may be different for each runner and is estimated as the minimizer of the residual sum of squares.
(2.d) Purdy: Prediction by calculation of equivalent performances using the Purdy points scheme [3]. Purdy
points are calculated by using the measurements given by the Portugese scoring tables which estimate the
maximum velocity for a given distance in a straight line, and adjust for the cost of traversing curves and
the time required to reach race velocity. The performance with the same number of points as the predicting
event is imputed.
The representatives of the state-of-the-art in matrix completion are:
(3.a) EM: Expectation maximization algorithm assuming a multivariate Gaussian model for the rows
of the performance matrix in log-time parameterisation.
Missing entries are initialized by the mean of
each column. The updates are terminated when the percent increase in log-likelihood is less than 0.1%.
For a review of the EM-algorithm see [4].
(3.b) Nuclear Norm: Matrix completion via nuclear norm
minimization [5, 6].
The variants of our proposed method are as follows:
(4.a-d) LMC rank r: local matrix completion for the low-rank model, with rank r = 1, 2, 3, 4. (4.a) is
LMC rank 1, (4.b) is LMC rank 2, and so on.
Our algorithm follows the local/entry-wise matrix completion paradigm in [7]. It extends the rank 1 local
matrix completion method described in [8] to arbitrary ranks.
Our implementation uses: determinants of size (r + 1 × r + 1) as the only circuits; the weighted variance
minimization principle in [8]; the linear approximation for the circuit variance outlined in the appendix of [9];
modelling circuits as independent for the co-variance approximation.
4
We further restrict to circuits supported on the event to be predicted and the r log-distance closest events.
For the convenience of the reader, we describe the exact way in which the local matrix completion
principle is instantiated, in the section “Prediction: Local Matrix Completion” below
In the supplementary experiments we also investigate two aggregate predictors to study the potential
benefit of using other lengths for prediction:
(5.a) bagged power law: bagging the power law predictor with estimated coefficient (2.b) by a weighted
average of predictions obtained from different events. The weighting procedure is described below. (5.b)
bagged LMC rank 2: estimation by LMC rank 2 where determinants can be supported at any three events,
not only on the closest ones (as in line 1 of Algorithm 1 below). The final, bagged predictor is obtained as a
weighted average of LMC rank 2 running on different triples of events. The weighting procedure is described
below.
The averaging weights for (5.a) and (5.b) are both obtained from the Gaussian radial basis function
kernel exp
Obtaining the Low-Rank Components and Coefficients
We obtain three low-rank components f1, . . . , f3 and corresponding coefficients λ1, . . . , λ3 for each runner
by considering the data in log-time coordinates. Each component fi is a vector of length 10, with entries
corresponding to events. Each coefficient is a scalar, potentially different for each runner.
To obtain the components and coefficients, we consider the data matrix for the specific target outcome,
sub-sampled to contain the runners who have attempted four or more events and the top 25% percentiles,
as described in “Prediction: Evaluation and Validation”. In this data matrix, all missing values are imputed
using the rank 3 local matrix completion algorithm, as described in (4.c) of “Prediction: Models and Algo-
rithms”, to obtain a complete data matrix M. For this matrix, the singular value decomposition M = USV ⊤
is computed, see [10].
We take the components f2, f3 to be the the 2-th and 3-rd right singular vectors, which are the 2-nd and
3-rd column of V . The component f1 is a re-scaled version of the 1-st column v of V , such that f1(s) ≈ log s,
where the natural logarithm is taken. More precisely, f1 := β−1v, where the re-scaling factor β is obtained
as the ordinary least-squares regression coefficient of the linear explanatory model v(s) = β log s + c, where
s ranges over the ten event distances, which is β = 0.0572. A more detailed study of v and the regression
coefficient can be found in supplementary experiment (II.b).
The three-number-summary referenced in the main corpus of the manuscript is obtained as follows: for
the k-th runner we obtain from the left singular vector the entries Ukj. The second and third score of the
three-number-summary are obtained as λ2 = Uk2 and λ3 = Uk3. The individual exponent is λ1 = β · Uj1.
The singular value decomposition has the property that the fi and λj are guaranteed to be least-squares
estimators for the components and the coefficients in a projection sense.
Computation of standard error and significance
Standard errors for the singular vectors (components of the model of Equation 1) are computed via inde-
pendent bootstrap sub-sampling on the rows of the data set (runners).
Standard errors for prediction accuracies are obtained by bootstrapping of the predicted performances
(1000 per experiment). A method is considered to perform significantly better than another when error
regions at the 95% confidence level (= mean over repetitions ± 1.96 standard errors) do not intersect.
Predictions and three-number-summary for elite runners
Performance predictions and three-number-summaries for the selected elite runners in Table 1 and Figure 4
are obtained from their personal best times. The relative standard error of the predicted performances is
estimated to be the same as the relative RMSE of predicting time, as reported in Table 1.
Calculating a fair race
Here we describe the procedure for calculating a fair racing distance with error bars between two runners:
runner 1 and runner 2. We first calculate predictions for all events. Provided that runner 1 is quicker on
some events and runner 2 is quicker on others, then calculating a fair race is feasible. If runner 1 is quicker on
shorter events then runner 2 is typically quicker on all longer events beyond a certain distance. In that case,
we can find the shortest race si whereby runner 2 is predicted to be quicker; then a fair race lies between si
and si−1. The performance curves in log-time vs. log-distance of both runners will be locally approximately
linear. We thus interpolate the performance curves between log(si) and log(si−1)—the crossing point gives
the position of a fair race in log-coordinates. We obtain confidence intervals by repeating this procedure
after sampling data points around the estimated performances with standard deviation equal to the RMSE
(see Table 1) on the top 25% of runners in log-time.
6
Supplementary Analyses
This appendix contains a series of additional experiments supplementing those in the main corpus. It con-
tains the following findings:
(I) Validation of the LMC prediction framework.
(I.a) Evaluation in terms of MAE. The results in terms of MAE are qualitatively similar to those in
RMSE; smaller MAEs indicate the presence of outliers.
(I.b) Evaluation in terms of time prediction. The results are qualitatively similar to measuring pre-
diction accuracy in RMSE and MAE of log-time. LMC rank 2 has an average error of approximately 2%
when predicting the top 25% of male runners.
(I.c) Prediction for individual events. LMC outperforms the other predictors on each type of event.
The benefit of higher rank is greatest for middle distances.
(I.d) Stability w.r.t. the unit measuring performance. LMC performs equally well in predicting (per-
formance in time units) when performances are presented in log-time or time normalized by event average.
Speed is worse when the rank 2 predictor is used.
(I.e) Stability w.r.t. the events used in prediction. LMC performs equally well when predicting from
the closest-distance events and when using a bagged version which uses all observed events for prediction.
(I.f) Stability w.r.t. the event predicted. LMC performs well both when the predicted event is close
to those observed and when the predicted event is further from those observed, in terms of event distance.
(I.g) Temporal independence of performances. There are negligible differences between predictions
made only from past events and predictions made from all available events (in the training set).
(I.h) Run-time comparisons. LMC is by orders of magnitude the fastest among the matrix completion
methods.
(II) Validation of the low-rank model.
(II.a) Synthetic validation. In a synthetic low-rank model of athletic performance that is a proxy to the
real data, the singular components of the model can be correctly recovered by the exact same procedure as
on the real data.
(II.b) The individual power law component, and the distance/time unit. The first singular com-
ponent can be explained by a linear model in log-distance (R-square 0.9997) with slope β = 0.0572 ± 0.0003
and intercept c = −0.136 ± 0.003.
(II.c) Universality in sub-groups. Quality of prediction, the low-rank model, its rank, and the singular
components remain mostly unchanged when considering subgroups male/female, older/younger, elite/amateur.
(III) Exploration of the low-rank model.
(III.a) Further exploration of the three-number-summary. The three number summary also corre-
lates with specialization and training standard.
(III.b) Preferred distance vs optimal distance. Most but not all runners prefer to attend the event
at which they are predicted to perform best. A notable number of younger runners prefer distances shorter
than optimal, and some older runners prefer distances longer than optimal.
(IV) Pivoting and phase transitions. The pivoting phenomenon described in Figure 1, right panel,
is found in the data for any three close-by distances up to the Mile, with anti-correlation between the shorter
and the longer distance. Above 5000m, a change in the shorter of the three distances positively correlates
with a change in the longer distance.
(I.a) Evaluation in terms of MAE. Table A reports on the goodness of prediction methods in terms
of MAE. Compared with the RMSE (Table 1, the MAE tend to be smaller than the RMSE, indicating the
presence of outliers. The relative prediction-accuracy of methods when compared to each other is qualita-
tively the same.
(I.b) Evaluation in terms of time prediction. Tables C and D report on the prediction accuracy of
the methods tested in terms the relative RMSE and MAE of predicting time. Relative measures are chosen
7
to avoid bias towards the longer events. The results are qualitatively and quantitatively very similar to the
log-time results in Tables 1 and A; this can be explained that mathematically the RMSE and MAE of a
logarithm approximate the relative RMSE and MAE well for small values.
(I.c) Individual Events. Prediction accuracy of LMC rank 1 and rank 2 on the ten different events is
displayed in Figure A. The reported prediction accuracy is out-of-sample RMSE in predicting log-time, on
the top 25 percentiles of Male runners who have attempted 3 or more events, of events in their best year
of performance. The reported RMSE for a given event is the mean over 1000 random prediction samples,
standard errors are estimated by the bootstrap.
The relative improvement of rank 2 over rank 1 tends to be greater for shorter distances below the Mile. This
is in accordance with observation (IV.i) which indicates that the individual exponent is the best descriptor
among the three-number summary for longer events, above the Mile.
(I.d) Stability w.r.t. the measure of performance. In the main experiment, the LMC model is learnt on
the same measure of performance (log-time, speed, normalized) which is predicted. We investigate whether
the measure of performance on which the model is learnt influences the prediction by learning the LMC model
on either measure and comparing all predictions using the log-time measure. Table G displays prediction
accuracy when the model is learnt in any one of the measures of performance. Here we check the effect of
calibration in one coordinates system and testing in another. The reported goodness is out-of-sample RMSE
of predicting log-time, on the top 25 percentiles of Male runners who have attempted 3 or more events,
of events in their best year of performance. The reported RMSE for a given event is the mean over 1000
random prediction samples, standard errors are estimated by the bootstrap.
We find that there is no significant difference in prediction goodness when learning the model in log-time
coordinates or normalized time coordinates. Learning the model in speed coordinates leads to a significantly
better prediction than log-time or normalized time when LMC rank 1 is applied, but to a worse prediction
with LMC rank 2. As overall prediction with LMC rank 2 is better, log-time or normalized time are the
preferable units for predicting performance.
(I.e) Stability w.r.t. the event predicted.
We consider here the effect of the ratio between the predicted event and the closest predictor. For data
of the best 25% of Males in the year of best performance (best), we compute the log-ratio of the closest
predicting distance and the predicted distance for Purdy Points, the power law formula and LMC rank 2.
See Figure B, where this log ratio is plotted against error. The results show that LMC is far more robust to
error for predicting distances far from the predicted distance.
(I.f) Stability w.r.t. the events used in prediction. We compare whether we can improve predic-
tion by using all events a runner has attempted, by using one of the aggregate predictors (5.a) bagged
power law or (5.b) bagged LMC rank 2. The kernel width γ for the aggregate predictors is chosen from
−0.001, −0.01, −0.1, −1, −10 as the minimizer of out-of-sample RMSE on five groups of 50 randomly chosen
validation data points from the training set. The validation setting is the same as in the main prediction
experiment.
Results are displayed in Table H. We find that prediction accuracy of (2.b) power law and (5.a) bagged
power law is not significantly different, nor is (4.b) LMC rank 2 significantly different from (5.b) bagged
LMC rank 2 (both p > 0.05; Wilcoxon signed-rank on the absolute residuals). Even though the kernel width
selected is in the majority of cases σ = −1 and not σ = −10, the incorporation of all events does not lead to
an improvement in prediction accuracy in our aggregation scheme. We find there is no significant difference
(p > 0.05; Wilcoxon signed-rank on the absolute errors) between the bagged and vanilla LMC for the top
95% of runners. This demonstrates that the relevance of closer events for prediction may be learnt from the
data. The same holds for the bagged version of the power law formula.
(I.g) Temporal independence of performances. We check here whether the results are affected by using
only temporally prior attempts in predicting a runner’s performance, see section “Prediction: Evaluation and
Validation” in “Methods”. To this end, we compute out-of-sample RMSEs when predictions are made only
from those events.
8
Table B reports out-of-sample RMSE of predicting log-time, on the top 25 percentiles of Male runners
who have attempted 3 or more events, of events in their best year of performance. The reported RMSE
for a given event is the mean over 1000 random prediction samples, standard errors are estimated by the
bootstrap.
The results are qualitatively similar to those of Table 1 where all events are used in prediction.
(I.h) Run-time comparisons. We compare the run-time cost of a single prediction for the three matrix
completion methods LMC, nuclear norm minimiziation, and EM. The other (non-matrix completion) meth-
ods are fast or depend only negligibly on the matrix size. We measure run time of LMC rank 3 for completion
of a single entry for matrices of 28, 29, . . . , 213 runners, generated as described in (II.a). This is repeated
100 times. For a fair comparison, the nuclear norm minimization algorithm is run with a hyper-parameter
already pre-selected by cross validation. The results are displayed in Figure C; LMC is faster by orders of
magnitude than nuclear norm and EM and is very robust to the size of the matrix. The reason computation
speeds up over the smallest matrix sizes is that 4 × 4 minors, which are required for rank 3 estimation are
not available, thus the algorithm must attempt all ranks lower than 3 to find sufficiently many minors.
(II.a) Synthetic validation. To validate the assumption of a low-rank generative model, we investigate
prediction accuracy and recovery of singular vectors in a synthetic model of athletic performance.
Synthetic data for a given number of runners is generated as follows:
For each runner, a three-number summary (λ1, λ2, λ3) is generated independently from a Gaussian dis-
tribution with the same mean and variance as the three-number-summaries measured on the real data and
with uncorrelated entries.
Matrices of performances are generated from the model
log(t) = λ1f1(s) + λ2f2(s) + λ3f3(s) + η(s)
(1)
where f1, f2, f3 are the three components estimated from the real data and η(s) is a stationary zero-mean
Gaussian white noise process with adjustable variance.
We take the components estimated in log-time
coordinates from the top 25% of male runners who have attempted at least 4 events as the three components
of the model. The distances s are the same ten distances as encountered in the real data. In each experiment
the standard deviation of η(s) is set to Std(η) = 0.01, which was shown to be plausible in the previous
section.
Accuracy of prediction: We synthetically generate a matrix of 1000 runners according to the model
of Equation (1), taking as distances the same distances measured on the real data. Missing entries are
randomized according to two schemes: (a) 6 (out of 10) uniformly random missing entries per row/runner.
(b) per row/runner, four in terms of distance-consecutive entries are non-missing, uniformly at random.
We then apply LMC rank 2 and nuclear norm minimization for prediction. This setup is repeated 100
times for ten different standard deviations of η between 0.01 and 0.1. The results are displayed in Figure D.
LMC performance outperforms nuclear norm; LMC performance is also robust to the pattern of miss-
ingness, while nuclear norm minimization is negatively affected by clustering in the rows. RMSE of LMC
approaches zero with small noise variance, while RMSE of nuclear norm minimization does not.
Comparing the performances with Table 1, an assumption of a noise variance of Std(η) = 0.01 seems
plausible. The performance of nuclear norm on the real data is explained by a mix of the sampling schemes
(a) and (b).
Recovery of model components. We synthetically generate a matrix which has a size and pattern of
observed entries identical to the matrix of top 25% of male runners who have attempted at least 4 events in
their best year. We set Std(η) = 0.01, which was shown to be plausible in the previous section.
We then complete all missing entries of the matrix using LMC rank 3. After this initial step we estimate
singular components using SVD, exactly as on the real data.
Confidence intervals are estimated by a
bootstrap on the rows with 100 iterations.
The results are displayed in Figure E.
We observe that the first two singular components are recovered almost exactly, while the third is a
slightly deformed. This is due to the smaller singular value of the third component.
9
(II.b) The individual power law component, and the distance/time unit. We examine linearity of
the first singular vector v, as listed in Table 1 and as described in methods section “Obtaining the Low-Rank
Components and Coefficients”. In an ordinary least squares regression model explaining v by log s and an
intercept, we find that v ≈ β log s + c with an R-squared of 0.9997 (Table I), where the scaling factor is
β = 0.0572 ± 0.0003 and the intercept is c = −0.136 ± 0.003. The intercept corresponds to a choice of
unit, the scaling factor to a choice of basis for the logarithm. Thus re-scaling v with β−1, that is, setting
f1 := β−1v in the low-rank model, and re-scaling the first individual coefficient with β, corresponds to the
choice of the natural basis.
The residuals of the the linear model appear to be plausibly explained by the second and third singular com-
ponent (Table I), though the small number of fitting nodes which is 10 does not allow a for an assessment
that is more than qualitative.
(II.c) Universality in sub-groups. We repeat the methodology for component estimation described above
and obtain the three components in the following sub-groups: female runners, older runners (> 30 years),
and amateur runners (25-95 percentile range of training standard). Male runners were considered in the
main corpus. For female and older runners, we restrict to the top 95% percentiles of the respective groups
for estimation.
Figure F displays the estimated components of the low-rank model. The individual power law is found
to be unchanged in all groups considered. The second and third component vary between the groups but
resemble the components for the male runners. The empirical variance of the second and third component is
higher, which may be explained by a slightly reduced consistency in performance, or a reduction in sample
size. Whether there is a genuine difference in form or whether the variation is explained by different three-
number-summaries in the subgroups cannot be answered from the dataset considered.
Table F displays the prediction results in the three subgroups. Prediction accuracy is similar but slightly
worse when compared to the male runners. Again this may be explained by reduced consistency in the
subgroups’ performances.
(III.a) Further exploration of the three-number-summary. Scatter plots of preferred distance and
training standard against the runners’ three-number-summaries are displayed in Figure G. The training
standard correlates predominantly with the individual exponent (score 1); score 1 vs. standard—r = −0.89
(p ≤ 0.001); score 2 vs. standard—r = 0.22 (p ≤ 0.001); score 3 vs. standard—r = 0.031 (p = 0.07); all
correlations are Spearman correlations with significance computed using a t-distribution approximation to
the correlation coefficient under the null. On the other hand preferred distance is associated with all three
numbers in the summary, especially the second; score 1 vs. log(specialization)—r = 0.29 (p ≤ 0.001); score
2 vs. log(specialization)—r = −0.58 (p ≤ 0.001); score 3 vs. log(specialization)—r = −0.14 (p =≤ 0.001);
The association between the third score and specialization is non-linear with an optimal value around the
middle distances. We stress that low correlation does not imply low predictive power; the whole summary
should be considered as a whole, and the LMC predictor is non-linear. Also, we observe that correlations
increase when considering only performances over certain distances, see Figure 2.
(III.b) Preferred event vs best event. For the top 95% male runners who have attempted 3 or more
events, we use LMC rank 2 to compute which percentile they would achieve in each event. We then determine
the distance of the event at which they would achieve the best percentile, to which we will refer as the “optimal
distance”. Figure H shows for each runner the difference between their preferred and optimal distance.
It can be observed that the large majority of runners prefer to attempt events in the vicinity of their
optimal event. There is a group of young runners who attempt events which are shorter than the predicted
optimal distance, and a group of old runners attempting events which are longer than optimal. One may
hypothesize that both groups could be explained by social phenomena: young runners usually start to train
on shorter distances, regardless of their potential over long distances. Older runners may be biased to at-
tempting endurance type events.
(IV) Pivoting and phase transitions. We look more closely at the pivoting phenomenon illustrated in
Figure 1 top right, and the phase transition discussed in observation (V). We consider the top 25% of male
runners who have attempted at least 3 events, in their best year.
10
We compute 10 performances of equivalent standard by using LMC rank 1 in log-time coordinates,
by setting a benchmark performance over the marathon and sequentially predicting each lower distance
(marathon predicts HM, HM predicts 10km etc.). This yields equivalent benchmark performances t1, . . . , t10.
We then consider triples of consecutive distances si−1, si, si+1 (excluding the Mile since close in distance
to the 1500m) and study the pivoting behaviour on the data set, by performing the analogous prediction
displayed in Figure 1.
More specifically, for each triple, we predict the performance on the distance si+1 using LMC rank 2,
from the performances over the distances si−1 and si. The prediction is performed in two ways, once with
and once without perturbation of the benchmark performance at si−1, which we then compare. Intuitively,
this corresponds to comparing the red to the green curve in Figure 1. In mathematical terms:
1. We obtain a prediction bti+1 for the distance si+1 from the benchmark performances ti, ti−1 and consider
this as the unperturbed prediction, and
2. We obtain a prediction bti+1 + δ(ϵ) for the distance si+1 from the benchmark performance ti on si
and the perturbed performance (1 + ϵ)ti−1 on the distance si−1, considering this as the perturbed
prediction.
We record these estimates for ϵ = −0.1, 0.09, . . . , 0, 0.01, . . . , 0.1 and calculate the relative change of the
perturbed prediction with respect to the unperturbed, which is δi(ϵ)/bti. The results are displayed in Figure I.
We find that for pivot distances si shorter than 5km, a slower performance on the shorter distance si−2
leads to a faster performance over the longer distance si, insofar as this is predicted by the rank 2 predictor.
On the other hand we find that for pivot distances greater than or equal to 5km, a faster performance over
the shorter distance also implies a faster performance over the longer distance.
References
[1] Blythe DAJ, Király FJ. Full data to “Prediction and Quantification of Individual Athletic Performance of
Runners"; 2016.
DOI: 10.6084/m9.figshare.3408202.v1.
Available from: https://figshare.com/articles/
thepowerof10/3408202.
[2] Blythe DAJ, Király FJ. Full code to “Prediction and Quantification of Individual Athletic Performance of Run-
ners"; 2016.
DOI: 10.6084/m9.figshare.3408250.v1.
Available from: https://figshare.com/articles/Ful_
code_to_Prediction_and_Quantification_of_Individual_Athletic_Performance_of_Runners_/3408250.
[3] Purdy JG. Computer generated track and field scoring tables: II. Theoretical foundation and development of a
model. Medicine and science in sports. 1974;7(2):111–115.
[4] Bishop CM, et al. Pattern recognition and machine learning. vol. 4. springer New York; 2006.
[5] Candès EJ, Recht B. Exact matrix completion via convex optimization. Foundations of Computational mathe-
matics. 2009;9(6):717–772.
[6] Tomioka R, Hayashi K, Kashima H. On the extension of trace norm to tensors. In: NIPS Workshop on Tensors,
Kernels, and Machine Learning; 2010. p. 7.
[7] Király FJ, Theran L, Tomioka R. The algebraic combinatorial approach for low-rank matrix completion. Journal
of Machine Learning Research. 2015;.
[8] Király FJ, Theran L. Obtaining error-minimizing estimates and universal entry-wise error bounds for low-rank
matrix completion. NIPS 2013. 2013;.
[9] Blythe DA, Theran L, Kiraly F.
Algebraic-Combinatorial Methods for Low-Rank Matrix Completion with
Application to Athletic Performance Prediction. arXiv preprint arXiv:14062864. 2014;.
[10] Golub GH, Reinsch C.
Singular value decomposition and least squares solutions.
Numerische Mathematik.
1970;14(5):403–420.
200m
800m
Mile
10km
Mar
0
0.005
0.01
0.015
0.02
0.025
0.03
0.035
0.04
0.045
0.05
log(distance)
RMSE
Accuracy by event rank 1 vs. rank 2
rank 1
rank 2
Figure A: The figure displays the results of prediction by event for the top 25% of male runners who attended ≥ 3 events in their
year of best performance. For each event the prediction accuracy of LMC rank 1 (blue) is compared to prediction accuracy in
rank 2 (red). RMSE is displayed on the y-axis against distance on the x-axis; the error bars extend two standard deviations of the
bootstrapped RMSE either side of the RMSE.
0
1
2
3
0
0.02
0.04
0.06
0.08
0.1
0.12
Absolute Relative Error
LMC r2 absolute relative errors
log(ratio) [DV]
0
1
2
3
power−law absolute relative errors
log(ratio) [DV]
0
1
2
3
Purdy absolute relative errors
log(ratio) [DV]
Figure B: The figure displays the absolute log ratio in distance predicted and predicting distance vs. absolute relative error per
runner. In each case the log ratio in distance is displayed on the x-axis and the absolute errors of single data points of the y-axis.
We see that LMC rank 2 is particularly robust for large ratios in comparison to the power law and Purdy Points. Data is taken
from the top 25% of male runners with no. events≥ 3 in the best year.
7
8
9
10
11
12
13
14
0.1
1
10
run time [secs.]
log(no.athletes)
Run Time
Nuclear Norm
LMC
EM algorithm
Figure C: The figure displays mean run-times for the 3 matrix completion algorithms tested in the paper: Nuclear Norm, EM
and LMC (rank 3). Run-times (y-axis) are recorded for completing a single entry in a matrix of size indicated by the x-axis. The
averages are over 100 repetitions, standard errors are estimated by the bootstrap.
0
0.02
0.04
0.06
0.08
0.1
0.04
0.06
0.08
0.1
0.12
0.14
0.16
0.18
0.2
RMSE in log−time
Uniform Missing
LMC rank 2
Nuclear Norm
0
0.02
0.04
0.06
0.08
0.1
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
Clustered Missing Entries
noise−level−nl [log(seconds)]
Figure D: LMC and Nuclear Norm prediction accuracy on the synthetic low-rank data. x-axis denotes the noise level (standard
deviation of additive noise in log-time coordinates); y-axis is out-of-sample RMSE predicting log-time. Left: prediction performance
when (a) the missing entries in each ros are uniform. Right: prediction performance when (b) the observed entries are consecutive.
Error bars are one standard deviation, estimated by the bootstrap.
100m
1500m
5km
Mar
−0.8
−0.6
−0.4
−0.2
0
0.2
0.4
0.6
Singular Components Estimated
from Complete Performances
100m
1500m
5km
Mar
−0.8
−0.6
−0.4
−0.2
0
0.2
0.4
0.6
log(time) [log(seconds)]
Singular Components Estimated
from Incomplete Performances
Singular Component 1
Singular Component 2
Singular Component 3
distance
Figure E: Accuracy of singular component estimation with missing data on synthetic model of performance. x-axis is distance,
y-axis is components in log-time. Left: singular components of data generated according to Equation 1 with all data present. Right:
singular components of data generated according to Equation 1 with missing entries estimated with LMC rank 3; the observation
pattern and number of runners is identical to the real data. The tubes denote one standard deviation estimated by the bootstrap.
100m
1km
10km
Mar
−1
−0.8
−0.6
−0.4
−0.2
0
0.2
0.4
0.6
0.8
Older Runners (30−80yrs) component scores
log(time) [log(seconds)]
Distances
100m
1km
10km
Mar
−1
−0.8
−0.6
−0.4
−0.2
0
0.2
0.4
0.6
0.8
1
Amateur Runners component scores
log(time) [log(seconds)]
Distances
100m
1km
10km
Mar
−0.8
−0.6
−0.4
−0.2
0
0.2
0.4
0.6
Female Runners component scores
log(time) [log(seconds)]
Distances
Figure F: The three components of the low-rank model in subgroups. Left: for older runners. Middle: for amateur runners =
best event below 25th percentile. Right: for female runners. Tubes around the components are one standard deviation, estimated
by the bootstrap. The components are the analogous components for the subgroups described as computed in the left-hand panel
of Figure 2.
Figure G: Scatter plots of training standard vs. three-number-summary (top) and preferred distance vs. three-number-summary.
In each case the individual exponents, 2nd and 3rd scores (λ2, λ3) are displayed on the y-axis and the log-preferred distance and
training standard on the x-axis.
Figure H: Difference of preferred distance and optimal distance, versus age of the runner, colored by specialization distance. Most
runners prefer the distance they are predicted to be best at. There is a mismatch of best and preferred for a group of younger
runners who have greater potential over longer distances, and for a group of older runners who’s potential is maximized over shorter
distances than attempted.
200m
400m
800m
1500m
5km
10km
HM
−0.1
−0.05
0
0.05
0.1
f(ε)/ti
distance
Short Performance peturbations
(1−ε)ti−2>ti−2
(1−ε)ti−2<ti−2
Figure I: Pivot phenomenon in the low-rank model. The figure quantifies the strength and sign of pivoting as in Figure 1, top
right, at different middle distances si (x-axis). The computations are based on equivalent log-time performances ti−1, ti, ti+1 at
consecutive triples si−1, si, si+1 of distances. The y-coordinate indicates the signed relative change of the LMC rank 2 prediction of
ti+1 from ti−1 and ti changes, when ti is fixed and ti−1 undergoes a relative change of 1%, 2%, . . . , 10% (red curves, line thickness
is proportional to change), or −1%, −2%, . . . , −10% (blue curves, line thickness is proportional to change). For example, the largest
peak corresponds to a middle distance of si = 400m. When predicting 800m from 400m and 200m, the predicted log-time ti+1 (=
800m performance) decreases by 8% when ti−1 (= 200m performance) is increased by 10% while ti (= 400m performance) is kept
constant.
Generic
Baselines
State of art
Performance Predictors
State of art
Matrix Completion
Proposed
Method: LMC
evaluation
percentiles
no.events
data type
r.mean
k-NN
individual
power
law
riegel
power law
purdy
nuclear
norm
EM
LMC
rank 1
LMC
rank 2
log time
0-95
3
best
0.1054
0.0421
0.0696
0.0661
0.0654
0.0423
0.1282
0.0387
0.0402
0.0336
±0.0025
±0.0014
±0.0024
±0.0023
±0.0023
±0.0014
±0.0115
±0.0013
±0.0014
±0.0012
normalized
0-95
3
best
0.1062
0.0441
0.0700
0.0681
0.0674
0.0441
0.0907
0.0400
0.0413
0.0347
±0.0027
±0.0018
±0.0026
±0.0026
±0.0025
±0.0017
±0.0051
±0.0016
±0.0016
±0.0014
speed
0-95
3
best
0.5463
0.2118
0.3989
0.3640
0.3600
0.2197
2.2846
0.2023
0.2130
0.1716
±0.0122
±0.0067
±0.0145
±0.0130
±0.0126
±0.0070
±0.8163
±0.0065
±0.0072
±0.0058
log time
0-95
3
random
0.1119
0.0373
0.0655
0.0656
0.0646
0.0411
0.1501
0.0376
0.0385
0.0320
±0.0026
±0.0012
±0.0021
±0.0021
±0.0021
±0.0014
±0.0131
±0.0014
±0.0013
±0.0011
normalized
0-95
3
random
0.1136
0.0397
0.0660
0.0686
0.0676
0.0438
0.1013
0.0395
0.0405
0.0338
±0.0029
±0.0016
±0.0021
±0.0025
±0.0023
±0.0016
±0.0055
±0.0016
±0.0015
±0.0013
speed
0-95
3
random
0.5727
0.1825
0.3795
0.3552
0.3497
0.2066
2.5636
0.1913
0.1995
0.1607
±0.0126
±0.0060
±0.0145
±0.0114
±0.0112
±0.0061
±0.7841
±0.0065
±0.0065
±0.0052
log time
0-95
4
best
0.1014
0.0514
0.0557
0.0551
0.0553
0.0406
0.0716
0.0350
0.0366
0.0310
±0.0024
±0.0016
±0.0017
±0.0020
±0.0020
±0.0013
±0.0051
±0.0013
±0.0013
±0.0010
log time
0-25
3
best
0.0424
0.0294
0.0559
0.0479
0.0507
0.0310
0.0970
0.0282
0.0300
0.0221
±0.0012
±0.0009
±0.0019
±0.0015
±0.0016
±0.0008
±0.0092
±0.0008
±0.0009
±0.0007
Table A: Exactly the same table as Table 1 but mean absolute errors reported.
Generic
Baselines
State of art
Performance Predictors
State of art
Matrix Completion
Proposed
Method: LMC
evaluation
percentiles
no.events
data type
r.mean
k-NN
individual
power
law
riegel
power law
purdy
nuclear
norm
EM
LMC
rank 1
LMC
rank 2
log time
0-95
3
best
0.1398
0.0637
0.1065
0.0100
0.0991
0.0639
0.3624
0.0574
0.0622
0.0545
±0.0066
±0.0057
±0.0054
±0.0063
±0.0063
±0.0066
±0.0849
±0.0060
±0.0054
±0.0050
normalized
0-95
3
best
0.1483
0.0792
0.1103
0.1051
0.1042
0.0724
0.1769
0.0658
0.0694
0.0620
±0.0097
±0.0104
±0.0068
±0.0066
±0.0068
±0.0097
±0.0222
±0.0096
±0.0078
±0.0086
speed
0-95
3
best
0.7153
0.3308
0.6553
0.5827
0.5772
0.3383
19.2009
0.3067
0.3410
0.2918
±0.0349
±0.0348
±0.0356
±0.0368
±0.0385
±0.0440
±9.9799
±0.0393
±0.0310
±0.0321
log time
0-95
3
random
0.1380
0.0544
0.0931
0.0931
0.0919
0.0591
0.4416
0.0561
0.0567
0.0471
±0.0032
±0.0027
±0.0035
±0.0039
±0.0038
±0.0027
±0.0435
±0.0031
±0.0027
±0.0023
normalized
0-95
3
random
0.1450
0.0623
0.0951
0.1011
0.0998
0.0682
0.2046
0.0634
0.0640
0.0538
±0.0044
±0.0037
±0.0038
±0.0049
±0.0049
±0.0039
±0.0124
±0.0041
±0.0038
±0.0033
speed
0-95
3
random
0.6935
0.2585
0.5917
0.5052
0.4979
0.2835
24.7206
0.2801
0.2863
0.2261
±0.0147
±0.0121
±0.0329
±0.0171
±0.0167
±0.0134
±10.7164
±0.0199
±0.0121
±0.0112
log time
0-95
4
best
0.1368
0.0763
0.0823
0.0859
0.0862
0.0620
0.2371
0.0608
0.0599
0.0531
±0.0075
±0.0060
±0.0042
±0.0060
±0.0059
±0.0038
±0.0423
±0.0064
±0.0041
±0.0040
log time
0-25
3
best
0.0539
0.0425
0.0810
0.0675
0.0710
0.0412
0.2479
0.0358
0.0417
0.0318
±0.0027
±0.0030
±0.0056
±0.0050
±0.0051
±0.0026
±0.0600
±0.0022
±0.0030
±0.0022
Table B: Prediction only from events which are earlier in time than the performance to be predicted. The table shows out-of-
sample RMSE for performance prediction methods on different data setups. Predicted performance is of the 25 top percentiles of
male runners, in their best year. Standard errors are bootstrap estimates over 1000 repetitions. Legend is as in Table 1.
Generic
Baselines
State of art
Performance Predictors
State of art
Matrix Completion
Proposed
Method: LMC
evaluation
percentiles
no.events
data type
r.mean
k-NN
individual
power
law
riegel
power law
purdy
nuclear
norm
EM
LMC
rank 1
LMC
rank 2
time
0-95
3
best
0.1295
0.0627
0.0959
0.0973
0.0964
0.0596
0.1785
0.0560
0.0569
0.0499
±0.0027
±0.0027
±0.0035
±0.0064
±0.0065
±0.0025
±0.0105
±0.0028
±0.0023
±0.0024
time
0-95
3
random
0.1357
0.0535
0.0874
0.0907
0.0895
0.0585
0.1961
0.0544
0.0550
0.0461
±0.0029
±0.0022
±0.0028
±0.0031
±0.0031
±0.0026
±0.0116
±0.0025
±0.0022
±0.0020
time
0-95
4
best
0.1232
0.0745
0.0750
0.0782
0.0785
0.0566
0.1167
0.0525
0.0522
0.0455
±0.0025
±0.0031
±0.0021
±0.0027
±0.0027
±0.0021
±0.0084
±0.0029
±0.0019
±0.0019
time
0-25
3
best
0.0559
0.0422
0.0760
0.0668
0.0704
0.0406
0.1579
0.0377
0.0402
0.0302
±0.0015
±0.0016
±0.0025
±0.0022
±0.0023
±0.0012
±0.0113
±0.0012
±0.0014
±0.0001
Table C: Exactly the same table as Table 1 but relative root mean squared errors reported in terms of time. Models are learnt on
the performances in log-time.
Generic
Baselines
State of art
Performance Predictors
State of art
Matrix Completion
Proposed
Method: LMC
evaluation
percentiles
no.events
data type
r.mean
k-NN
individual
power
law
riegel
power law
purdy
nuclear
norm
EM
LMC
rank 1
LMC
rank 2
time
0-95
3
best
0.1057
0.0424
0.0669
0.0654
0.0647
0.0420
0.0876
0.0384
0.0397
0.0333
±0.0023
±0.0015
±0.0022
±0.0023
±0.0024
±0.0014
±0.0048
±0.0013
±0.0013
±0.0012
time
0-95
3
random
0.1116
0.0372
0.0635
0.0651
0.0642
0.0410
0.0980
0.0373
0.0381
0.0318
±0.0024
±0.0012
±0.0018
±0.0019
±0.0020
±0.0013
±0.0055
±0.0013
±0.0013
±0.0011
time
0-95
4
best
0.1006
0.0519
0.0547
0.0540
0.0543
0.0401
0.0605
0.0348
0.0362
0.0307
±0.0023
±0.0016
±0.0016
±0.0018
±0.0018
±0.0013
±0.0032
±0.0013
±0.0012
±0.0011
time
0-25
3
best
0.0425
0.0296
0.0542
0.0476
0.0504
0.0308
0.0688
0.0280
0.0297
0.0220
±0.0011
±0.0001
±0.0017
±0.0015
±0.0016
±0.0008
±0.0046
±0.0008
±0.0008
±0.0007
Table D: Exactly the same table as Table 1 but relative mean absolute errors reported in terms of time. Models are learnt on the
performances in log-time.
no events.
r1
r2
r3
r4
3
0.0411
0.0306
—
—
±0.0014
±0.0011
4
0.0446
0.0328
0.0309
—
±0.0016
±0.0013
±0.0012
5
0.0518
0.0408
0.0400
0.0408
±0.0032
±0.0033
±0.0034
±0.0036
Table E: Determination of the true rank of the model. Table displays out-of-sample RMSE for predicting performance with LMC
rank 1-4 (columns) Predicted performance is of the 25 top percentiles of male runners, in their best year, who have attempted at
least the number of events indicated by the row. The model is learnt on performances in log-time coordinates. Standard errors are
bootstrap estimates over 1000 repetitions. The entries where no. events ≥ rank are empty, as LMC rank r needs r + 1 attempted
events for leave-one-out-validation. Prediction with LMC rank 3 is always better or equally good compared to using a different
rank, in terms of out-of-sample prediction accuracy.
subgroup
RMSE
Amateur
0.0305
±0.0002
Female
0.0305
±0.0003
Old
0.0326
±0.0003
Table F: Prediction in three different subgroups: amateur runners, female runners, older runners. Table displays out-of-sample
RMSE for predicting performance with LMC rank 2.
rank
log time
speed
normalized
1
0.0410
0.0376
0.0399
±0.0014
±0.0011
±0.0013
2
0.0304
0.0315
0.0305
±0.0011
±0.0011
±0.0001
Table G: Effect of performance measure in which the LMC model is learnt. The model is learnt on three different measures of
performance: log-time, time normalized by event mean, speed (columns).
The table shows out-of-sample RMSE for predicting
log-time performance with LMC rank 1,2. Standard errors are bootstrap estimates over 1000 repetitions. Performance is of the 25
top percentiles of male runners, in their best year of performance.
percentiles
no.event
bagged LMC r2
bagged power-law
LMC r2
power-law
0-25
3
0.0310
0.0654
0.0308
0.0666
±0.0011
±0.0025
±0.0011
±0.0025
0-95
3
0.0529
0.0898
0.0512
0.0948
±0.0031
±0.0040
±0.0028
±0.0039
0-95
4
0.0480
0.0762
0.0467
0.0825
±0.0034
±0.0029
±0.0021
±0.0030
Table H: Comparison of prediction using all distances, to prediction using only closest distances. Table displayes out-of-sample
RMSE of predicting log-time, for (5.a) the bagged power law and (5.b) the bagged LMC rank 2 predictor, compared with the un-
bagged variants, (2.b) and (4.b). Predicted performance is of the 25 top percentiles of male runners, in their best year. Standard
errors are bootstrap estimates over 1000 repetitions. The results of the bagging predictors are very similar to the unbagged one.
variables
β
β2
β3
c
model 1
log s
0.0572 ± 0.0003
−0.136 ± 0.003
model 2
log s, f2
0.0547 ± 0.0007
−0.017 ± 0.004
−0.115 ± 0.006
model 3
log s, f2, f3
0.0554 ± 0.0007
−0.013 ± 0.004
0.002 ± 0.001
−0.120 ± 0.006
t1
p(X > |t1|)
t2
p(X > |t2|)
t3
p(X > |t3|)
tc
p(X > |tc|)
model 1
168
1.7e-15
-51
2.3e-11
model 2
81
1.1e-12
-3.9
5.9e-3
-21
1.5e-7
model 3
80
2.5e-10
-3.0
2.5e-2
1.8
0.13
-21
7.1e-7
F
P (X > F )
RSE
R-squared
model 1
2.8e+4
1.7e-15
0.0020
0.9997
model 2
3.9e+4
6.6e-15
0.0012
0.9999
model 3
3.4e+4
4.4e-13
0.0011
0.9999
Table I: Explaining the first singular component, v. The following explanatory linear models are fitted: v explained by β log s + c
(model 1); v explained by β log s + β2f2 + c (model 2); v explained by β log s + β2f2 + β3f3 + c. The β, β2, β3 are the estimated
coefficients, ± one standard error. t1, t2, t3 are the t-statistics of β, β2, β3; tc is the t-statistic of c. The F-statistic of the respective
model is F , RSE is the residual standard error.
| Prediction and Quantification of Individual Athletic Performance of Runners. | 06-23-2016 | Blythe, Duncan A J,Király, Franz J | eng |
PMC4783109 | RESEARCH ARTICLE
Effects of Heavy Strength Training on
Running Performance and Determinants of
Running Performance in Female Endurance
Athletes
Olav Vikmoen1*, Truls Raastad2, Olivier Seynnes2, Kristoffer Bergstrøm2, Stian Ellefsen1,
Bent R. Rønnestad1
1 Section for Sport Science, Lillehammer University College, Lillehammer, Norway, 2 Department of
Physical Performance, Norwegian School of Sport Sciences, Oslo, Norway
* [email protected]
Abstract
Purpose
The purpose of the current study was to investigate the effects of adding strength training to nor-
mal endurance training on running performance and running economy in well-trained female
athletes. We hypothesized that the added strength training would improve performance and
running economy through altered stiffness of the muscle-tendon complex of leg extensors.
Methods
Nineteen female endurance athletes [maximal oxygen consumption (VO2max): 53±3 mlkg-1min-1,
5.8 h weekly endurance training] were randomly assigned to either normal endurance training
(E, n = 8) or normal endurance training combined with strength training (E+S, n = 11). The
strength training consisted of four leg exercises [3 x 4–10 repetition maximum (RM)], twice a
week for 11 weeks. Muscle strength, 40 min all-out running distance, running performance
determinants and patellar tendon stiffness were measured before and after the intervention.
Results
E+S increased 1RM in leg exercises (40 ± 15%) and maximal jumping height in counter
movement jump (6 ± 6%) and squat jump (9 ± 7%, p < 0.05). This was accompanied by
increased muscle fiber cross sectional area of both fiber type I (13 ± 7%) and fiber type II
(31 ± 20%) in m. vastus lateralis (p < 0.05), with no change in capillary density in m. vastus
lateralis or the stiffness of the patellar tendon. Neither E+S nor E changed running economy,
fractional utilization of VO2max or VO2max. There were also no change in running distance
during a 40 min all-out running test in neither of the groups.
Conclusion
Adding heavy strength training to endurance training did not affect 40 min all-out running
performance or running economy compared to endurance training only.
PLOS ONE | DOI:10.1371/journal.pone.0150799
March 8, 2016
1 / 18
OPEN ACCESS
Citation: Vikmoen O, Raastad T, Seynnes O,
Bergstrøm K, Ellefsen S, Rønnestad BR (2016)
Effects of Heavy Strength Training on Running
Performance and Determinants of Running
Performance in Female Endurance Athletes. PLoS
ONE 11(3): e0150799. doi:10.1371/journal.
pone.0150799
Editor: Massimo Sacchetti, University of Rome Foro
Italico, ITALY
Received: June 16, 2015
Accepted: February 20, 2016
Published: March 8, 2016
Copyright: © 2016 Vikmoen et al. This is an open
access article distributed under the terms of the
Creative Commons Attribution License, which permits
unrestricted use, distribution, and reproduction in any
medium, provided the original author and source are
credited.
Data Availability Statement: All relevant data are
within the paper.
Funding: This work was supported by grant 203961
from the Regional Science Fund—Innlandet of
Norway. (http://www.regionaleforskningsfond.no/
prognett-innlandet/Forside/1253953746925). The
funders had no role in study design, data collection
and analysis, decision to publish, or preparation of
the manuscript.
Introduction
The effects of strength training on running performance has been examined in a number of
studies with the majority reporting improved running performance [1–6]. However, the litera-
ture is far from conclusive, as some studies report no beneficial effect of strength training on
running performance [7–10]. Running performance is mainly determined by the maximal oxy-
gen consumption (VO2max), fractional utilization of VO2max and running economy [11]. Addi-
tion of strength training has neither a negative nor a positive effect on VO2max (e.g., [2, 6, 12]).
The effect of combining strength and endurance training on fractional utilization of VO2max
has not been directly investigated, but the indirect measure of VO2 at the lactate threshold,
expressed as percent of VO2max, seems to be unchanged [2, 12]. Running economy on the
other hand seems to be positively affected by strength training (e.g., [2, 6, 12–14]). An
improved running performance following strength training is therefore suggested to be mainly
related to improved running economy [2, 6].
One of the most frequent proposed mechanisms behind improved running economy after
strength training is changes in the stiffness of lower limb muscles and tendons [2, 14, 15]. Dur-
ing the first part of the contact phase in the running stride, elastic energy is stored in the mus-
cles, tendons and ligaments acting across joints [16]. A partial return of this stored energy
during the second part of the contact phase limits the muscle energy expenditure and amplifies
the mechanical output of the muscle-tendon complex [16]. Hence, the stiffness of series elastic
component, mainly tendons, can affect both the utilization of this elastic energy and the muscle
contraction mechanics during the running stride. In fact, stiffer Achilles tendons have been
associated with better running economy [17]. Intriguingly, more compliant patellar tendons
were associated with better running economy [17], whereas heavy strength training has been
shown to increase patellar tendon stiffness [18, 19]. A more compliant patellar tendon may
indeed allow the muscle to operate at mechanically efficient lengths and velocities during the
contact phase [17]. However, for a given tendon stiffness a stronger muscle would enable larger
energy storage. Consequently, heavy strength training might induce changes in muscle and
tendon properties with both potential beneficial and negative effects on running economy. It is
therefore important to gain insight into the effects of strength training on patellar tendon
mechanical properties, and if possible effects induces changes in running economy. However,
to our best knowledge, no studies to date have investigated this.
Most research on the effects of strength training on running performance are performed
with male athletes (e.g., [1, 3, 6, 15]) or a combination of male and female athletes (e.g., [2, 5, 7,
20]). Unfortunately, there is performed a substantial lower volume of research in this area
using only female athletes [10, 13]. Therefore, there is a need for more research with female
athletes. This is especially true regarding the effect of strength training induced changes in
patellar tendon stiffness on running economy since it seems that male and female tendons may
react differently to increased loading [21].
Even though strength training may enhance middle to long distance running performance
through improved running economy it will also normally increase cross sectional area (CSA)
of the muscle fibers [22]. Therefore, it can be speculated that strength training can increase dif-
fusion distances from the capillaries to the interior of muscle cells, which will be negative for
performance. In untrained individuals there are reports of increased or unchanged numbers of
capillaries around each muscle fiber [23, 24] and no change in capillaries per fiber area [24]
after strength training. However, as performing endurance training concurrently with strength
training may blunt the hypertrophic response (e.g., [25]), and endurance trained athletes have
larger numbers of capillaries than untrained [26, 27] these findings may not apply for
Strength Training and Running Performance
PLOS ONE | DOI:10.1371/journal.pone.0150799
March 8, 2016
2 / 18
Competing Interests: The authors have declared
that no competing interests exist.
endurance athletes. Consequently, there is a need to look closer into the effects of combined
strength and endurance training on capillarization and fiber CSA in well-trained endurance
athletes.
The purpose of this study was to investigate the effects of 11 weeks of heavy strength train-
ing on running performance during a 40 min all-out test and running economy in well-trained
female endurance athletes. Furthermore, we wanted to assess the effects of the strength training
on the mechanical properties of the patellar tendon to elucidate whether this could be related
to changes in running performance and running economy. To investigate if strength training
would have any effect on capillarization in endurance athletes we measured muscle fiber CSA
and capillarization in m. vastus lateralis.
We hypothesized that the addition of heavy strength training would result in improved 40
min all-out performance and improved running economy and that these changes would be
related to changes in mechanical properties of the patellar tendon, together with no negative
effect on capillarization.
Materials and Methods
Ethical approval
The study was approved by the Local Ethics Committee at Lillehammer University College.
Written informed consent was obtained from all athletes prior to inclusion, and the study was
carried out in accordance with the Declaration of Helsinki.
Participants
Twenty-eight female endurance athletes active in both cycling and running and that fulfilled at
least two of Jeunkedrup et al.’s [28] training and race status descriptions of a well-trained
endurance athlete were recruited to this study. None of the athletes had performed systematic
strength training for the last 12 months leading up to the study (less than one session per
week). The athletes were matched on VO2max and randomly assigned to either adding heavy
strength training to the ongoing endurance training (E+S, n = 14) or endurance training only
(E, n = 14). During the study, three athletes in E+S left the project for reasons unrelated to the
project protocol: one because of an injury, one because of a prolonged period of illness during
the last part of the intervention and one because of other medical reasons. In E, six athletes left
the study for reasons unrelated to the project protocol (injuries in training, pregnancy and lack
of time). Therefore, the final numbers of athletes in E+S and E were 11 and 8, respectively.
Experimental overview
This study is part of a larger study investigating the effects of heavy strength training on various
aspects of endurance performance. Some of the basic characteristics as anthropometrics and
endurance training have been reported previously [29].
The strength training program for the E+S group consisted of two strength training sessions
per week and lasted for 11 weeks (during the competition period from April to July). Testing
before and after the intervention period was performed over four test-days. During pre-tests,
day one was utilized to sample muscle biopsies from the right m. vastus lateralis, and measure
the mechanical properties of the left patellar tendon. At day two 1RM in one-legged leg press
and half squat was measured. Day 3 consisted of an incremental running test for determination
of blood lactate profile, a VO2max test and testing of maximal squat jump (SJ) and counter
movement jump (CMJ) height. Day 4 consisted of a 40 min all-out running test. There were at
least 7 days between day one and two and 3–7 days between the remaining test-days. All tests
Strength Training and Running Performance
PLOS ONE | DOI:10.1371/journal.pone.0150799
March 8, 2016
3 / 18
for each participant were completed within 2–3 weeks. During post tests, athletes in E+S main-
tained their strength training with one session per week until all testing were complete. In gen-
eral, post-tests were performed in the same order as pre-tests. However, muscle biopsies and
patellar tendon measurements were moved to the last test day. The athletes did not perform
any systematic periodization so neither the pre tests nor post tests were performed in a particu-
lar phase of periodization.
Training
Endurance training duration and intensity were calculated based on heart rate (HR) record-
ings. Endurance training was divided into three HR zones: 1) 60%-82%, 2) 83%-87%, and 3)
88%-100% of maximal HR. The endurance training performed has been previously reported
[29]. In brief, there were no significant differences between groups in the mean weekly duration
of the endurance training, the distribution of this training within the three intensity zones
(across groups values were: zone 1: 3.7 ± 1.6 h, zone 2: 1.1 ± 0.5 h, zone 3: 0.8 ± 0.5) and the
numbers of endurance training sessions per week (across groups values were 4.3 ± 1 session
week-1).
The heavy strength training sessions for the E+S groups targeted leg muscles and were per-
formed twice per week during the 11-week intervention period. Adherence to the strength
training was high, with E+S athletes completing 21.4 ± 1.0 (range 19–22) of the planned 22
strength-training sessions. The strength-training program was performed as reported in Vik-
moen et al. [29]. Briefly, each strength training session consisted of four leg exercises: half
squat in a smith machine (Gym 80 International, Gelsenkirchen, Germany), leg press with one
leg at a time (Gym 80 International, Gelsenkirchen, Germany), standing one-legged hip flexion
in a cable cross machine (Gym 80 International, Gelsenkirchen, Germany), and ankle plantar
flexion in a smith machine. For a detailed description of the exercises, see Ronnestad et al. [30].
Three sets were performed per exercise. An investigator supervised the athletes at all workouts
during the first two weeks and at least one workout per week thereafter. During weeks one to
three, athletes trained with 10RM sets at the first session and 6RM sets at the second session.
These alternating loads were adjusted to 8RM and 5RM during weeks four to six, and was fur-
ther adjusted to 6RM and 4RM during weeks seven to eleven (Table 1). The numbers of repeti-
tions was always the same as the prescribed RM load meaning that the sets were performed
until failure, and the athletes adjusted the absolute load as they got stronger to correspond to
the prescribed RM load. The athletes were allowed assistance on the last repetition if necessary.
Because a proposed mechanisms behind improved running performance after strength training
is an increased rate of force development [2] the athletes were instructed to perform the concen-
tric phase of the exercises with focus on maximal effort (duration around 1 s) while the eccentric
phase was performed more slowly (duration 2–3 s). During each strength training session the
athletes consumed a bar containing 15 g protein (Squeezy recovery bar, Squeezy Sports Nutrition,
Braunschweig, Germany) to ensure adequate protein intake in conjunction with the strength
Table 1. Training loads used during the strength training intervention
Week
Load session 1
Load session 2
1–3
10RM
6RM
4–6
8RM
5RM
7–11
6RM
4RM
RM: Repetition maximum
doi:10.1371/journal.pone.0150799.t001
Strength Training and Running Performance
PLOS ONE | DOI:10.1371/journal.pone.0150799
March 8, 2016
4 / 18
training sessions. The athletes were encouraged to perform the strength training and endurance
training on separate days. On days the athletes had to perform both endurance and strength
training, they were encouraged to perform strength training in the first session of the day.
Strength, jumping and running tests
The athletes were instructed to refrain from intense exercise the day preceding testing, to pre-
pare for the tests as they would have done for a competition, and to consume the same type of
meal before each test. Running tests was performed on a motor driven treadmill (Woodway
Desmo Evo, Waukesha, Wisconsin, USA). The inclination of the treadmill was set to 5.3% at
all tests. All testing were performed under similar environmental conditions (18–20˚C).
1RM tests. 1RM was tested in one-legged leg press and half squat and the mean value
from these two exercises were used for statistical analyses. Prior to the testing day, each athlete
was given a supervised familiarization session to learn proper lifting technique, find individual
equipment settings and practice SJ and CMJ. During this session, the load was gradually
increased to allow estimation of a proper starting point for the 1RM testing.
The 1RM tests in both exercises were performed as previously described (Vikmoen et al.
2015). Briefly, testing started with a specific warm-up, consisting of three sets with gradually
increasing load (40, 75 and 85% of expected 1RM) and decreasing number of repetitions
(10!6!3). The first attempt was performed with a load approximately 5% below the expected
1RM. If a lift was successful, the load was increased by approximately 5%. The test was termi-
nated when the athletes failed to lift the load in 2–3 attempts and the highest successful load
lifted was noted as 1RM. Athletes were given 3 min rest between lifts.
Blood lactate profile.
The blood lactate profile tests started with 5 min running at 7 kmh-1,
which was subsequently increased every 5 min by 1 kmh-1. Between consecutive 5 min bouts there
was a 1 min break, wherein blood was sampled from a finger-tip and analyzed for whole blood lac-
tate concentration ([la-]) using a Lactate Pro LT-1710 analyzer (Arcray Inc., Kyoto, Japan), and the
rating of perceived exertion (RPE) was recorded. The test was terminated when a [la-] of 4 mmolL-1
or higher was measured. VO2 and HR were measured during the last 3 min of each bout, and
mean values were used for statistical analysis. VO2 was measured (30 s sampling time) using a
computerized metabolic system with mixing chamber (Oxycon Pro, Erich Jaeger, Hoechberg, Ger-
many). The gas analyzers were calibrated with certified calibration gases of known concentrations
before every test. The flow turbine (Triple V, Erich Jaeger, Hoechberg, Germany) was calibrated
before every test with a 3 l, 5530 series, calibration syringe (Hans Rudolph, Kansas City, USA). HR
was recorded using a Polar S610i heart rate monitor (Polar, Kempele, Finland). From this incre-
mental running test, the running velocity at 3.5 mmolL-1 [la-] was calculated for each athlete from
the relationship between [la-] and running velocity using linear regression between data points.
Running economy was determined by the mean VO2 at a running velocity of 10 kmh-1.
VO2max.
After termination of the blood lactate profile test the athletes ran for 10 min at a
freely chosen submaximal workload. The VO2max test was then initiated with 1 min running at
8 kmh-1 and the speed was increased by 1 kmh-1 every minute until exhaustion. The athletes
received strong verbal encouragement to run for as long as possible. VO2 was measured contin-
uously, and VO2max was calculated as the mean of the two highest 30 s VO2 measurements.
The VO2max test was considered valid when two or more of the following criteria were met: a
plateau in VO2 was despite increased workload, a respiratory exchange ration above 1.1 and
HRpeak ± 10 bpm pf the predicted maximal HR (220-age) [31]. Peak running performance dur-
ing the test (Vmax) was calculated as the mean running velocity during the last 2 min of the
incremental test. The highest HR recorded during the test was taken as HRpeak and immedi-
ately after the test blood [la-] and RPE were recorded.
Strength Training and Running Performance
PLOS ONE | DOI:10.1371/journal.pone.0150799
March 8, 2016
5 / 18
SJ and CMJ.
Twenty min after termination of the VO2max test, explosive strength was
tested as maximal jumping height in SJ and CMJ. These jumps were performed on a force plate
(SG-9, Advanced Mechanical Technologies, Newton, MA, USA, sampling frequency of 1kHz).
After 3–5 submaximal warm up jumps, the athletes performed three SJ and three CMJ with
2–3 min rest between jumps. The highest SJ and CMJ were utilized for statistical analyses. Dur-
ing all jumps the athletes were instructed to keep their hands placed on their hips and aim for
maximal jumping height. The SJ was performed from ~90 degrees knee angle. In this position,
they paused for 3 s before jumping. No downward movement was allowed prior to the jump
and the force curves were inspected to verify this. During the eccentric phase of the CMJ the
athletes were instructed to turn at a knee angle they felt was optimal for achieving maximal
jumping height.
40 min all-out test.
Prior to the 40 min all-out test, athletes performed 10 min warm up at
self-selected submaximal velocities, containing three submaximal sprints performed during the
last 2 min. These sprints were standardized from pre to post in each athlete. During the first 5
min of the test, the investigators set the velocity. This individual selected velocity was based on
the lactate profile test and corresponded to the velocity at 2.5 mmolL-1 [la-]. Thereafter, run-
ning velocity were controlled by the athletes themselves, with instructions to run the greatest
distance possible during the 40 min. Measurements of VO2 was made during the last minute of
every 5 min section, to allow estimation of performance VO2 and fractional utilization of
VO2max,. During this minute, athletes were not allowed to adjust the running velocity. The
mean VO2 during this minute was estimated to reflect the mean VO2 during the corresponding
5 min section. During the last 5 min of the test, VO2 was measured continuously as pilot testing
showed that athletes performed numerous velocity adjustments during this part of the test.
Performance VO2 was calculated as the mean VO2 of all 5 min sections, and fractional utiliza-
tion of VO2max was calculated as performance VO2 in percentage of VO2max. During the test,
the athletes were allowed to drink water ad libitum.
Measurements of the mechanical and material properties of the patellar
tendon
All the measurements of the mechanical and material properties of the patellar tendon were
performed on the left leg and were done as previously described [32]. Briefly, the athletes were
seated with a 90° angle in both knee and hip joint in a knee extension apparatus (Knee exten-
sion, Gym 2000, Geithus, Norway) instrumented with a force cell (U2A, Hottinger Baldwin
Messtechnik GmbH, Darmstadt, Germany). To measure patellar tendon CSA, transversal
scans were performed proximally, medially and distally along the tendon length using an B-
mode ultrasound apparatus (HD11XE, Phillips, Bothell, WA, USA). Sagittal scanning was used
to measure tendon length. To measure tendon force and elongation the ultrasound probe was
attached to the left knee with a custom-made device. The athletes performed ramp contractions
at a constant rate of 100 Ns-1. To correct for hamstring co-activation when calculating tendon
force (see below), a maximal isometric knee flexion were performed after the knee extension
test. In addition, EMG data were recorded (TeleMyo 2400 G2 telemetry Systems, Noraxon
Inc., Scottsdale, AZ, USA) from the biceps femoris muscle during isometric knee extension and
flexion. Patellar tendon force (FPT) was calculated as the force measured in the force cell, cor-
rected for hamstring co-activation, internal and external moment arms as follows:
FPT ¼ ððFq þ FhÞMeÞ=Mi
Where Fq is force measured by the force cell, Fh is estimated hamstrings co-activation force, Mi
and Me corresponds to internal and external moment arm respectively.
Strength Training and Running Performance
PLOS ONE | DOI:10.1371/journal.pone.0150799
March 8, 2016
6 / 18
Tendon morphology data were analysed as previously described [32], using an image analy-
sis software (ImageJ 1.45s, National Institute of Health, Austin, TE, USA). Tendon elongation
data were analyzed using a video analysis software (Tracker Video Analysis and Modeling
Tool, Open Source Physics, Douglas Brown, 2012). The patellar apex and the tibia plateau were
digitally marked within a common coordinate system. The actual elongation of the tendon was
calculated as the change in the distance between coordinates of these anatomical landmarks.
To calculate tendon material and mechanical properties force-elongation curves were fitted
with a 2nd degree polynomial. All the recordings used in the results had a fit of R2 = 0,92 or
higher. Stiffness was calculated as the slope of the force–elongation curve, between 90 and
100% of each athlete’s maximal force. The Young’s modulus was calculated by multiplying the
stiffness values by the ratio between the patellar tendon resting length (l0) and mean CSA.
Patellar tendon l0 and maximal length (lmax) was used to calculate the patellar tendon strain.
Two sets of ultrasound data (from two E+S athletes) had to be discarded because of an insuffi-
cient quality to enable analysis. Therefore, the number of athletes included in the data from
tendon testing is 9 in E+S and 8 in E.
Muscle biopsy sampling
Muscle biopsies were sampled from m. vastus lateralis using the Bergström procedure. Athletes
were instructed to refrain from physical activity for the last 24h leading up to biopsy sampling.
During each biopsy sampling-event, two separate muscle biopsies were retrieved and pooled in
a petri dish filled with sterile physiological salt water. An appropriately sized muscle sample
(mean wet weight: 29 ± 8 mg) was selected for immunohistochemical analyses and mounted in
Tissue-Tek (Sakura Finetek USA, Inc., Torrance, CA, USA) and quickly frozen in isopentane
cooled on liquid nitrogen. Muscle samples were stored at– 80°C until later analyses.
Immunohistochemistry
Cross-sections 8 μm thick were cut using a microtome at −20°C (CM3050; Leica Microsystems
GmbH, Wetzlar, Germany) and mounted on microscope slides (Superfrost Plus; Thermo
Fisher Scientific, Inc., Waltham, MA, USA). The sections were then air-dried and stored at
−80°C. Prior to antibody labelling, muscle sections were blocked in a solution containing 1%
BSA (cat. no. A4503; Sigma-Aldrich Corp., St Louis, MO, USA) and 0.05% PBS-T solution
(cat. no. 524650; Calbiochem, EMD Biosciences, Inc., San Diego, CA, USA) for 30 min. Then
they were incubated overnight at 4°C with antibodies against the capillary marker CD31
(1:200; clone JC70A, M0823; Dako A/S, Glostrup, Denmark), followed by incubation with
appropriate secondary antibodies (Alexa Fluor, cat. no. A11005).
Following staining, muscle sections were visualized and pictures taken using a high-resolu-
tion camera (DP72; Olympus Corp., Tokyo, Japan) mounted on a microscope (BX61; Olympus
Corp.) with a fluorescence light source (X-Cite 120PCQ; EXFO Photonic Solutions Inc., Mis-
sissauga, Ontario, Canada).
These muscle sections were then incubated for 1 hour at room temperature with antibodies
against myosin heavy chain type II (1:1000; SC71; gift from Professor S. Schiaffino) and dystro-
phin (1:1000; cat. no. ab15277; Abcam Plc), followed by incubation with appropriate secondary
antibodies (Alexa Fluor, cat. no. A11005 or A11001; Invitrogen, Inc.).
Muscle sections were then covered with a coverslip and glued with ProLong Gold Antifade
Reagent with DAPI (cat.no. P36935; Invitrogen Molecular Probes, Eugene, OR,USA) and left
to dry overnight at room temperature. Muscle sections were again visualized and new pictures
was taken at the exactly same location in the section as the CD31 picture. Between all stages,
sections were washed for 3 × 5 min using a 0.05% PBS-T solution.
Strength Training and Running Performance
PLOS ONE | DOI:10.1371/journal.pone.0150799
March 8, 2016
7 / 18
Fiber type distribution, fiber cross-sectional area and capillaries were identified using
TEMA software (CheckVision, Hadsund, Denmark). Capillarization was expressed as capillar-
ies around each fiber (CAF) and capillaries related to fiber area (CAFA), for type I and type II
(IIA and IIX) fibers. Because of technical problems with some analyses, the number of athletes
in the immunohistochemistry data is 8 in E+S and 5 in E.
Statistical analyses
All data in the text, Figs and tables are presented as mean ± standard deviation, unless other-
wise stated. Data were analyzed using two-way (group x time) repeated measures ANOVA.
Effect sizes (ES) were calculated for key performance and physiological adaptations to elucidate
on the practical significance of strength training. ES were calculated as Cohen’s d and the crite-
ria to interpret the magnitude were the following: 0–0.2 = trivial, 0.2–0.6 = small, 0.6–
1.2 = moderate, 1.2–2.0 = large and > 2 = very large [33].
Correlations analyses were done using the Pearson product-moment method. Analyses was
performed in Excel 2013 (Microsoft Corporation, Redmon, WA, USA). Analyses resulting in
p 0.05 were considered statistically significant.
Results
There were no significant differences between E+S and E in any of the measured variables at
baseline.
Body mass, maximal leg strength and muscle fiber area
Body mass remained unchanged in E+S (from 62.4 ± 5.2 kg to 63.1 ± 5.6 kg) but was slightly
reduced in E (from 65.6 ± 8.4 kg to 64.8 ± 8.0 kg, p < 0.05). There was a significant interaction
(p < 0.05) between group and time (pre vs post) indicating that the change in body mass was
different between the groups.
1RM in the leg exercises increased 40.4 ± 14.7% in E+S (p < 0.01, Fig 1) with no change in
E. There was a significant interaction (p < 0.01) between group and time (pre vs post). In addi-
tion, the effect size analyses revealed a very large practical effect of E+S compared to E
(ES = 3.20).
In E+S, CSA of both type I and type II muscle fibers increased in m. vastus lateralis
(13.2 ± 6.8% and 30.8 ± 19.6%, respectively, p < 0.01), with no changes occurring in E (Fig 2).
E+S had a moderate practical effect on muscle fiber CSA compared to E (ES = 0.83).
SJ and CMJ
E+S increased SJ and CMJ height by 8.9 ± 6.8% and 5.9 ± 6.0% respectively (p < 0.05) while no
changes occurred in E (Fig 1). The effect size analyses revealed a moderate practical effect in
favor of E+S in both SJ (ES = 1.06) and CMJ (ES = 0.65).
Capillarization
None of the groups had any change in CAF or CAFA around neither type I nor type II fibers
(Fig 3).
Mechanical and material properties of the patellar tendon
There were no significant changes in stiffness or Young’s modulus of the patellar tendon in nei-
ther E+S nor E (Table 2). The mean CSA of the patellar tendon increased by 5.2 ± 3.6% in E+S
(p < 0.01) while no significant changes occurred in E (Table 2).
Strength Training and Running Performance
PLOS ONE | DOI:10.1371/journal.pone.0150799
March 8, 2016
8 / 18
Fig 1. Maximal strength and jumping performance. Individual values (dotted lines) and mean values (solid
lines) before (Pre) and after (Post) the intervention period for athletes adding strength training to their normal
endurance training (E+S) and athletes performing normal endurance training only (E). a: Squat jump (SJ)
height. b: Counter movement jump (CMJ) height. c: Mean 1 repetition maximum (1RM) in half-squat and one-
legged leg press (leg exercises). * Different from pre (p < 0.05), # significant interaction between group and
time (p < 0.05)
doi:10.1371/journal.pone.0150799.g001
Strength Training and Running Performance
PLOS ONE | DOI:10.1371/journal.pone.0150799
March 8, 2016
9 / 18
VO2max and Vmax
Both VO2max and Vmax was unchanged in both groups during the intervention period
(Table 3).
Running economy and running velocity at 3.5 mmolL-1 [la-]
There were no changes in running economy measured at 10 kmh-1 during the blood lactate
profile test (Fig 4) or running velocity at 3.5 mmolL-1 [la-] (Fig 4) in neither of the groups.
40 min all-out test
There were no change in running distance or performance VO2 during the 40 min all-out test
in neither of the groups during the intervention (Fig 4). Fractional utilization of VO2max did
not change in E+S (from 85.3 ± 3.9 to 85.3 ± 4.3, Fig 4), but increased in E, going from
83.2 ± 3.1% to 86.0 ± 3.0% (p < 0.05).
Before the intervention the performance in the 40 min all-out test correlated with velocity at
3.5 mmolL-1 [la-], VO2max and Vmax (r = 0.65, r = 0. 58, r = 0.79, respectively), but not with
running economy (r = -0.24). No significant correlations between changes in these variables
and changes in 40 min all-out running distance were observed.
Discussion
The main results from the current study were that adding heavy strength training to well-
trained female athletes`normal endurance training did not affect the mechanical properties of
the patellar tendon or running economy. Furthermore, there was no effect on running perfor-
mance during a 40 min all-out running test. Strength training had no negative effect on capil-
lary density despite increased muscle fiber CSA and muscle strength.
Maximal strength and muscle fiber cross sectional area
The strength-training program used in the current study was effective in increasing maximal
leg strength as shown by an increase in 1RM in the leg exercises. This is in accordance with
Fig 2. Muscle fiber cross sectional area. Individual values (dotted lines) and mean values (solid lines) before (Pre) and after (Post) the intervention period
for athletes adding strength training to their normal endurance training (E+S, left panel) and athletes performing normal endurance training only (E, right
panel). Muscle fiber cross sectional area (CSA) for both type I muscle fibers and type II muscle fibers * Different from pre (p < 0.05)
doi:10.1371/journal.pone.0150799.g002
Strength Training and Running Performance
PLOS ONE | DOI:10.1371/journal.pone.0150799
March 8, 2016
10 / 18
previously observed increases in 1RM in endurance athletes adding heavy strength training to
their normal endurance training (e.g., [2, 4, 13]). The results from the current study confirms
previous results [2, 34] that a quite large increase in muscular strength can be achived without
an increased body mass. This is important for runners since increased body mass can negatively
influence running performance. In spite of this, the improved strength seemed to be at least par-
tially dependent on increased muscle mass, as evident from the increased muscle fiber CSA. The
present muscle hypertrophy is in agreement with other studies using similar strength training
protocols in endurance athletes [34–36]. Interestingly, there were no difference in the CSA of the
type I and type II fibers in the current athletes, confirming the notion that in endurance athletes
the type I fibers may be just as large [37] or even larger [38] than the type II fibers.
Fig 3. Capillarization. Individual values (open symbols) and mean values (solid squares) for athletes adding
strength training to their normal endurance training (E+S) and athletes performing normal endurance training
only (E). a: Percent change in capillaries around each muscle fiber (CAF) for both muscle fiber type I and
muscle fiber type II for E+S and E. b: Percent change in capillaries related to fiber area (CAFA) for both
muscle fiber type I and muscle fiber type II for E+S and E.
doi:10.1371/journal.pone.0150799.g003
Strength Training and Running Performance
PLOS ONE | DOI:10.1371/journal.pone.0150799
March 8, 2016
11 / 18
SJ and CMJ
The current strength training protocol was also effective in increasing leg muscle power, as evi-
dent from the increased SJ and CMJ performance. This is in line with previous reports of effects
of heavy strength training on jumping ability in untrained participants [39, 40]. However, pre-
vious data from endurance athletes are more unclear, as some studies report improved jumping
performance [36, 41] whereas others do not [14, 20]. The current study indicate that quite
large improvements in jumping ability and explosive strength can be achived with heavy
strength traning despite concurrently performing endurance training.
Capillarization
Eleven weeks of heavy strength training did not affect capillarization expressed as either CAF
or CAFA, despite leading to significant muscle fiber hypertrophy. Importantly, this suggest
that the potentially negative effect of increased muscle fiber CSA on diffusion distances
between blood and inner parts of muscle fibers was counteracted by a non-significant increase
in CAF. However, this data should be treated with caution because of the limited sample size.
Despite of this, our finding are in line with previous studies in untrained participants that have
reported either no change or a slight increase in CAF [23, 24] and no change in CAFA [24]
after a period of heavy strength training. Our finding is also in agreement with results reported
in elite male cyclists after 16 weeks of heavy strength training [35]. Therefore, it seems like
endurance athletes should not be afraid of reduced capillarization when they consider adding
heavy strength training to their ongoing endurance training.
Mechanical properties of the patellar tendon
The lack of changes in mechanical properties of the patellar tendon following heavy strength
training is in contrast to most studies, typically reporting increased patellar tendon stiffness, at
Table 2. Stiffness, Young’s modulus and mean cross section area (CSA) of the patellar tendon.
E+S
E
Pre
Post
Pre
Post
Stiffness (Nmm-1)
2752 ± 402
2483 ± 733
2753 ± 947
2692 ± 697
Young’s Modulus (MPa)
1038 ± 194
925 ± 162
1251 ± 296
1158 ± 273
Mean CSA (mm2)
65.9 ± 7.1
69.2 ± 6.9*
60.3 ± 4.2
59.9 ± 4.4
Stiffness, Young’s modulus and mean cross section area (CSA) of the patellar tendon before (Pre) and after (Post) the intervention period for athletes
adding strength training to their normal endurance training (E+S) and athletes performing normal endurance training only (E). Values are mean ± SD
* Different from pre (p < 0.05).
doi:10.1371/journal.pone.0150799.t002
Table 3. Data from the maximal oxygen consumption test.
E+S
E
Pre
Post
Pre
Post
VO2max (mlkg-1min-1)
52.2 ± 2.3
52.7 ± 3.3
54.2 ± 2.9
53.1 ± 1.9
Vmax (kmh-1)
12.8 ± 0.7
13.0 ± 0.9
13.1 ± 0.5
13.3 ± 0.6
HRpeak (beatsmin-1)
193 ± 9
192 ± 9
189 ± 8
187 ± 7
RPE
19 ± 1
20 ± 1
19 ± 1
19 ± 1
[La-1]peak (mmoll-1)
9.7 ± 3.0
8.1 ± 3.8
8.9 ± 2.2
7.7 ± 1.8
Data from the maximal oxygen consumption (VO2max) test before (Pre) and after (Post) the intervention period for athletes adding strength training to their
normal endurance training (E+S) and athletes performing normal endurance training only (E). Values are mean ± SD.
doi:10.1371/journal.pone.0150799.t003
Strength Training and Running Performance
PLOS ONE | DOI:10.1371/journal.pone.0150799
March 8, 2016
12 / 18
least in previously untrained participants [19, 42–44]. A possible reason for the discrepancy in
results between our study and that by others is that we included female participants while the
other studies included males [19, 42–44]. In fact, female tendons have been reported to show a
lower rate of new connective tissue formation in response to mechanical loading [21]. Differ-
ences in the strength training protocol may also explain the lack of changes in the current
study. Indeed, most previous studies reporting increased patellar tendon stiffness following
strength training have included heavy knee extension exercise [19, 42, 44] or isometric muscle
actions [45]. In the current study, the exercises involved were more complex involving multiple
joints that perhaps reduced the absolute mechanical loading on the patellar tendon compared
to a pure knee extension exercise. In addition, the athletes were instructed to perform the con-
centric phase of the exercises as fast as possible making the time under tension quite low.
In contrast to the lack of effect of strength training on patellar tendon properties, it led to
increases in its CSA. In line with these findings, some studies on the effect of strength training,
yet not all [18], reports an increase in patellar tendon CSA [19, 42]. Without changes in
mechanical properties, the tendon hypertrophy measured here suggests that material proper-
ties may also have been altered. The lack of change in Young’s modulus following training may
highlight the limitation of this parameter based on finite tendon sections to reflect whole ten-
don material properties. Interpreting the mechanisms driving tendon hypertrophy extends
beyond the scope of the present article. One could speculate that increasing tendon CSA may
Fig 4. Determinants of running performance and running performance. Individual values (dotted lines) and mean values (solid lines) before (Pre) and
after (Post) the intervention period for athletes adding strength training to their normal endurance training (E+S) and athletes performing normal endurance
training only (E). a: Body mass adjusted oxygen consumption at 10 kmh-1. b: Running velocity at 3.5 mmolL-1 [la-] calculated during the blood lactate profile
test. c: The fractional utilization of VO2max during the 40 min all-out test. d: Running distance during the 40 min all-out test. * Different from pre (p < 0.05).
doi:10.1371/journal.pone.0150799.g004
Strength Training and Running Performance
PLOS ONE | DOI:10.1371/journal.pone.0150799
March 8, 2016
13 / 18
shield this tissue against damage caused by excessive and/or unusual stresses. Taken together,
the present measurements indicate that resistance training triggers an adaptive response in the
patellar tendon of female runners, without affecting the mechanical properties of this tissue.
Whether this adaptation may affect injury rates or have other effects amongst runners warrants
further investigation.
VO2max, fractional utilization of VO2max, running velocity at 3.5 mmolL-1
blood [La-], and running economy
The lack of change in VO2max after strength training is not surprising and is actually in accor-
dance with the current literature (e.g., [6, 12, 46]). Fractional utilization of VO2max measured
during the 40 min all-out test did not change in E+S during the course of the study. To our
knowledge, this is the first study directly measuring fractional utilization of VO2max in running
after addition of heavy strength training in endurance athletes. However, VO2 at lactate thresh-
old in percentage of VO2max, is often taken as an indirect measure of fractional utilization of
VO2max [11]. The few studies measuring this variable in running reports no effect after addition
of heavy strength training [2, 12]. Notably, there was a slight increase in fractional utilization
of VO2max in E over the course of the intervention. This was likely due to a combination of two
factors; a small but non-significant reduction in VO2max, largely due to one athlete exhibiting a
large reduction, and a small but non-significant increase in performance VO2.
Surprisingly, we found no effect of heavy strength training on running economy, contrast-
ing the majority of previous studies, typically reporting improvements from 3–8% [2, 6, 13,
14]. However, some studies supports the lack of an effect of strength training on running econ-
omy [1, 7, 8, 47]. In two of this studies [7, 47] the lack of improved running economy might be
because the strength training program only consisted of one session for the legs per week.
Supporting the lack of an effect on VO2max, fractional utilization of VO2max and running
economy, strength training had no effect on running velocity at 3.5 mmolL-1 blood [La-]. The
latter is in accordance with the majority of the current literature which reports no change in
velocity at a certain blood [la-1] or ventilatory threshold after adding various forms of strength
training to runners’ normal training [2, 7, 12], although exceptions exist [14]. This is quite sur-
prising considering that improved running economy in theory should affect the running speed
at a certain lactate threshold [11].
Running performance
The lack of changes in 40 min all-out performance is not in line with many of the studies in
this area where improved running performance have been reported [1, 2, 4–6]. However, no
change in performance is in line with the present lack of changes in the important performance
determining factors like VO2max, running economy and fractional utilization of VO2max. Since
strength training does not affect VO2max and the fractional utilization of VO2max, the mecha-
nism for the observations of improved running performance in some other studies seems to be
improved running economy [2, 6]. However, not all studies have found strength training to be
beneficial for running performance [7, 8, 10, 47], and are in accordance with the current study.
Interestingly, these studies do also report no improvements in running economy. Therefore,
the lack of improved running performance in the current study is probably because of no
changes in running economy.
Whilst unclear, the discrepancies in training-induced running performance measures
between the current study and that by others may be attributed to methodological differences.
In the current study, all running tests were performed at 5.3% inclination. This inclination
resulted in a quite low running velocity compared to some other studies. Indeed, changes in
Strength Training and Running Performance
PLOS ONE | DOI:10.1371/journal.pone.0150799
March 8, 2016
14 / 18
running economy after strength training have previously been found to be related to running
velocity [46]. However, improvements in running economy after strength training have also
been reported at similar velocities [20, 36] and at the same inclination [48] used in the current
study. Therefore, the inclination used is probably not the only explanation why no changes in
running economy and performance were observed.
The fact that this study includes only female athletes while most previous studies include
either only males or a mix of both male and female runners this may perhaps explain why no
effects of strength training on running economy was observed. However, strength training
have been reported to improve running economy in female runners [13] making this explana-
tion unlikely.
One of the most frequent proposed mechanism for the possible ergogenic effect of strength
training on running economy is changes in stiffness of the muscle or tendons in the legs [2,
14]. Despite this speculation, studies have yet to investigate the effect of heavy strength training
on patellar tendon mechanical properties in conjunction with running economy. In the current
study, the unchanged tendon stiffness and increased strength suggest that more elastic energy
may be stored in the patellar tendon during the stance phase, amplifying muscular power out-
put and efficiency. However, the lack of changes in running economy do not support this
hypothesis, and conclusion cannot be drawn regarding the influence of patellar tendon proper-
ties in the present study.
Vmax has been reported to be the best laboratory measure to predict performance in various
running distances [49] and can actually be considered a measure of running performance [50].
The lack of increased Vmax further indicates that heavy strength training did not lead to
improved running performance in the current study. It has previously been reported both
improved [6, 36] and no change [14] in Vmax after heavy strength training in trained runners.
Conclusion
In contrast to our hypothesis, adding heavy strength training to endurance training in well-
trained female endurance athletes did not affect running performance measured as running
distance during a 40 min all-out test. The lack of effect on performance was probably because
the strength training intervention did not improve running economy or changed the mechani-
cal properties of the patellar tendon. However, strength training had no negative effect capillary
density.
Acknowledgments
The authors would like to thank the participants for their time and effort. Students Øyvind
Trøen, Roger Kristoffersen, Allan Sørgaard Nielsen and Sondre Prestkvern for assistant during
the intervention follow-up and data sampling.
Author Contributions
Conceived and designed the experiments: OV TR OS SE BRR. Performed the experiments: OV
TR OS KB SE BRR. Analyzed the data: OV TR OS KB SE BRR. Contributed reagents/materi-
als/analysis tools: OV TR OS KB SE BRR. Wrote the paper: OV TR OS KB SE BRR.
References
1.
Damasceno MV, Lima-Silva AE, Pasqua LA, Tricoli V, Duarte M, Bishop DJ, et al. Effects of resistance
training on neuromuscular characteristics and pacing during 10-km running time trial. Eur J Appl Phy-
siol. 2015; doi: 10.1007/s00421-015-3130-z
Strength Training and Running Performance
PLOS ONE | DOI:10.1371/journal.pone.0150799
March 8, 2016
15 / 18
2.
Storen O, Helgerud J, Stoa EM, Hoff J. Maximal strength training improves running economy in dis-
tance runners. Med Sci Sports Exerc. 2008; 40(6): 1087–1092. doi: 10.1249/MSS.0b013e318168da2f
PMID: 18460997
3.
Paavolainen L, Hakkinen K, Hamalainen I, Nummela A, Rusko H. Explosive-strength training improves
5-km running time by improving running economy and muscle power. J Appl Physiol. 1999; 86(5):
1527–1533. PMID: 10233114
4.
Hickson RC, Dvorak BA, Gorostiaga EM, Kurowski TT, Foster C. Potential for strength and endurance
training to amplify endurance performance. J Appl Physiol. 1988; 65(5): 2285–2290. PMID: 3209573
5.
Barnes KR, Hopkins WG, McGuigan MR, Northuis ME, Kilding AE. Effects of resistance training on run-
ning economy and cross-country performance. Med Sci Sports Exerc. 2013; 45(12): 2322–2331. doi:
10.1249/MSS.0b013e31829af603 PMID: 23698241
6.
Sedano S, Marin PJ, Cuadrado G, Redondo JC. Concurrent training in elite male runners: the influence
of strength versus muscular endurance training on performance outcomes. J Strength Cond Res. 2013;
27(9): 2433–2443. doi: 10.1519/JSC.0b013e318280cc26 PMID: 23287831
7.
Ferrauti A, Bergermann M, Fernandez-Fernandez J. Effects of a concurrent strength and endurance
training on running performance and running economy in recreational marathon runners. J Strength
Cond Res. 2010; 24(10): 2770–2778. doi: 10.1519/JSC.0b013e3181d64e9c PMID: 20885197
8.
Roschel H, Barroso R, Tricoli V, Batista MA, Acquesta FM, Serrao JC, et al. Effects of strength training
associated with whole body vibration training on running economy and vertical stiffness. J Strength
Cond Res. 2015; doi: 10.1519/JSC.0000000000000857
9.
Bertuzzi R, Pasqua LA, Bueno S, Damasceno MV, Lima-Silva AE, Bishop D, et al. Strength-training
with whole-body vibration in long-distance runners: a randomized trial. Int J Sports Med. 2013; 34(10):
917–923. doi: 10.1055/s-0033-1333748 PMID: 23559412
10.
Kelly CM, Burnett AF, Newton MJ. The effect of strength training on three-kilometer performance in rec-
reational women endurance runners. J Strength Cond Res. 2008; 22(2): 396–403. doi: 10.1519/JSC.
0b013e318163534a PMID: 18550953
11.
Bassett DR, Howley ET. Limiting factors for maximum oxygen uptake and determinants of endurance
performance. Med Sci Sports Exerc. 2000; 32(1): 70–84. PMID: 10647532
12.
Millet GP, Jaouen B, Borrani F, Candau R. Effects of concurrent endurance and strength training on
running economy and.VO(2) kinetics. Med Sci Sports Exerc. 2002; 34(8): 1351–1359. PMID:
12165692
13.
Johnston RE, Quinn TJ, Kertzer R, Vroman NB. Strength training in female distance runners: Impact on
running economy. J Strength Cond Res. 1997; 11(4): 224–229.
14.
Guglielmo LG, Greco CC, Denadai BS. Effects of strength training on running economy. Int J Sports
Med. 2009; 30(1): 27–32. doi: 10.1055/s-2008-1038792 PMID: 18975259
15.
Spurrs RW, Murphy AJ, Watsford ML. The effect of plyometric training on distance running perfor-
mance. Eur J Appl Physiol. 2003; 89(1): 1–7. PMID: 12627298
16.
Roberts TJ, Azizi E. Flexible mechanisms: the diverse roles of biological springs in vertebrate move-
ment. J Exp Biol. 2011; 214(Pt 3): 353–361. doi: 10.1242/jeb.038588 PMID: 21228194
17.
Arampatzis A, De Monte G, Karamanidis K, Morey-Klapsing G, Stafilidis S, Bruggemann GP. Influence
of the muscle-tendon unit's mechanical and morphological properties on running economy. J Exp Biol.
2006; 209(Pt 17): 3345–3357. PMID: 16916971
18.
Reeves ND, Maganaris CN, Narici MV. Effect of strength training on human patella tendon mechanical
properties of older individuals. J Physiol. 2003; 548(Pt 3): 971–981. PMID: 12626673
19.
Seynnes OR, Erskine RM, Maganaris CN, Longo S, Simoneau EM, Grosset JF, et al. Training-induced
changes in structural and mechanical properties of the patellar tendon are related to muscle hypertro-
phy but not to strength gains. Journal of applied physiology. 2009; 107(2): 523–530. doi: 10.1152/
japplphysiol.00213.2009 PMID: 19478195
20.
Francesca PM, Giulia DI, Stefania C, Alessandro S, Gianluca V, Antonio LT. Concurrent strength and
endurance training effects on running economy in master endurance runners. J Strength Cond Res.
2012; doi: 10.1519/JSC.0b013e3182794485
21.
Magnusson SP, Hansen M, Langberg H, Miller B, Haraldsson B, Westh EK, et al. The adaptability of
tendon to loading differs in men and women. International journal of experimental pathology. 2007; 88
(4): 237–240. PMID: 17696904
22.
Folland JP, Williams AG. The adaptations to strength training: morphological and neurological contribu-
tions to increased strength. Sports Med. 2007; 37(2): 145–168. PMID: 17241104
23.
Bell GJ, Syrotuik D, Martin TP, Burnham R, Quinney HA. Effect of concurrent strength and endurance
training on skeletal muscle properties and hormone concentrations in humans. Eur J Appl Physiol.
2000; 81(5): 418–427. PMID: 10751104
Strength Training and Running Performance
PLOS ONE | DOI:10.1371/journal.pone.0150799
March 8, 2016
16 / 18
24.
Hather BM, Tesch PA, Buchanan P, Dudley GA. Influence of eccentric actions on skeletal muscle
adaptations to resistance training. Acta physiologica Scandinavica. 1991; 143(2): 177–185. PMID:
1835816
25.
Kraemer WJ, Patton JF, Gordon SE, Harman EA, Deschenes MR, Reynolds K, et al. Compatibility of
high-intensity strength and endurance training on hormonal and skeletal muscle adaptations. J Appl
Physiol. 1995; 78(3): 976–989. PMID: 7775344
26.
Ingjer F, Brodal P. Capillary supply of skeletal muscle fibers in untrained and endurance-trained
women. Eur J Appl Physiol Occup Physiol. 1978; 38(4): 291–299. PMID: 668683
27.
Brodal P, Ingjer F, Hermansen L. Capillary supply of skeletal muscle fibers in untrained and endurance-
trained men. The American journal of physiology. 1977; 232(6): H705–712. PMID: 879309
28.
Jeukendrup AE, Craig NP, Hawley JA. The bioenergetics of World Class Cycling. Journal of science
and medicine in sport / Sports Medicine Australia. 2000; 3(4): 414–433. PMID: 11235007
29.
Vikmoen O, Ellefsen S, Troen O, Hollan I, Hanestadhaugen M, Raastad T, et al. Strength training
improves cycling performance, fractional utilization of VO and cycling economy in female cyclists.
Scand J Med Sci Sports. 2015; doi: 10.1111/sms.12468
30.
Ronnestad BR, Hansen EA, Raastad T. In-season strength maintenance training increases well-
trained cyclists' performance. Eur J Appl Physiol. 2010; 110(6): 1269–1282. doi: 10.1007/s00421-010-
1622-4 PMID: 20799042
31.
Howley ET, Bassett DR Jr., Welch HG. Criteria for maximal oxygen uptake: review and commentary.
Med Sci Sports Exerc. 1995; 27(9): 1292–1301. PMID: 8531628
32.
Helland C, Bojsen-Moller J, Raastad T, Seynnes OR, Moltubakk MM, Jakobsen V, et al. Mechanical
properties of the patellar tendon in elite volleyball players with and without patellar tendinopathy. British
journal of sports medicine. 2013; 47(13): 862–868. doi: 10.1136/bjsports-2013-092275 PMID:
23833044
33.
Hopkins WG, Marshall SW, Batterham AM, Hanin J. Progressive statistics for studies in sports medi-
cine and exercise science. Med Sci Sports Exerc. 2009; 41(1): 3–13. doi: 10.1249/MSS.
0b013e31818cb278 PMID: 19092709
34.
Ronnestad BR, Hansen EA, Raastad T. Effect of heavy strength training on thigh muscle cross-sec-
tional area, performance determinants, and performance in well-trained cyclists. Eur J Appl Physiol.
2010; 108(5): 965–975. doi: 10.1007/s00421-009-1307-z PMID: 19960350
35.
Aagaard P, Andersen JL, Bennekou M, Larsson B, Olesen JL, Crameri R, et al. Effects of resistance
training on endurance capacity and muscle fiber composition in young top-level cyclists. Scand J Med
Sci Sports. 2011; 21(6): e298–307. doi: 10.1111/j.1600-0838.2010.01283.x PMID: 21362056
36.
Taipale RS, Mikkola J, Nummela A, Vesterinen V, Capostagno B, Walker S, et al. Strength training in
endurance runners. Int J Sports Med. 2010; 31(7): 468–476. doi: 10.1055/s-0029-1243639 PMID:
20432192
37.
Sjogaard G. Muscle morphology and metabolic potential in elite road cyclists during a season. Int J
Sports Med. 1984; 5(5): 250–254. PMID: 6500791
38.
Costill DL, Fink WJ, Pollock ML. Muscle fiber composition and enzyme activities of elite distance run-
ners. Medicine and science in sports. 1976; 8(2): 96–100. PMID: 957938
39.
Tricoli V, Lamas L, Carnevale R, Ugrinowitsch C. Short-term effects on lower-body functional power
development: weightlifting vs. vertical jump training programs. J Strength Cond Res. 2005; 19(2): 433–
437. PMID: 15903387
40.
Wilson GJ, Newton RU, Murphy AJ, Humphries BJ. The optimal training load for the development of
dynamic athletic performance. Med Sci Sports Exerc. 1993; 25(11): 1279–1286. PMID: 8289617
41.
Ronnestad BR, Kvamme NH, Sunde A, Raastad T. Short-term effects of strength and plyometric train-
ing on sprint and jump performance in professional soccer players. J Strength Cond Res. 2008; 22(3):
773–780. doi: 10.1519/JSC.0b013e31816a5e86 PMID: 18438241
42.
Kongsgaard M, Reitelseder S, Pedersen TG, Holm L, Aagaard P, Kjaer M, et al. Region specific patellar
tendon hypertrophy in humans following resistance training. Acta physiologica. 2007; 191(2): 111–121.
PMID: 17524067
43.
Kubo K, Kanehisa H, Ito M, Fukunaga T. Effects of isometric training on the elasticity of human tendon
structures in vivo. J Appl Physiol. 2001; 91(1): 26–32. PMID: 11408409
44.
Kubo K, Komuro T, Ishiguro N, Tsunoda N, Sato Y, Ishii N, et al. Effects of low-load resistance training
with vascular occlusion on the mechanical properties of muscle and tendon. Journal of applied biome-
chanics. 2006; 22(2): 112–119. PMID: 16871002
45.
Kubo K, Yata H, Kanehisa H, Fukunaga T. Effects of isometric squat training on the tendon stiffness
and jump performance. Eur J Appl Physiol. 2006; 96(3): 305–314. PMID: 16328192
Strength Training and Running Performance
PLOS ONE | DOI:10.1371/journal.pone.0150799
March 8, 2016
17 / 18
46.
Saunders PU, Telford RD, Pyne DB, Peltola EM, Cunningham RB, Gore CJ, et al. Short-term plyomet-
ric training improves running economy in highly trained middle and long distance runners. J Strength
Cond Res. 2006; 20(4): 947–954. PMID: 17149987
47.
Mikkola J, Rusko H, Nummela A, Pollari T, Hakkinen K. Concurrent endurance and explosive type
strength training improves neuromuscular and anaerobic characteristics in young distance runners. Int
J Sports Med. 2007; 28(7): 602–611. PMID: 17373596
48.
Hoff J, Helgerud J. Maximal strength training enhances running economy and aerobic endurance per-
formance In: Hoff J, Helgerud J, editors. Football (Soccer) New Developments in Physical Training
Research. Trondheim: Norwegian University of Science and Technology, Dept. of Physiology and Bio-
medical Engineering; 2003. p. 37–53.
49.
Noakes TD, Myburgh KH, Schall R. Peak treadmill running velocity during the VO2 max test predicts
running performance. Journal of sports sciences. 1990; 8(1): 35–45. PMID: 2359150
50.
Hill DW, Rowell AL. Running velocity at VO2max. Med Sci Sports Exerc. 1996; 28(1): 114–119. PMID:
8775363
Strength Training and Running Performance
PLOS ONE | DOI:10.1371/journal.pone.0150799
March 8, 2016
18 / 18
| Effects of Heavy Strength Training on Running Performance and Determinants of Running Performance in Female Endurance Athletes. | 03-08-2016 | Vikmoen, Olav,Raastad, Truls,Seynnes, Olivier,Bergstrøm, Kristoffer,Ellefsen, Stian,Rønnestad, Bent R | eng |
PMC3299590 | STUDY PROTOCOL
Open Access
Effect of running therapy on depression
(EFFORT-D). Design of a randomised controlled
trial in adult patients [ISRCTN 1894]
Frank R Kruisdijk1,3*, Ingrid JM Hendriksen2,3, Erwin CPM Tak2,3, Aartjan TF Beekman4,5 and
Marijke Hopman-Rock2,3,5
Abstract
Background: The societal and personal burden of depressive illness is considerable. Despite the developments in
treatment strategies, the effectiveness of both medication and psychotherapy is not ideal. Physical activity,
including exercise, is a relatively cheap and non-harmful lifestyle intervention which lacks the side-effects of
medication and does not require the introspective ability necessary for most psychotherapies. Several cohort
studies and randomised controlled trials (RCTs) have been performed to establish the effect of physical activity on
prevention and remission of depressive illness. However, recent meta-analysis’s of all RCTs in this area showed
conflicting results. The objective of the present article is to describe the design of a RCT examining the effect of
exercise on depressive patients.
Methods/Design: The EFFect Of Running Therapy on Depression in adults (EFFORT-D) is a RCT, studying the
effectiveness of exercise therapy (running therapy (RT) or Nordic walking (NW)) on depression in adults, in addition
to usual care. The study population consists of patients with depressive disorder, Hamilton Rating Scale for
Depression (HRSD) ≥ 14, recruited from specialised mental health care. The experimental group receives the
exercise intervention besides treatment as usual, the control group receives treatment as usual. The intervention
program is a group-based, 1 h session, two times a week for 6 months and of increasing intensity. The control
group only performs low intensive non-aerobic exercises. Measurements are performed at inclusion and at 3,6 and
12 months.
Primary outcome measure is reduction in depressive symptoms measured by the HRSD. Cardio-respiratory fitness is
measured using a sub maximal cycling test, biometric information is gathered and blood samples are collected for
metabolic parameters. Also, co-morbidity with pain, anxiety and personality traits is studied, as well as quality of life
and cost-effectiveness.
Discussion: Exercise in depression can be used as a standalone or as an add-on intervention. In specialised
mental health care, chronic forms of depression, co-morbid anxiety or physical complaints and treatment
resistance are common. An add-on strategy therefore seems the best choice. This is the first high quality large
trial into the effectiveness of exercise as an add-on treatment for depression in adult patients in specialised
mental health care.
Trial registration: Netherlands Trial Register (NTR): NTR1894
* Correspondence: [email protected]
1GGZ Centraal Centers for Mental Health, Symfora-Meander Centre for
Psychiatry, Utrechtseweg 266, 3800 DB Amersfoort, The Netherlands
Full list of author information is available at the end of the article
Kruisdijk et al. BMC Public Health 2012, 12:50
http://www.biomedcentral.com/1471-2458/12/50
© 2012 Kruisdijk et al; BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons
Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in
any medium, provided the original work is properly cited.
Background
Depression is a common disorder. The lifetime preva-
lence of depressive disorders in Dutch adults is 19% [1].
Recurrence of symptoms occurs in an unfavourable and
fluctuating course in 44% and a severe chronic course in
32% of the patients. Depressive disorders obviously have
negative effects on wellbeing and daily personal and
professional functioning.
Antidepressants and psychotherapy
Treatment frequently includes prescription of antide-
pressants. However, the effectiveness of such treatment
may be limited because of poor compliance and limited
effectiveness. A recent analysis of data of the US Food
and Drugs Administration, showed relatively small drug-
placebo differences in antidepressant efficacy [2]. Other
disadvantages of antidepressants are unpleasant side
effects and increased risk of hypertension in depressed
patients combined with Diabetes Mellitus type II (DM
II) [3,4].
A systematic review of randomised controlled trials
into effectiveness and cost-effectiveness of brief psycho-
logical treatment for depression showed that some
forms,
especially
cognitive-behavioural
based
approaches, were beneficial in the treatment of outpati-
ents [5]. A meta-analysis, regarding a mostly adult popu-
lation, found a favourable outcome for a combination of
psychotherapy and pharmacotherapy compared to psy-
chological treatment alone for depression in the short-
term [6]. A recent meta-analysis examined whether the
quality of the studies investigating psychotherapy for
adult depression was associated with the effect-size
found in these studies [7]. It showed that the effects of
psychotherapy for adult depression have been overesti-
mated in meta-analytical studies. The authors stated
that the effects of psychotherapy are much smaller than
is assumed.
The traditional treatment strategies, medication and
psychotherapy, are still not ideal and a new intervention,
exercise, focusing on a different psychopathological
mechanism of depression is needed.
Exercise
Exercise is a potential alternative low-cost therapy, but
more studies are needed to define the place of exercise
in a stepped care program for depression [8]. Exercise is
relatively safe and has less negative side effects than
antidepressants. Obviously, exercise has also many bene-
ficial effects on physical health [9,10], and is expected to
have additional advantages in depressed patients who
suffer from a combination of mental and physical pro-
blems, such as pain complaints or increased risk of car-
diac morbidity. Exercise is suitable for most individuals,
participating in an exercise program promotes social
integration and successful adaptation can increase self-
esteem. Finally, exercise may reduce the negative side
effects of antidepressants, such as fatigue, thus increas-
ing compliance in medication use.
Scientific evidence for the efficacy of exercise
Recent reviews and meta-analyses suggest that exercise
leads to improvements in depressive symptoms [11,12].
Following the cumulative evidence from prospective
cohort studies and RCTs, exercise has proven protective
benefits for several aspects of mental health in general
and for symptoms of depression in particular. However,
study results on the curative effect on a present depres-
sion are conflicting. A recent Cochrane meta-analysis
[13], updating an earlier systematic review in 2001 by
Lawlor and Hopker [14], concluded that the statistically
weak findings didn’t support the efficacy of exercise in
the treatment of depression. Methodological weaknesses
of the trials were identified, including the lack of treat-
ment concealment, intention to treat analysis and a clin-
ical interview to confirm the diagnosis of depression.
Rethorst et al. [15] argue that the earlier mentioned
meta-analysis of Lawlor and Hopker suffered from
incomplete research data and a lack of moderating vari-
able analysis. The meta-analysis by Rethorst et al., with
58 RCTs included, calculated an overall effect size of
-0.80 for the effects of exercise on depression. The
recommendation of both Lawlor et al. and Rethorst et
al. is that further, conclusive research is still needed into
the use of exercise for depression. Exercise as an adjunct
to recognised treatments, such as psychotherapy and/or
medication, should also be studied. Furthermore, follow-
up research is needed that examines the sustainability of
the effects after exercise therapy has ended.
Very recently, Krogh et al. [16] performed a systematic
review and meta-analysis in which only 13 trials were
selected because of the stringent selection criteria used
(i.e. recording clinical depression according to any diag-
nostic system, adequate allocation concealment, blinded
outcome and intention to treat analysis). Besides an
inverse association between duration of intervention and
the magnitude of the association of exercise with
depression, the effect size was 0.40 for the pooled effect
sizes of 13 studies and 0.19 for the selected three high
quality design studies. The authors concluded that large
high quality design studies are required. The recommen-
dations from the three meta-analysis cited above are
integrated in the design of the EFFORT-D study.
Besides reduction of depressive symptoms, exercise
can be useful for other treatment indications, such as
cardiovascular condition, metabolic syndrome, pain
modulation and the immune system. Depression is
Kruisdijk et al. BMC Public Health 2012, 12:50
http://www.biomedcentral.com/1471-2458/12/50
Page 2 of 9
associated with cardiovascular disease. It is an estab-
lished risk factor for mortality after acute myocardial
infarction. A recent meta-analysis of cohort studies also
shows the aetiological and prognostic effects of depres-
sion on coronary heart disease (CHD) [17]: the pooled
relative risk of depression on future coronary heart dis-
ease was 1.80. Another meta-analysis shows an overall
relative risk of dying in depressed subjects of 1.80 com-
pared to non-depressed subjects, with no major differ-
ences between men and women [18]. The increased
relative risk on mortality was also found in subclinical
forms of depression.
These findings suggest that exercise may also be pre-
scribed to improve the cardiovascular condition and
prevent future heart disease. Depressed patients should
also be more closely monitored regarding cardiovascular
parameters.
The body of evidence showing an association between
depression and obesity and metabolic deregulation is
growing. Recent meta-analysis of cross-sectional studies
in the general population found a significant positive
association between depression and obesity, affecting
women more than men. A reciprocal risk relation
between obesity and depression [19,20], and an U-
shaped, non-linear trend in the association between
BMI and depression was found [21]. The Dutch NESDA
study [22] shows an association between severity of
depression and unfavourable cholesterol high-hdl/low-
ldl concentration [23]. This tendency for a higher risk of
depression on metabolic syndrome, in which various
cardiovascular risk factors appear, emphasizes the need
of careful screening on this parameters [24], and exer-
cise could positively influence these parameters.
A recent meta-analysis of studies into cytokines in
depressed patients [25] showed that depression and an
activation of the immune-system do co-exist. Both
tumornecrosisfactor-a (TNF-a) and interleukine-6-con-
centration (IL-6) were significantly raised and C-reactive
protein (CRP) was linearly associated with several con-
ventional risk factors and inflammatory markers [26].
The relevance of CRP however still remains controver-
sial [27]. Therefore, measurement of CRP blood levels in
an intervention exercise study could contribute to a bet-
ter understanding of the clinical importance in relation
to heart condition.
The reciprocal relation between pain and depression
has been reported in several studies where Major
Depressive Disorder (MDD) was associated with chronic
pain (> 6 months) and nearly 50% of patients had at
least one chronic painful physical condition [28,29]. The
duration of depressive symptoms was found to be pro-
longed in the presence of chronic pain. The therapeutic
prognosis of co-morbid pain and depression is poor and
the pain-depression association was found to be
stronger in men than in women and in older adults
compared to younger. Productivity was decreased when
depression and pain existed in a co-morbid pattern [30].
Quality of life and cost-effectiveness
Systematic evaluation of cost-effectiveness and quality of
life, including life-events in the treatment of depression,
may contribute to the development of more evidence
based care models [31], for instance by disease
management.
A population sample study showed that, in patients
with a remitted MDD, the quality of life was lower than
in the general population. A higher depressive severity
was associated with a lower quality of life. Ten
Doesschate et al. argue that, even in depression in
remission, attention is needed for the quality of life, and
above that, residual symptoms must be treated aggres-
sively to achieve a higher quality of life [32].
Summarized aims and hypothesis of EFFORT-D
The main objective of the study is to assess the effec-
tiveness of exercise therapy (running therapy (RT) or
Nordic walking (NW)), in addition to usual care, on
depression in adult patients. We hypothesize that adding
exercise therapy to usual care will result in a larger
reduction in depressive symptoms (as measured with
the Hamilton Rating Scale of Depression, (HRSD)) dur-
ing a 6 month treatment program as well as at 6
months follow-up, compared to usual care without exer-
cise therapy.
Secondary aims are to assess the effectiveness of exer-
cise therapy (RT or NW) on the following outcome
measures: 1) Cardiovascular and metabolic risk para-
meters as fitness (VO2 max, grip strength), BMI, waist
circumference, body composition, blood pressure, fasting
blood glucose, cholesterol HDL/LDL ratio and CRP, 2)
Co-morbid symptoms of anxiety and pain 3) Quality of
life, and 4) Cost-effectiveness.
To assess whether there are subgroups of patients who
show a larger or less effect of exercise therapy on
depression, subgroup analysis will be conducted differ-
entiated on demographic variables (age, sex) and level
and type of depression based on the HRSD, personality
traits and level of fitness.
Methods/Design
EFFORT-D is a randomized controlled trial, designed
along the Consort-statement guidelines [33], in which
outpatients as well as hospitalised depressed patients
will be included and randomized in two groups: a con-
trol group or an intervention group (see flow chart in
Figure 1). The control group outpatients receive usual
care (i.e. anti-depressive medication and/or cognitive
and/or interpersonal therapy). The control group
Kruisdijk et al. BMC Public Health 2012, 12:50
http://www.biomedcentral.com/1471-2458/12/50
Page 3 of 9
inpatients have their usual treatment program (consist-
ing of pharmacotherapy and/or sociotherapy, psy-
chotherapy, psycho-education and indicated nonverbal
therapies) and are allowed to exercise at low intensity as
part of their daily program.
The intervention group receives 6 months supervised
exercise therapy for 1 h a week and is instructed to
train unsupervised for 1 h a week during this period
(combined about 40 exercise sessions during the inter-
vention period). Both training sessions follow an indivi-
dualised intervention protocol and are in addition to the
usual care program.
Included and randomised patients in the intervention
group are invited to take part in Running Therapy (RT).
Patients will only be referred to Nordic Walking (NW)
in case of clear medical contra-indications against RT,
such as muscular-skeleton problems, or if patients have
a
strong
dislike
for
running
which
obstructs
participation and compliance. The study will run for 27
months, with an 18 months inclusion period during
which patients are recruited and randomized. There are
four measurements for each patient: at baseline (T0),
halfway the 6 month intervention period (T3), at the
end of the intervention period (T6), and at follow up, 12
months after baseline (T12). In Table 1 the timetable is
shown.
Study population
The study population consists of adult patients diag-
nosed by a clinician with a depression or bipolar disor-
der with depressive mood, who are or will be treated in
GGZ Centraal Mental Hospitals or Symfora-Meander
Hospital. Patients aged between 18-65, with a DSM-IV
diagnoses of unipolar, bipolar depression or seasonal
depression not responding to light therapy (10 sessions
of 1 h), a baseline Hamilton Rating Scale of Depression
Psychiatrist diagnosis of depression
In -or-outpatients
Screening by
research-assistant
Check inclusion/
exclusion criteria
HRSD
Biometric values = BV
Cycle test = CT
Electronic Questionnary = EQ
Blood samples = BS
Blinded randomisation
by random envelopes
Control Group
Treatment as usual
Psychotherapy
Medication
Lifestyle
psychoeducation
Intervention Group
Treatment as usual
Psychotherapy
Medication
Lifestyle
psychoeducation
+
20 sessions 1 hr. RT/NW supervised
20 sessions 1 hr. RT/NW individually
HR freq. Polar
Informed consent
+
Inclusion in EFF-D
T= 0
T = 3 BV,CT,EQ
T = 6 BV,CT,EQ,BS
T = 12 BV,CT,EQ
End of
Intervention period
Figure 1 Flow chart.
Kruisdijk et al. BMC Public Health 2012, 12:50
http://www.biomedcentral.com/1471-2458/12/50
Page 4 of 9
Table 1 Events and time table
Intervention period
Follow-up period
Procedure
Source
Person
No. of items
Duration (min.)
T0
T3
T6
T12
Written informed consent
letter
patient
X
Demographics1
patient file
research assistant
X
Depression
Depression history2
patient file
research assistant
2
X
HRSD3
interview
blind rater
17
20
X
X
X
X
IDS-SR4
questionnaire
patient
30
10
X
X
X
X
Bearableness (VAS)
questionnaire
patient
1
X
X
X
X
Metabolic syndrome
Length
physical test
research assistant
1
X
Weight
physical test
research assistant
1
X
X
X
X
Waist circumference
physical test
research assistant
1
X
X
X
X
Blood pressure
physical test
research assistant
1
X
X
X
X
Smoking/alchohol intake
questionnaire
patient
1
X
X
X
X
Laboratory assesment6
physical test
laboratory
2
X
X
X
X
Quality of Life
WHO-DAS7
questionnaire
patient
36
X
X
X
X
Pain
GCPS8
questionnaire
patient
7
X
X
X
X
Bearableness (VAS)
questionnaire
patient
1
X
X
X
X
Anxiety
BAI9
questionnaire
patient
21
X
X
X
X
Cost effectiveness
TIC-P10
quest./pnt file
pnt/research assistant
26
X
X
X
Euroqol
questionnaire
patient
5
X
X
X
Subjective health (VAS)
questionnaire
patient
1
X
X
X
Fitness
Submaximal cycle test
physical test
research assistant
10
X
X
X
X
Grip strength
physical test
research assistant
2
X
X
X
X
Exercise intensity (HR)11
physical test
exercise instructor
during training sessions
Personality
NEO PI-R12
questionnaire
patient
60
X
Physical activity
SQUASH13
questionnaire
patient
12
X
X
X
X
Additional Measures
LEQ14
questionnaire
patient
12
X
X
X
X
Compliance
registration
exercise instructor
during training sessions
POMS15
questionnaire
patient
32
during training sessions
Evaluation Intervention16
questionnaire
patient
X
1: Date of birth, sex, education, ethnicity, income, professional status, living situation, marital status
2: Number and duration of former episodes, duration current episode
3: Hamilton Rating Scale of Depression (or HAMD)
4: Inventory of Depressive Symptomatology - Self Report
6: fasting glucose, triglycerides, total cholesterol, HDL- cholesterol, cholesterol/HDL-ratio, creatinine, MDRD
7: World Health Organisation- Disability Assessment Schedule
8: Graded Chronic Pain Scale
9: Beck Anxiety Inventory
10: Trimbos/iMTA questionnaire for Costs associated with Psychiatric Illness
11: Polar RS 800 CX Run
12: Revised NEO Personality Inventory
13: Short QUestionnaire to ASses Health enhancing physical activity
14: Life Events Questionnaire
15: Profile of Mood State
16: Only for patients who participated in the intervention
Kruisdijk et al. BMC Public Health 2012, 12:50
http://www.biomedcentral.com/1471-2458/12/50
Page 5 of 9
(HRSD) score ≥14 and (will be) treated for depression,
are included.
Criteria for exclusion are: a depression as part of a
psychotic disorder, schizophrenia, schizoaffective disor-
der or obsessive compulsive disorder, anxiety disorder
as primary diagnosis, patients in long stay facilities
(including day-care) or with complex pathology and
treatment resistant depression (inpatients, treated by
protocol more than 6 months with no remission);
patients with significant cardiovascular disease or other
medical conditions as contra-indication for exercise
therapy, walking and/or running such as joint and hip
pathology; alcohol/drugs dependence as a primary diag-
nosis, pregnancy, high suicide risk with treatment on a
closed ward, or already being physically active on a reg-
ular basis (2-3 times a week on a high-intensity).
Sample size
Following earlier RCTs [34] it is expected that patients
in the usual care group (controls) will respond with a
mean reduction in HRSD of six points. Adding exercise
to usual care is expected to result in a decline of at least
eight points in HRSD score (thus two extra points). To
detect this difference, with an a (two-tailed) of 5% and a
power (1-b) of 80%, using two equal groups and a stan-
dard deviation of 5 points, 100 patients are needed in
each group. Taking 30% drop-out into account, 140
patients have to be included in each group.
Procedures and study instruments
Names of eligible patients with their registration number
of the electronic patient file (EPD) are provided by the
diagnosing psychiatrists to the research assistant, who
will make a first check on inclusion and exclusion cri-
teria, informs the participants and asks them to join the
study. Written informed consent will be obtained
according to prevailing legal requirements before the
start of the study. Eligible patients, HDRS ≥ 14 as pri-
mary outcome measure, perform the Åstrand submaxi-
mal cycling test, physical measures and fill out the
questionnaire, after which the participants are rando-
mized. All outcome parameters measured during base-
line will be repeated three, six and 12 months after
baseline, except for the blood samples (only at T0 and
T6). Height is measured according to protocol (Seca
214, Hamburg, Germany) and a bio-impedance scale is
used to measure weight and body composition (Omron
HBF-510, Omron Healthcare Europe BV, The Nether-
lands). Waist circumference is measured twice with a
tape measure (Seca 201, Hamburg, Germany) at the
midpoint between the lower border of the ribs and the
upper border of the pelvis. Systolic and diastolic blood
pressure are registered twice at rest, using an electronic
blood pressure meter (Omron M6 comfort, Omron
Healthcare Europe BV, The Netherlands) with an ade-
quate cuff size. Grip-strength is tested according to pro-
tocol using a hydraulic hand dynamometer (Jamar
J00105, Sammons Preston Rolyan, Bolingbrook, USA).
The submaximal Åstrand test [35] will be performed on
a stationary bicycle ergometer (Examiner, Lode BV, The
Netherlands) and the mean heart rate of the last 2 min
of the test will be used to estimate the VO2max. Heart
rate during this test is registered by a heart rate monitor
(Polar RS 800, Electro Oy, Finland).
Questionnaires
At each measurement moment the participant will be
asked to fill out a digital questionnaire containing the
following instruments:
Demographics and personal life events
Socio-demographic data are collected using standard
questions on age, sex, marital status, ethnicity and
household composition. Socio-economic variables
include highest education and income. Personal history
is evaluated by the Life Events Questionnaire (LEQ), a
12-item inventory-type questionnaire in which subjects
mark the exposure to negative life events such as unem-
ployment, separation from a partner and death of a
close family member which have occurred in the past
year [36].
Mental and physical health and its consequences
The Hamilton Rating Scale of Depression (HRSD) mea-
sures depression with a 17-item list performed by
trained interviewers [37]., using the Dutch translation of
the version of Bech et al. [38], in which the items of
depressive symptoms are extensively operationalised and
this version is often used in international research [39].
The Inventory of Depressive Symptomatology - Self-
Reported (IDS-SR) measures the severity of depression
with a 30-item self-report list [40]. It has good respon-
siveness to change and is more sensitive for atypical
depressive criteria than the HRSD. History of depression
is evaluated by a single question into the number and
duration of depressive episodes for which treatment was
necessary.
Bearableness of depression is measured with a visual
analogue scale (VAS) ranging from 0 (very unbearable)
to 100 (very well bearable). Anxiety is measured by the
Beck Anxiety Inventory (BAI), a 21-item multiple-choice
self report inventory that measures the severity of gener-
alized anxiety and panic symptoms in adults and adoles-
cents [41].
Pain complaints are evaluated with the Graded
Chronic Pain Scale (GCPS) [42], a 7-item scale measur-
ing aspects of pain, physical ability and social interfer-
ence, resulting in a 5-class hierarchical scale ranging
from 0 (no pain problem) to IV (high disability/severely
limiting). Next to the GCPS, bearableness of pain is
Kruisdijk et al. BMC Public Health 2012, 12:50
http://www.biomedcentral.com/1471-2458/12/50
Page 6 of 9
evaluated with a VAS-scale ranging from 0 (very
unbearable) to 100 (very well bearable).
Other secondary outcomes are disability during the
last 30-days associated with both physical and mental
problems and is measured by a shortened version of the
World Health Organisation - Disability Assessment
Schedule II (WHO-DAS-II) [43], resulting in a disability
score ranging from 0-100 with a higher score reflecting
greater disability. Quality of life data are collected using
the EQ5D [44], a standardized instrument for describing
and valuing health related quality of life.
Subjective health is evaluated by a visual analogue
scale ranging from 0 (the worst imaginable health condi-
tion) to 100 (the best imaginable health condition).
Health care use and work productivity are evaluated
by the Trimbos/iMTA Questionnaire for Costs asso-
ciated with Psychiatric Illness (TIC-P), a 29-item list
which focuses on establishing costs related to loss of
productivity at work and health care utilization [45].
Variables expected to modify the effect of the inter-
vention Personality, as a possible effect modifier, is mea-
sured at baseline by the NEO-PI-R, a 60-item validated
questionnaire measuring the five domains of personality
including neuroticism, extraversion, agreeableness, con-
scientiousness and openness to experience [46].
Confounders: Lifestyle behaviours Self-reported level of
physical activity is assessed by means of the validated
Short QUestionnaire to ASsess Health enhancing physi-
cal activity (SQUASH), a 12-item questionnaire which
evaluates the frequency and duration of physical activ-
ities in the domains of work, domestic and leisure time
[47]. Tobacco and alcohol intake is measured with a
standard single question on the frequency of use per
day.
Additional measures in the intervention group
The Profile of Mood States (POMS) registers partici-
pants’ mood after a running session for in total three
times during the intervention period (at the beginning,
halfway and at the end). The Dutch shortened version
of the POMS [48] consists of 32 items divided over
seven subscales including tension, depression, anger,
fatigue, vigour, positive and negative affect.
At regular intervals participants score their exertion
on a Borg-scale ranging from 6 (very, very light) to 20
(very, very severe) [49]. At the end of the intervention
period, or when participants dropped out of the RT or
NW therapy, a short questionnaire is administered eval-
uating their satisfaction and experience with the
intervention.
Randomisation, blinding and treatment allocation
Randomisation takes place at every location separately.
This way, every location will have an equal distribution
of participants between the intervention and control
group. The SPSS random generator (SPSS version 14.0)
[50] will be used to allocate patients. Ten closed envel-
opes with allocation numbers are presented to the parti-
cipants. They choose an envelope, after which the
research assistant tells the patients in which arm of the
study they are included. Evaluators of the main outcome
measure (HRSD) are blinded for group allocation and
are trained regularly for inter-rater reliability. All other
measures will be evaluated by a research assistant, who
is not blinded for group allocation.
Exercise intervention
The exercise sessions will take place twice a week (40
sessions in total): once a week a supervised group ses-
sion is offered and once a week the patient does an
individual training, with clear instructions beforehand
and an evaluation at the beginning of the next super-
vised session. Each supervised session, in which the
trainers are working according to a standardized proto-
col, lasts one hour, of which 30 min are spent running
(RT)/Nordic walking (NW). The remaining time is
spent on warming-up and cooling-down. Each patient
follows an individualised intervention protocol with a
gradually increasing training intensity. The goal is to
achieve a 30 min period of continuous running in the
last sessions (two times a week 30 min continuous
aerobic exercise at least 60% of the maximal heart
rate). The NW program follows a comparable progres-
sive schedule with increased time spent Nordic walking
with high intensity.
Intensity in RT as in NW is monitored by the instruc-
tor during every supervised session by counting the
heart rate and three times during the intervention per-
iod by electronic registration (Polar RS 400, Electro Oy,
Finland). The control group receives usual care for
depression in accordance with the revised Dutch guide-
line [51] and are advised to exercise regularly. Hospita-
lised and day-care patients in the control group are
supposed not to participate in organised high intensity
aerobic exercise during the intervention period. Only
low-intensity activity psycho-motor therapy is allowed.
Compliance and withdrawal
In order to improve compliance during the intervention
period, a protocol will be followed concerning missed
exercise therapy sessions by participants. This protocol
includes: 1) active approach by the exercise instructor in
case of no show, and 2) encouragement of other partici-
pants to contact each other in case of no show. Partici-
pants can withdraw at any time for any reason without
any consequences. Also, the investigator can decide to
withdraw a participant from the study for urgent medi-
cal reasons. Participants who withdraw from the inter-
vention will be asked the reason(s) for drop-out but will
Kruisdijk et al. BMC Public Health 2012, 12:50
http://www.biomedcentral.com/1471-2458/12/50
Page 7 of 9
be retained in the study for the intention to treat
analysis.
Statistical analyses
Comparability of the intervention and control groups
will be examined for the baseline measurements. If
necessary, analyses will be adjusted for baseline differ-
ences. The primary analysis of the data set will be
according to the ‘intention to treat’ principle. A second-
ary ‘per protocol’ analysis will be done taking into
account the level of compliance and the amount of exer-
cise during the intervention period. Usual daily physical
activity, tobacco smoking and alcohol intake will be
treated as confounders.
Differences in remission rates (and other categorical
outcomes) between the experimental groups are exam-
ined by contingency table Chi-square statistics. Differ-
ences in mean scores on continuous outcomes (e.g.
HRSD) between the intervention groups are examined
by analysis-of-variance.
Ethical principles and safety
The study has been designed and will be carried out in
accordance with the principles of the Helsinki Declara-
tion (Edinburgh, Scotland Amendment, October 2000).
The study protocol has been approved by the Medical
ethical committee for mental health (Metigg Kamer
Noord).
Discussion
The aim of this study is to investigate the effectiveness
of aerobic exercise therapy (RT or NW) on depression
in adult patients in addition to usual care. Also, the
effect of physical exercise on frequently existing co-mor-
bid diseases or risk factors for such disorders as meta-
bolic syndrome is an objective of this study.
This is the first well conducted add-on randomised
controlled high-quality trial into the effect of aerobic
exercise on depression with a correct randomisation
procedure, blinded outcome assessment, intention to
treat analysis, study into cost-effectiveness, quality of life
and long-term follow-up as was advised in earlier publi-
cations. EFFORT-D can therefore contribute to stronger
evidence for this type of intervention, which can result
in more specified recommendations for clinical practice.
Another strength of this study is the fact that also
severely clinically depressed patients, who are mostly
excluded in other studies, will be included. A relatively
greater effect of exercise in this subgroup of severely
depressed patients is possible. A challenge of the study
lies in the motivational techniques needed to exercise
with depressed patients, which is proven to be difficult
[52] and it will take a serious effort not to exceed the
calculated 30% drop-out patients in the intervention
group. Because this study is supported by a strong
hypothesis and minimal negative side effects are
expected, one-tailed statistical analysis is also possible if
the large number of included participants can’t be
reached within the planned time. Furthermore, the
diversity of outcome measures makes it possible to dis-
tinguish more explicitly those depressed patients that
could benefit most from exercise.
Acknowledgements
The former “Open Ankh Foundation” (since March 2008 “Zorgcoöperatie
Nederland”) funded part of this project, together with the Symfora Group of
Mental Hospitals (January 2010 “GGZ Centraal”), and TNO (Netherlands
Organisation for Applied Scientific Research).
The authors would also like to thank Professor B.W. Penninx and Professor E.
de Geus for their contribution in the design discussion and
recommendations for biometric diagnostics.
Author details
1GGZ Centraal Centers for Mental Health, Symfora-Meander Centre for
Psychiatry, Utrechtseweg 266, 3800 DB Amersfoort, The Netherlands. 2TNO
Expert Center Life Style, Wassenaarseweg 56, 2333 AL Leiden, The
Netherlands. 3Body@Work, Research Center Physical Activity, Work and
Health, TNO-VUmc, Van der Boechhorststraat 7, 1081 BT Amsterdam, The
Netherlands. 4Department of Psychiatry, VU University Medical Centre, A.J.
Ernststraat 887, 1081 HL Amsterdam, The Netherlands. 5The EMGO Institute
for Health and Care Research (EMGO+), VU University Medical Centre, Van
der Boechhorststraat 7, 1081 BT Amsterdam, The Netherlands.
Authors’ contributions
FRK is principle investigator, psychiatrist and project leader of the project in
GGZ Centraal and Symfora-Meander Hospitals and drafted the manuscript,
IH is project leader for the study at Body@Work, designed the study and has
been involved in drafting the manuscript, ET was involved in the study
design, organizing the data and commented the manuscript, AJB was
involved in the study design and revised the manuscript critically and gave
approval for publication, MHR was involved in the study design, revised the
manuscript critically and gave final approval of the version to be published.
Competing interests
The authors declare that they have no competing interests.
Received: 20 December 2011 Accepted: 19 January 2012
Published: 19 January 2012
References
1.
Bijl RV, Ravelli A, van Zessen G: Prevalence of psychiatric disorder in the
general population: results of The Netherlands Mental Health Survey
and Incidence Study (Nemesis). Soc Psychiatry Psychiatr Epidemiol 1998,
33:587-595.
2.
Kirsch I, Deacon BJ, Huedo-Medina TB, Scoboria A, Moore TJ, Johnson BT:
Initial severity and antidepressant benefits: a meta-analysis of data
submitted to the Food and Drug Administration. PLoS Med 2008, 5(2):e45.
3.
Licht CM, de Geus EJ, Seldenrijk A, van Hout HP, Zitman FG, van Dyck R,
Penninx BW: Depression is associated with decreased blood pressure,
but antidepressant use increases the risk for hypertension. Hypertension
2009, 53(4):631-638.
4.
Mezuk B, Eaton WW, Albrecht S, Golden SH: Depression and type 2
diabetes over the lifespan: a meta-analysis. Diabetes Care 2008,
31(12):2383-2390.
5.
Churchill R, Hunot V, Corney R, Knapp M, McGuire H, Tylee A, Wessely S: A
systematic review of controlled trials of the effectiveness and cost-
effectiveness of brief psychological treatments for depression. Health
Technol Assess 2001, 5(35):1-173.
6.
Cuijpers P, Dekker J, Hollon SD, Andersson G: Adding psychotherapy to
pharmacotherapy in the treatment of depressive disorders in adults: a
meta-analysis. J Clin Psychiatry 2009, 70(9):1219-1229.
Kruisdijk et al. BMC Public Health 2012, 12:50
http://www.biomedcentral.com/1471-2458/12/50
Page 8 of 9
7.
Cuijpers P, van Straten A, Bohlmeijer E, Hollon SD, Andersson G: The effects
of psychotherapy for adult depression are overestimated: a meta-
analysis of study quality and effect size. Psychol Med 2010, 40(2):211-223.
8.
Stammes R, Spijker J: Physical training to treat depression. Tijdschr
Psychiatr 2009, 51(11):821-830.
9.
Blumenthal JA, Sherwood A, Babyak MA, Watkins LL, Waugh R,
Georgiades A, Bacon SL, Hayano J, Coleman RE, Hinderliter A: Effects of
exercise and stress management training on markers of cardiovascular
risk in patients with ischemic heart disease: a randomized controlled
trial. JAMA 2005, 293(13):1626-1634.
10.
Warburton DE, Nicol CW, Bredin SS: Health benefits of physical activity:
the evidence. CMAJ 2006, 174(6):801-809.
11.
Craft LL, Perna FM: The Benefits of Exercise for the Clinically Depressed.
Prim Care Companion J Clin Psychiatry 2004, 6(3):104-111.
12.
Physical Activity Guidelines Advisory Committee: Physical Activity
Guidelines Advisory Committee Report. Washington, DC U.S. Department
of Health and Human Services; 2008:G-8:1-58.
13.
Mead GE, Morley W, Campbell P, Greig CA, McMurdo M, Lawlor B: Exercise
for depression (Review). In The Cochrane Library, The Cochrane
Collaboration. Volume 1. John Wiley & Sons, Ltd; 2009:1-60.
14.
Lawlor DA, Hopker SW: The effectiveness of exercise as an intervention in
the management of depression: systematic review and meta-regression
analysis of randomised controlled trials. BMJ 2001, 322(7289):763-767.
15.
Rethorst CD, Wipfli BM, Landers DM: The antidepressive effects of
exercise: a meta-analysis of randomized trials. Sports Med 2009,
39(6):491-511.
16.
Krogh J, Nordentoft M, Sterne JA, Lawlor DA: The effect of exercise in
clinically depressed adults: systematic review and meta-analysis of
randomized controlled trials. J Clin Psychiatry 2011, 72(4):529-538.
17.
Nicholson A, Kuper H, Hemingway H: Depression as an aetiologic and
prognostic factor in coronary heart disease: a meta-analysis of 6362
events among 146 538 participants in 54 observational studies. Eur Hear
J 2006, 27(23):2763-2774.
18.
Cuijpers P, Smit F: Excess mortality in depression: a meta-analysis of
community studies. J Affect Disord 2002, 72(3):227-236.
19.
de Wit L, Luppino F, van Straten A, Penninx B, Zitman F, Cuijpers P:
Depression and obesity: A meta-analysis of community-based studies.
Psychiatry Res 2010, 178(2):230-235.
20.
Luppino FS, de Wit LM, Bouvy PF, Stijnen T, Cuijpers P, Penninx BW,
Zitman FG: Overweight, obesity, and depression: a systematic review
and meta-analysis of longitudinal studies. Arch Gen Psychiatry 2010,
67(3):220-229.
21.
de Wit LM, van Straten A, van Herten M, Penninx BW, Cuijpers P:
Depression and body mass index, a u-shaped association. BMC Publ
Health 2009, 9:14.
22.
Penninx BW, Beekman AT, Smit JH, Zitman FG, Nolen WA, Spinhoven P,
Cuijpers P, De Jong PJ, Van Marwijk HW, Assendelft WJ, et al: The
Netherlands Study of Depression and Anxiety (NESDA): rationale,
objectives and methods. Int J Methods Psychiatr Res 2008, 17(3):121-140.
23.
Penninx BWJH, van Dyck R: Depression and somatic comorbidity. Ned
Tijdschr Geneeskd 2010, 154(15):722-727.
24.
Vogelzangs N, Kritchevsky SB, Beekman AT, Newman AB, Satterfield S,
Simonsick EM, Yaffe K, Harris TB, Penninx BW: Depressive symptoms and
change in abdominal obesity in older persons. Arch Gen Psychiatry 2008,
65(12):1386-1393.
25.
Dowlati Y, Herrmann N, Swardfager W, Liu H, Sham L, Reim EK, Lanctot KL:
A meta-analysis of cytokines in major depression. Biol Psychiatry 2010,
67(5):446-457.
26.
Kaptoge S, Di Angelantonio E, Lowe G, Pepys MB, Thompson SG, Collins R,
Danesh J: C-reactive protein concentration and risk of coronary heart
disease, stroke, and mortality: an individual participant meta-analysis.
Lancet 2010, 375(9709):132-140.
27.
Whooley MA, Caska CM, Hendrickson BE, Rourke MA, Ho J, Ali S:
Depression and inflammation in patients with coronary heart disease:
findings from the Heart and Soul Study. Biol Psychiatry 2007,
62(4):314-320.
28.
Demyttenaere K, Bonnewyn A, Bruffaerts R, Brugha T, De Graaf R, Alonso J:
Comorbid painful physical symptoms and depression: prevalence, work
loss, and help seeking. J Affect Disord 2006, 92(2-3):185-193.
29.
Ohayon MM, Schatzberg AF: Using chronic pain to predict depressive
morbidity in the general population. Arch Gen Psychiatry 2003, 60(1):39-47.
30.
Geerlings SW, Twisk JW, Beekman AT, Deeg DJ, van Tilburg W: Longitudinal
relationship between pain and depression in older adults: sex, age and
physical disability. Soc Psychiatry Psychiatr Epidemiol 2002, 37(1):23-30.
31.
Neumeyer-Gromen A, Lampert T, Stark K, Kallischnigg G: Disease
Management Programs for depression, A systematic review and meta-
analysis of randomized controlled trials. Medical Care 2004,
42(12):1211-1221.
32.
ten Doesschate MC, Koeter MW, Bockting CL, Schene AH: Health related
quality of life in recurrent depression: a comparison with a general
population sample. J Affect Disord 2010, 120(1-3):126-132.
33.
Moher D, Schulz KF, Altman DG: The Consort Statement. Lancet 2001,
357(9263):1191-1194.
34.
Dunn AL, Trivedi MH, Kampert JB, Clark CG, Chambliss HO: Exercise
treatment for depression: efficacy and dose response. Am J Prev Med
2005, 28(1):1-8.
35.
Astrand PO, Rhyming I: A nomogram for calculation of aerobic capacity
(Physical Fitness) from pulse rate during submaximal work. J Appl Physiol
1954, 7:218-221.
36.
Sarason IG, Johnson JH, Siegel JM: Assessing the impact of life changes:
Development of the life experiences survey. J Consult Clin Psychol 1978,
46:932-946.
37.
Hamilton M: A rating scale for depression. J Neurol Neurosurg Psychiatry
1960, 23:56-62.
38.
Bech P, Jorgensen B, Jeppesen K, Loldrup Poulsen D, Vanggaard T:
Personality in depression: concordance between clinical assessment and
questionnaires. Acta Psychiatr Scand 1986, 74(3):263-268.
39.
Nolen WA, Dingemans PMAJ: Instruments for measuring mood disorders.
Tijdschrift voor psychiatrie 2004, 46(10):681-686.
40.
Rush AJ, Gullion CM, Basco MR, Jarrett RB, Trivedi MH: The Inventory of
Depressive Symptomatology (IDS): psychometric properties. Psychol Med
1996, 26(3):477-486.
41.
Beck AT, Epstein N, Brown G, Steer RA: An inventory for measuring clinical
anxiety: psychometric properties. J Consult Clin Psychol 1988,
56(6):893-897.
42.
Mv K, Ormel J, Keefe F, Dworkin S: Grading the severity of chronic pain.
Pain 1992, 50(2):133-149.
43.
Janca A, Kastrup M, Katschnig H, Lopez-Ibor JJ Jr, Mezzich JE, Sartorius N:
The World Health Organization Short Disability Assessment Schedule
(WHO DAS-S): a tool for the assessment of difficulties in selected areas
of functioning of patients with mental disorders. Soc Psychiatry Psychiatr
Epidemiol 1996, 31(6):349-354.
44.
EuroQol-Group: EuroQol-a new facility for the measurement of health-
related quality of life. Health Policy 1990, 19.
45.
Hakkaart-van Roijen L: Manual Trimbos/iMTA questionnaire for costs
associated with psychiatric illness. Insitute for Medical Technology
Assessment; 2002.
46.
The revised NEO Personality Inventory (NEO-PI-R). Edited by: Costa PT,
McCrae RR. London: SAGA Publications Ltd.; 2008:.
47.
Wendel-Vos GC, Schuit AJ, Saris WH, Kromhout D: Reproducibility and
relative validity of the short questionnaire to assess health-enhancing
physical activity. J Clin Epidemiol 2003, 56(12):1163-1169.
48.
Wald FDM, Mellenbergh GJ: De verkorte versie van de Nederlandse
vertaling van de Profile of Mood States POMS. Nederlands Tijdschrift voor
de Psychologie 1990, 45:86-90.
49.
Borg S: Borg’s Percieved Exertion and Pain scales. Human Kinetics 1998.
50.
de Vocht A: SPSS 14 for Windows Utrecht: Bijleveld Press; 2007.
51.
Trimbos: Multidisciplinaire Richtlijn Depressie (eerste revisie). Utrecht:
Stuurgroep Multidisciplinaire Richtlijn Ontwikkeling 2010.
52.
Seime RJ, Vickers KS: The challenges of treating depression with exercise;
from evidence to practice. Clin Psychol Sci Pract 2006, 13:194-197.
Pre-publication history
The pre-publication history for this paper can be accessed here:
http://www.biomedcentral.com/1471-2458/12/50/prepub
doi:10.1186/1471-2458-12-50
Cite this article as: Kruisdijk et al.: Effect of running therapy on
depression (EFFORT-D). Design of a randomised controlled trial in adult
patients [ISRCTN 1894]. BMC Public Health 2012 12:50.
Kruisdijk et al. BMC Public Health 2012, 12:50
http://www.biomedcentral.com/1471-2458/12/50
Page 9 of 9
| Effect of running therapy on depression (EFFORT-D). Design of a randomised controlled trial in adult patients [ISRCTN 1894]. | 01-19-2012 | Kruisdijk, Frank R,Hendriksen, Ingrid J M,Tak, Erwin C P M,Beekman, Aartjan T F,Hopman-Rock, Marijke | eng |
PMC5015676 | original article
http://dx.doi.org/10.1590/bjpt-rbf.2014.0165
289
Braz J Phys Ther. 2016 July-Aug; 20(4):289-297
Is heart rate variability a feasible method to determine
anaerobic threshold in progressive resistance exercise in
coronary artery disease?
Milena P. R. Sperling1,2, Rodrigo P. Simões2, Flávia C. R. Caruso2,
Renata G. Mendes2, Ross Arena3, Audrey Borghi-Silva1,2,3
ABSTRACT | Background: Recent studies have shown that the magnitude of the metabolic and autonomic responses
during progressive resistance exercise (PRE) is associated with the determination of the anaerobic threshold (AT).
AT is an important parameter to determine intensity in dynamic exercise. Objectives: To investigate the metabolic
and cardiac autonomic responses during dynamic resistance exercise in patients with Coronary Artery Disease (CAD).
Method: Twenty men (age = 63±7 years) with CAD [Left Ventricular Ejection Fraction (LVEF) = 60±10%] underwent
a PRE protocol on a leg press until maximal exertion. The protocol began at 10% of One Repetition Maximum Test
(1‑RM), with subsequent increases of 10% until maximal exhaustion. Heart Rate Variability (HRV) indices from
Poincaré plots (SD1, SD2, SD1/SD2) and time domain (rMSSD and RMSM), and blood lactate were determined at
rest and during PRE. Results: Significant alterations in HRV and blood lactate were observed starting at 30% of 1-RM
(p<0.05). Bland‑Altman plots revealed a consistent agreement between blood lactate threshold (LT) and rMSSD threshold
(rMSSDT) and between LT and SD1 threshold (SD1T). Relative values of 1‑RM in all LT, rMSSDT and SD1T did not
differ (29%±5 vs 28%±5 vs 29%±5 Kg, respectively). Conclusion: HRV during PRE could be a feasible noninvasive
method of determining AT in CAD patients to plan intensities during cardiac rehabilitation.
Keywords: autonomic nervous system; anaerobic threshold; blood lactate; cardiac rehabilitation; cardiac disease; 1‑RM test.
BULLET POINTS
• Parasympathetic modulation was reduced during lower extremity resistance exercise.
• Anaerobic Threshold occurred at ≈30% of 1-RM in patients with CAD.
• HRV may prove to be a feasible tool in clinical practice to determine Anaerobic Threshold.
• HRV can be safe and appropriate method to determine exercise intensity in patients with CAD.
HOW TO CITE THIS ARTICLE
Sperling MPR, Simões RP, Caruso FCR, Mendes RG, Arena R, Borghi‑Silva A. Is heart rate variability a feasible method to determine
anaerobic threshold in progressive resistance exercise in coronary artery disease? Braz J Phys Ther. 2016 July-Aug; 20(4):289-297 .
http://dx.doi.org/10.1590/bjpt‑rbf.2014.0165
1 Interunidades Bioengenharia (EESC/FMRP/IQSC), Universidade de São Paulo (USP), São Carlos, SP, Brazil
2 Laboratório de Fisioterapia Cardiopulmonar, Departamento de Fisioterapia, Universidade Federal de São Carlos (UFSCar), São Carlos, SP, Brazil
3 Integrative Physiology Laboratory, Department of Physical Therapy, College of Applied Health Sciences, University of Illinois at Chicago (UIC),
Chicago, USA
Received: Mar 02, 2015 Revised: Aug 25, 2015 Accepted: Jan 28, 2016
Introduction
It is known that the combination of aerobic and
resistance exercise for cardiac patients synergistically
improves muscular strength and endurance, functional
capacity, quality of life, cardiovascular function,
metabolism, and cardiovascular risk profile1. In addition,
resistance exercise is considered safe for both healthy
elderly individuals and cardiac patients1‑4.
The magnitude of the cardiovascular and ventilatory
responses to exertional demands depends on the
type of physical exercise and the intensity of effort1.
With respect to exercise intensity, the anaerobic
threshold (AT) is defined as a point above a given power
value when the production of lactic acid is greater
than the capacity for its utilization by body tissues5‑7.
The point past which blood lactate concentration
increases progressively5 is an important parameter in
determining submaximal exercise tolerance. The use of
discontinuous protocols to assess functional capacity
Sperling MPR, Simões RP, Caruso FCR, Mendes RG, Arena R, Borghi-Silva A
290
Braz J Phys Ther. 2016 July-Aug; 20(4):289-297
and determine AT are potentially advantageous as
they reduce the inherent added risks incurred during
maximum stress intensities2.
The ability of Heart Rate Variability (HRV) to
determine changes in blood lactate and AT during
resistance and aerobic exercise in healthy individuals
has already been investigated8,9. Other studies have
also examined the behavior of HRV indices during
exercise in diabetic10, heart failure11, and elderly12‑14
cohorts. However, parameters that indicate safe training
intensities with resistance exercise, particularly in
patients with cardiac conditions, remain unclear.
While HRV indices are important predictors of
cardiovascular risk and risk of sudden cardiac death
and may be used as potential indices of relative risk15,
the use of HRV to determine the point of transition
between aerobic and anaerobic metabolism (i.e., AT)
during incremental resistance exercise in patients with
cardiac disease is unknown. Therefore, the objectives
of this study were to: 1) evaluate the behavior of
HRV and blood lactate; 2) determine the AT during
an incremental leg‑press protocol with an incremental
percentage of One Repetition Maximum Test (1‑RM);
and 3) evaluate the degree of agreement between HRV
indices and blood lactate in relation to the AT in a
cohort diagnosed with coronary artery disease (CAD).
Method
Study design and population
This is an observational cross‑sectional study
involving 20 males with clinically stable CAD (sample
of convenience) participating in an outpatient cardiac
rehabilitation program. Inclusion criteria consisted of
1) being at least 12 months post an acute event (i.e.,
myocardial infarction) or 12 months after a surgical or
percutaneous revascularization procedure and 2) being
clinically stable on a regular pharmacologic regimen.
The experimental protocol was approved by the
Research Ethics Committee of Centro Universitário
de Araraquara, Araraquara, SP, Brazil (n. 1331‑11).
All procedures were conducted in accordance with
the Declaration of Helsinki. All participants signed
an informed consent form.
Experimental procedures
Subjects did not ingest caffeine or alcohol during
the 24‑hour period prior to any of the testing protocols
and did not perform any rigorous physical activity
during the 48 hours prior to testing. All trials were
performed at the same period of the day to avoid
any influence of circadian rhythm on cardiovascular
variables. The experiments were carried out in a
climate‑controlled room (21‑24 °C) with a relative
air humidity of 40‑60%.
Clinical examination was performed by a physician
(cardiologist) before study initiation. This examination
consisted of anamnesis and resting 12‑lead
electrocardiography. A transthoracic echocardiogram
was also performed for all patients.
Cardiopulmonary exercise testing – CPX
A symptom‑limited incremental exercise test
(CPX) was performed on a recumbent cycle ergometer
(Corival, Lode, Groningen, The Netherlands) with the
collection of gas exchange and ventilatory variables
using a calibrated computer‑based exercise system
(Oxycon Mobile, JaegerTM, Hoechberg, Germany).
The workload (W) was continuously increased in a
linear “ramp” pattern of 15 W.min–1. The test finished
when subjects reached physical exhaustion or when
abnormal test responses warranted test termination16,17.
The incremental exercise testing duration was between
8 and 12 minutes18.
Peak VO2 was defined as the highest value during
the last 15 seconds of exercise and peak respiratory
exchange ratio (RER) was the 15‑second averaged
VCO2 divided by VO2 at peak exercise16.
One Repetition Maximum test – (1-RM - leg
press)
This test was applied by gradually increasing the
resistance until the patients succeeded in performing no
more than one repetition on the leg press at a 45 degree
angle (Pro‑Fitness, São Paulo, Brazil). The resistance
load for 1‑RM was estimated (1‑RM‑E) before the
test by multiplying subject body weight by 3.5, based
on pilot testing. The details of this test protocol have
been described previously13.
Discontinuous resistance exercise testing
(DRET-leg press)
72 hours after the 1‑RM test, Discontinuous
Resistance Exercise Testing (DRET–leg press) was
performed, starting at a load of 10% of 1‑RM with
subsequent increases of 10% until exhaustion. At each
percentage of effort, patients underwent 2 minutes of
exercise at a movement rhythm of 12 repetitions/minute,
maintaining respiratory cadence. The period between
trials was 5 minutes. The details of this test protocol
have been described previously13.
HRV in resistance exercise in CAD patients
291
Braz J Phys Ther. 2016 July-Aug; 20(4):289-297
Heart Rate (HR) and R‑R intervals were recorded
with a wireless HR monitor (Polar S810i, Kempele,
Finland) and blood samples (via earlobe puncture)
were taken at rest and immediately after the final
repetition completed at each load (i.e., % of 1‑RM).
Blood samples were analyzed using a YSI 1500
Sports Lactate Analyzer (YSI Inc., Yellow Springs,
OH, USA).
Measurement of HRV
The R‑R intervals were recorded continuously
with the wireless HR monitor (Polar S810i) during
all exercise protocols. The R‑Ri captured with the
monitor can be analyzed with both linear and nonlinear
models. After data capture, the signals were transmitted
to a receiver and interface connected to a computer
for subsequent analysis. The details of this technique
have been described previously13.
Safety during exercise protocols (CPX,
1-RM–leg press and DRET-leg press)
During all exercise protocols, HR was recorded
with the HR monitor (Polar S810i) and the ECG was
constantly monitored using a USB electrocardiogram
(WinCardio, Micromed Biotecnologia, Brasilia,
Brazil) to detect any potential arrhythmias or signs
of ischemia that would indicate the protocol should
be interrupted. Blood pressure (auscultation) and
symptoms (muscle fatigue, chest pain, and breathing/
dyspnea), assessed by means of the 10-point modified
Borg Scale Rating16,19‑21, were measured and recorded
immediately after each effort.
The criteria for protocol termination were a systolic
blood pressure >200 mmHg, symptoms of lower limb
fatigue, angina or shortness of breath, development
of any cardiac arrhythmias, or achieving maximum
voluntary exhaustion13.
Data analysis
To evaluate the responses of HR and R‑Ri during
DRET–leg press, the first step in the data analysis
involved a visual inspection of R‑Ri (ms) distribution
in the ECG in order to select the sections corresponding
to the final minute of each load (second minute) of
resistance exercise maneuvers, as this was considered
to be a more stable phase for analysis22‑24.
Ectopic beats, arrhythmic events, missing data, and
noise effects that might alter the estimation of HRV
were excluded15. HRV analysis was carried out using
the following linear and nonlinear methods: 1) Linear
methods ‑ RMSSD (square root of the mean of the sum
of the squares of differences between adjacent RRi
divided by the number of RRi minus one, expressed in
ms) and RMSM (square root of the sum of the squares
of differences of individual values compared to the
mean value, divided by the number of RRi in a period
for the time domain); and 2) nonlinear method ‑ SD1
(instantaneous R‑R interval variability from Poincaré
plots), SD2 (standard deviation of continuous long‑
term R‑R interval variability), and the SD1/SD2 ratio
carried out by the Poincaré method of quantitative
two‑dimensional vector analysis15.
The Poincaré plots were analyzed quantitatively,
based on the premise of different temporal effects
of changes in vagal and sympathetic modulation of
HR on the R‑R intervals without a requirement for
a stationary quality of the data24. RR‑interval series
were processed using Kubios HRV 2.0 (University of
Kuopio, Finland). The details of this technique have
been described previously24.
Determination of Anaerobic Threshold (AT)
in resistance exercise
To determine AT, changes in blood lactate curves
were generated for each subject and the AT was
defined as the exercise intensity at which the blood
lactate concentration began to increase exponentially,
i.e., breakpoint8,12‑14.
To determine the HRV threshold, the rMSSD and
SD1 for each stage of exercise were plotted against
work rate. This HRV deflection point was defined as
the HRV threshold9. The point at which there was
an initial decline in indices during exercise, thus
indicating vagal withdrawal.
The determination of the lactate and HRV threshold
occurred through visual inspection of lactate and HRV
curves, respectively, by two independent experienced
examiners. When there was no agreement between
the two evaluators, a third evaluator was called to
give the casting vote.
Statistical analysis
The sample size for the current study was estimated
considering previous studies with the same resistance
exercise protocol for healthy and elderly13,14 subjects
and also different resistance exercise protocols for
coronary artery disease22,25. Considering the presence
of coronary artery disease, we doubled the sample
size (n=20) to increase the chance of having less
variability of the resulting data.
Initially, we used the Kolmogorov‑Smirnov test
to verify the normality of the data and subsequently
Sperling MPR, Simões RP, Caruso FCR, Mendes RG, Arena R, Borghi-Silva A
292
Braz J Phys Ther. 2016 July-Aug; 20(4):289-297
one‑way ANOVA with repeated measures was used
to analyze the behavior of the HRV indices, blood
lactate curves, R‑Ri and RPP responses during the
DRET–leg press (at different percentages of 1‑RM),
and the different methods of identifying AT (blood
lactate curves, rMSSD, and SD1 threshold). When
appropriate, post‑hoc analyses were performed using
the Tukey test. The degree of agreement between the
methods used to determine AT was verified using
Bland‑Altman plots26‑28.
Data are reported as mean and standard deviation
and the significance level was set at 5%. The statistical
analysis was carried out using Sigma Plot for Windows
version 11.0 (Sigma Plot, San Jose, CA, USA) and
MedCalc version 12.6.1.0 (MedCalc Software,
Ostend, Belgium).
Results
Over a one‑year period, 42 patients were assessed
for eligibility, 26 were recruited, five did not meet the
inclusion criteria, and one was excluded for having
an inadequate blood pressure response during CPX.
Among the remaining subjects, 20 completed the
protocol successfully with no abnormalities that
would contraindicate enrollment in the present study
and were included in the final analysis.
The clinical characteristics of the subjects are
summarized in Table 1. All subjects had normal left
ventricular systolic function (and mild left ventricle
diastolic dysfunction in 45% of the study population)
measured by echocardiography. The majority of
patients had hypertension, history of smoking, and a
family history of CAD. Myocardial infarction was the
predominant clinical diagnosis and all patients were
NYHA class I. Pharmacologic treatment commonly
included antiplatelets, statins, beta‑blockers, ACE
inhibitors, and hypoglycemic agents. Mean CPX
values indicate this sample had a well‑preserved
functional capacity and exerted maximal effort
during the exercise test, according to American Heart
Association (AHA) standards16.
1‑RM testing and DRET (30% and 50%) responses
are summarized in Table 2. Regarding the response
to the 1‑RM testing, the criterion for termination
for almost all subjects was muscle fatigue (rate of
perceived exertion – RPE = 9.2±2.0), with only one test
interrupted due to chest pain. No test was interrupted
due to ECG alterations or an excessive rise in SBP
(>200 mmHg). In relation to the resistance load
achieved during 1‑RM, values were similar to those
Table 1. Baseline characteristics of the study population.
CAD, n=20
Demographics/anthropometrics
Age, years
63±7
Height, m
1.7±0.1
Body mass, kg
75.7±12.7
BMI, kg/m2
26.6±2.9
Transthoracic echocardiography
LVEF, %
60±10
LV diastolic diameter, cm
5.3±0.6
LV diastolic volume, ml
143±40
Septal thickness, cm
0.9±0.2
Posterior wall thickness, cm
0.9±0.1
Doppler echocardiography
LV diastolic function*:
Normal
11 (55)
Mild dysfunction
9 (45)
Risk Factors, n (%)
Diabetes
5 (25)
Hypertension
13 (65)
History of smoking
12 (60)
Family history of CAD
18 (90)
Functional Class (NYHA): I, n (%)
20 (100)
History of Myocardial infarction, n (%)
15 (75)
Intervention, n (%)
CABG
9 (45)
PCI
18 (90)
Medications, n (%)
Antiplatelet (aspirin)
20 (100)
Statin
20 (100)
Beta‑blocker
14 (70)
ACE inhibitor
7 (35)
Hypoglycemic
5 (25)
CPX
Peak VO2, ml.Kg–1.min–1
24±5
PredictiveVO2, ml.Kg–1.min–1 (%)*
85±14
Peak workload, W
134±23
Data are presented as mean±SD or number (percentage) of subjects. CAD:
coronary artery disease; BMI: body mass index; NYHA: New York Heart
Association; CABG: coronary artery bypass grafting; PCI: percutaneous
coronary intervention; LVEF: left ventricular ejection fraction; ACE:
angiotensin‑converting enzyme; CPX: cardiopulmonary exercise
testing; VO2: oxygen uptake; W: watts. *Clinical recommendations
for Cardiopulmonary Exercise Testing data assessment in specific
patient populations16.
HRV in resistance exercise in CAD patients
293
Braz J Phys Ther. 2016 July-Aug; 20(4):289-297
stipulated previously during the pilot test (3.5 times the
body weight of the patient). Regarding the response
to the DRET, the criterion for termination for almost
all subjects was muscle fatigue (RPE = 6.6±2.8) or
an excessive rise in SBP (>200 mHg) and only one
test was interrupted due to chest pain.
Figure 1 illustrates the behavior of HRV indices,
blood lactate, and R‑Ri and RPP at rest and with
the increasing resistance exercise loads through the
common maximum load achieved by all patients (i.e.,
50% of 1‑RM). Both rMSSD and SD1 indices, which
are representative of parasympathetic modulation,
demonstrated a significant decrease at peak load
compared to resting values, with a significant drop at
30% of 1‑RM (Figure 1A, C) with a parallel significant
increase in blood lactate at 30% of 1‑RM (Figure 1B).
The SD1/SD2 ratio had a significant decrease from
40% of 1‑RM (Figure 1E). The RMSM and SD2
indices (Figure 1A, C) were significantly increased
at 50% of 1‑RM, although there was an increasing
trend starting in 30% of 1‑RM, perceived visually.
R‑Ri showed a progressive reduction (Figure 1D)
and RPP showed a progressive increase concurrently
(Figure 1F), reflecting the progressive increase in the
intensity of effort at 50% of 1‑RM.
The AT was determined for each patient through
the analysis of blood lactate curves, rMSSD, and
SD1, expressed in both absolute and relative values
(Table 3). There were no significant differences in
relation to different methods for identifying absolute
values in Kg (lactate threshold ‑ LT: 81±19, rMSSD
threshold ‑ rMSSDT: 78 ± 14; SD1 threshold ‑ SD1T:
79±13) and relative AT values at ≈30% of 1-RM (29±5;
28±5; 29±5; respectively), as presented in Table 3.
The analysis of agreement between methods of
determining the AT was carried out using Bland‑Altman
plots, considering the blood lactate analyses as the
“gold standard“. LT vs. rMSSDT and LT vs. SD1T were
plotted. The mean of the differences for identifying AT
using the LT and rMSSDT methods was 2.7±20 Kg
(Figure 2A), and the mean difference between LT and
SD1T was 1.3±19.1 Kg (Figure 2B), demonstrating
good agreement.
Discussion
In this observational cross‑sectional study, we were
able to demonstrate that the fall in parasympathetic
indices is associated with an increase in blood lactate,
starting at ≈30% of 1-RM using a leg-press maneuver.
Table 2. Cardiovascular, metabolic, and cardiac autonomic variables obtained during peak 1‑RM testing and DRET (30% and 50% of
the 1‑RM testing).
CAD, n=20
1-RM testing
DRET 30%
DRET 50%
SBP, mmHg
137±24
156±25
172±23
F=9.07
DBP, mmHg
76±14
87±11
89±18
F=4.47
HR, bpm
95±13
88±11
104a±16
F=6.15
RPE, 0‑10
Lower limb fatigue
9.2±2.0
3.0±2.3
6.6±2.8
F=35.79
Chest pain (angina)
*
−−
*
F=9.22
Breathing (dyspnea)
5.9±3.2
2.8±2.4
5.9±2.9
Load, Kg
282±46
85±13c
144a,b±23
F=211
Load 1RM/total body mass
3.8±0.9
−−
−−
Lactate, mlmol.L–1
−−
1.5±0.8
3.4±1.7a
F=28.60
HRV indices
rMSSD
−−
8.6±4.1
7.1±4.7
F=23.61
RMSM
−−
15.8±7.3
25.5a±16.0
F=2.53
SD1
−−
6.1±2.9
5.1±3.3
F=23.66
SD2
−−
21.2±10.2
35.5a±22.4
F=3.24
SD1/SD2
−−
0.3±0.2
0.2±0.1
F=21.14
Data are presented as mean ± SD. SBP: systolic blood pressure; DBP: diastolic blood pressure; HR: heart rate; 1‑RM: one repetition maximum;
DRET: discontinuous resistance exercise testing; RPE: rate of perceived exertion. *Only one patient had chest pain. a: difference between DRET
50% and DRET 30%. b: difference between DRET 50% and 1‑RM testing. c: difference between DRET 30% and 1‑RM testing (p value <0.05,
one‑way ANOVA with repeated measures).
Sperling MPR, Simões RP, Caruso FCR, Mendes RG, Arena R, Borghi-Silva A
294
Braz J Phys Ther. 2016 July-Aug; 20(4):289-297
The good agreement between HRV indices and blood
lactate curves may represent the importance and value
of HRV in CAD patients for exercise prescription
and monitoring.
Responses during discontinuous resistance
exercise
The rMSSD and SD1 indices reflect parasympathetic
heart activity15,24,28 and they both demonstrated a
significant drop from ≈30% of 1-RM (Figure 1A, C).
The total HRV, represented by RMSM and SD2
indices (Figure 1A, C) were significantly increased
from ≈50%of 1-RM, although the increasing trend
started at ≈30% of 1-RM, observed visually. Lastly,
the SD1/SD2 ratio appears stable up to ≈30% of
1‑RM, followed by changes thereafter (Figure 1E).
All of these changes indicate a shift in sympathovagal
balance towards sympathetic predominance and
reduced vagal tone29,30. This increase in sympathetic
tone appears to correspond with AT, which in this
study, corresponded to ≈30% of 1-RM.
Figure 1. Behavior of variables in Discontinuous Resistance Exercise Testing (DRET) in percentage of 1 repetition maximum (1‑RM;
x axis), starting from rest until the load in common for all patients (50% of 1‑RM). Data are presented as mean±SD. (A) rMSSD (square
root of the difference in the sum of squares between R‑R interval on the recording, divided by the determined time minus one) and
RMSM (root mean square of the differences from the mean interval); (B) Blood Lactate; (C) SD1 (standard deviation of instantaneous
beat‑to‑beat R‑R interval variability) and SD2 (the standard deviation of continuous long‑term R‑R interval. &: difference in relation
to rest. +: difference in relation to 10% of 1‑RM. ‡: difference in relation to 20% of 1‑RM. #: difference in relation to 30% of 1‑RM.
•: difference in relation to 40% of 1-RM. *: difference in relation to 50% of 1-RM (one-way ANOVA with repeated measures; p<0.05).
HRV in resistance exercise in CAD patients
295
Braz J Phys Ther. 2016 July-Aug; 20(4):289-297
Anaerobic threshold determination by HRV
and blood lactate
The determination of AT through indices of HRV
was effective and associated with blood lactate
responses in patients with CAD who are receiving
standard pharmacological therapy. This is an important
topic, since these results can be more representative
of the CAD population seen clinically. In this context,
Machado et al.25 assessed HRV indices during
progressive upper limb exercise in CAD patients and
found medications did not influence the HRV response.
In the present study, the load corresponding to
the AT, considering the blood lactate threshold as a
parameter during DRET‑leg press was obtained at
≈30% of the peak load reached during the 1-RM test
(Table 3), which is in accordance with other studies
in assessing apparently healthy subjects14,31.
Figure 2 demonstrates that, although there were
agreements among the methods for determining the
AT (the mean of the differences was close to zero),
the limits of agreement were clinically wide. Other
studies have shown the potential use of HRV for
the determination of AT/ventilator threshold on a
cycle ergometer using rMSSD and SD1 in healthy
adults9 and in patients with type‑2 diabetes10. To our
knowledge, this is the first study to analyze the behavior
of metabolic and autonomic responses during lower
limb resistance exercise in CAD patients.
The mean CPX values (peak VO2 and predictive
VO2) indicate that these patients had a well‑preserved
functional capacity and confirm a maximal effort
during the exercise test according to AHA standards16.
Regarding the response to the DRET, the criteria for
interrupting the test was muscle fatigue or excessive
rise in SBP (>200 mmHg). All of these patients were
included in the data analysis.
Study perspectives
Our results suggest that HRV may also be considered
a useful tool in clinical practice to determine the intensity
corresponding to AT. AT was approximately 30% of
1‑RM testing for CAD patients with well‑preserved
Table 3. Comparison of relative and absolute resistance values for anaerobic threshold measured with different methods of identification
during discontinuous resistance exercise testing (DRET).
LT
rMSSDT
SD1T
DRET
Absolute values, Kg
81±19
78±14
79±13
p=0.43; F=0.94
Relative values, %
29±5
28±5
29±5
p=0.52; F=0.76
Data are presented as mean±SD. LT: Lactate threshold; rMSSDT: rMSSD threshold; SD1T: SD1 threshold. No significant differences among
the three methods of identifying the anaerobic threshold (one‑way ANOVA with repeated measures).
Figure 2. Bland‑Altman plot showing the agreement between LT and rMSSDT (A) and LT and SD1T (B). BIAS = mean of the differences
among the averages; ± 1.96 SD = 95% limits of agreement. LT = lactate threshold; rMSSDT = rMSSD threshold (rMSSD: square root
of the mean of the sum of the squares of differences between adjacent RR‑intervals on the recording, divided by the determined time
minus one); SD1T = SD1 threshold (SD1: standard deviation of instantaneous R‑R interval variability). Horizontal lines indicate mean
(solid lines) and 95% confidence intervals (dashed lines) of differences between two measurements.
Sperling MPR, Simões RP, Caruso FCR, Mendes RG, Arena R, Borghi-Silva A
296
Braz J Phys Ther. 2016 July-Aug; 20(4):289-297
functional capacity. HRV analysis using linear and
nonlinear methods could be considered an important
method for evaluating and understanding cardiac
autonomic modulation in CAD patients during dynamic
resistance exercise.
In order to establish the correct intensity, it is
important to consider that the same exercise may
lead to different levels of stress in different patient
populations. Several factors, such as body weight,
coordination, intention, and perception of the level
of effort during resistance exercise, directly interfere
with measurements of effort2.
Limitations of this study
The current study has limitations that should be
recognized. RMSM and SD2 HRV indices reflect both
sympathetic and parasympathetic influences15,29 and a
pure index representative of the sympathetic modulation
was not assessed in this study. Moreover, during the
exercise protocol with increased load increments every
2 minutes, HRV indices reached steady state in the last
minute of each stage of exercise up to AT. However,
after AT, this equilibrium condition was not maintained,
which is inherent to high exercise intensities. Thus, it
is possible that exercise intensities after AT may have
affected HRV data capture. Even so, clear trends were
apparent in the current investigation. The results found
in this study may be protocol‑dependent, considering
the duration of each load and rest periods between
them. The leg press was chosen because it induces
more changes in cardiac, ventilatory, and metabolic
parameters, but it is necessary to investigate other kinds
of resistance exercise. Once the resistance activity
stops, the blood pressure decreases quite rapidly so
that measuring by auscultation at the end of exercise
would do not give a reliable estimation of the blood
pressure during exercise. The evaluation of the blood
pressure response was limited to the evaluation of
discontinuous blood pressure monitoring, measured
at the end of the exercise. However, this is still the
most widely used method in clinical practice.
Conclusion
Our results suggest that parasympathetic modulation
was reduced during lower extremity resistance exercise,
beginning at AT, which occurred at ≈30% of 1-RM.
Moreover, HRV may prove to be a feasible tool in
clinical practice to determine AT, aiding in setting
safe and appropriate exercise intensity parameters
in patients with CAD.
Acknowledgements
This work was funded by Coordenação de
Aperfeiçoamento de Pessoal de Nível Superior ‑ CAPES
(grant no. 00011/07‑0) and Fundação de Amparo à
Pesquisa do Estado de São Paulo ‑ FAPESP (grant
no. 2009/01842‑0).
References
1. Williams MA, Haskell WL, Ades PA, Amsterdam EA, Bittner
V, Franklin BA, et al. Resistance exercise in individuals with
and without cardiovascular disease: 2007 update: a scientific
statement from the American Heart Association Council
on Clinical Cardiology and Council on Nutrition, Physical
Activity, and Metabolism. Circulation. 2007;116(5):572‑84.
http://dx.doi.org/10.1161/CIRCULATIONAHA.107.185214.
PMid:17638929.
2. Bjarnason‑Wehrens B, Mayer‑Berger W, Meister ER, Baum
K, Hambrecht R, Gielen S. Recommendations for resistance
exercise in cardiac rehabilitation. Recommendations of
the German Federation for Cardiovascular Prevention and
Rehabilitation. Eur J Cardiovasc Prev Rehabil. 2004;11(4):352‑
61. http://dx.doi.org/10.1097/01.hjr.0000137692.36013.27.
PMid:15292771.
3. Delagardelle C, Feiereisen P, Autier P, Shita R, Krecke R,
Beissel J. Strength/endurance training versus endurance
training in congestive heart failure. Med Sci Sports Exerc.
2002;34(12):1868‑72. http://dx.doi.org/10.1097/00005768‑
200212000‑00002. PMid:12471289.
4. Pollock ML, Franklin BA, Balady GJ, Chaitman BL, Fleg
JL, Fletcher B, et al. Resistance exercise in individuals with
and without cardiovascular disease: benefits, rationale,
safety, and prescription: an advisory from the Committee
on Exercise, Rehabilitation, and Prevention, Council
on Clinical Cardiology, American Heart Association.
Circulation. 2000;101(7):828‑33. http://dx.doi.org/10.1161/01.
CIR.101.7.828. PMid:10683360.
5. Davis JA, Rozenek R, DeCicco DM, Carizzi MT, Pham PH.
Comparison of three methods for detection of the lactate
threshold. Clin Physiol Funct Imaging. 2007;27(6):381‑
4. http://dx.doi.org/10.1111/j.1475‑097X.2007.00762.x.
PMid:17944661.
6. Wasserman K, Hansen JE, Sue D, Whipp BJ, Casaburi R.
Principles of exercise testing and interpretation. 3rd ed.
Philadelphia: Lea & Febiger; 1999.
7. Ekkekakis P, Hall EE, Petruzzello S. Practical markers of
the transition from aerobic to anaerobic metabolism during
exercise: rationale and a case for affect‑based exercise
prescription. Prev Med. 2004;38(2):149‑59. http://dx.doi.
org/10.1016/j.ypmed.2003.09.038. PMid:14715206.
8. Souza NM, Magosso RF, Pereira GB, Leite RD, Arakelian
VM, Montagnolli AN, et al. The measurement of lactate
threshold in resistance exercise: a comparison of methods.
Clin Physiol Funct Imaging. 2011;31(5):376‑81. http://dx.doi.
org/10.1111/j.1475‑097X.2011.01027.x. PMid:21771257.
9. Karapetian GK, Engels HJ, Gretebeck RJ. Use of heart
rate variability to estimate LT and VT. Int J Sports Med.
HRV in resistance exercise in CAD patients
297
Braz J Phys Ther. 2016 July-Aug; 20(4):289-297
2008;29(8):652‑7. http://dx.doi.org/10.1055/s‑2007‑989423.
PMid:18213538.
10. Sales MM, Campbell CS, Morais PK, Ernesto C, Soares‑
Caldeira LF, Russo P, et al. Noninvasive method to
estimate anaerobic threshold in individuals with type 2
diabetes. Diabetol Metab Syndr. 2011;3(1):1. http://dx.doi.
org/10.1186/1758‑5996‑3‑1. PMid:21226946.
11. Leprêtre PM, Bulvestre M, Ghannem M, Ahmaidi S, Weissland
T, Lopes P. Determination of ventilatory threshold using
heart rate variability in patients with heart failure. Surgery.
2013; S12:003. http://dx.doi.org/10.4172/2161‑1076.S12‑003.
12. Simões RP, Castello‑Simões V, Mendes RG, Archiza B, Santos
DA, Bonjorno JC Jr, et al. Identification of anaerobic threshold
by analysis of heart rate variability during discontinuous
dynamic and resistance exercise protocols in healthy older
men. Clin Physiol Funct Imaging. 2014;34(2):98‑108. http://
dx.doi.org/10.1111/cpf.12070. PMid:23879324.
13. Simões RP, Castello‑Simões V, Mendes RG, Machado
HG, Archisa B, Santos DA, et al. Lactate and Heart Rate
Variability Threshold during Resistance Exercise in the
Young and Elderly. Int J Sports Med. 2013;34(11):991‑6.
http://dx.doi.org/10.1055/s‑0033‑1337946. PMid:23606341.
14. Simões RP, Mendes RG, Castello V, Machado HG, Almeida
LB, Baldissera V, et al. Heart rate variability and blood‑lactate
threshold interaction during progressive resistance exercise
in healthy older men. J Strength Cond Res. 2010;24(5):1313‑
20. http://dx.doi.org/10.1519/JSC.0b013e3181d2c0fe.
PMid:20393353.
15. Electrophysiology TFESCNAS, and the Task Force of the
European Society of Cardiology and the North American
Society of Pacing and Electrophysiology. Heart rate variability:
standards of measurement, physiological interpretation and
clinical use. Circulation. 1996;93(5):1043‑65. http://dx.doi.
org/10.1161/01.CIR.93.5.1043. PMid:8598068.
16. Guazzi M, Adams V, Conraads V, Halle M, Mezzani
A, Vanhees L, et al. Clinical recommendations for
Cardiopulmonary Exercise Testing data assessment in
specific patient populations: EACPR/AHA scientific
statement. Circulation. 2012;126(18):2261‑74. http://dx.doi.
org/10.1161/CIR.0b013e31826fb946. PMid:22952317.
17. Balady GJ, Arena R, Sietsema K, Miers J, Coke L, Fletcher
GF, et al. Clinician’s guide to cardiopulmonary exercise
testing in adults: a cientific statement from the American
Heart Association. Circulation. 2010;122(2):191‑225. http://
dx.doi.org/10.1161/CIR.0b013e3181e52e69. PMid:20585013.
18. Buchfuhrer MJ, Hansen JE, Robinson TE, Sue DY, Wasserman
K, Whipp BJ. Optimizing the exercise protocol for
cardiopulmonary assessment. J Appl Physiol. 1983;55(5):1558‑
64. PMid:6643191.
19. Myers J, Arena R, Franklin B, Pina I, Kraus WE, Mcinnis
K, et al. Recommendations for clinical exercise laboratories:
a scientific statement from the American Heart Association.
Circulation. 2009;119(24):3144‑61. http://dx.doi.org/10.1161/
CIRCULATIONAHA.109.192520. PMid:19487589.
20. Myers JN. Perception of chest pain during exercise testing
in patients with coronary artery disease. Med Sci Sports
Exerc. 1994;26(9):1082‑6. http://dx.doi.org/10.1249/00005768‑
199409000‑00003. PMid:7808240.
21. Borg GA. Psychophysical bases of perceived exertion.
Med Sci Sports Exerc. 1982;14(5):377‑81. http://dx.doi.
org/10.1249/00005768‑198205000‑00012. PMid:7154893.
22. Machado‑Vidotti HG, Mendes RG, Simões RP, Castello‑
Simões V, Catai AM, Borghi‑Silva A. Cardiac autonomic
responses during upper versus lower limb resistance exercise
in healthy elderly men. Braz J Phys Ther. 2014;18(1):9‑
18. http://dx.doi.org/10.1590/S1413‑35552012005000140.
PMid:24675908.
23. Moreira SR, Arsa G, Oliveira HB, Lima LC, Campbell CS,
Simões HG. Methods to identify the lactate and glucose
thresholds during resistance exercise for individuals with type
2 diabetes. J Strength Cond Res. 2008;22(4):1108‑15. http://
dx.doi.org/10.1519/JSC.0b013e31816eb47c. PMid:18545200.
24. Tulppo MP, Makikallio TH, Takala TES, Seppanen T, Huikuri
HV. Quantitative beat‑to‑beat analysis of heart rate dynamics
during exercise. Am J Physiol 1996;271:H244‑H252.
25. Machado HG, Simões RP, Mendes RG, Castello V, Di
Thommazo L, Almeida LB, et al. Cardiac autonomic
modulation during progressive upper limb exercise by
patients with coronary artery disease. Braz J Med Biol
Res. 2011;44(12):1276‑84. http://dx.doi.org/10.1590/S0100‑
879X2011007500134. PMid:22002089.
26. Bland JM, Altman DG. Agreed statistics: measurement
method comparison. Anesthesiology. 2012;116(1):182‑
5. http://dx.doi.org/10.1097/ALN.0b013e31823d7784.
PMid:22129533.
27. Bland JM, Altman DG. Measuring agreement in method
comparison studies. Stat Methods Med Res. 1999;8(2):135‑
60. http://dx.doi.org/10.1191/096228099673819272.
PMid:10501650.
28. Bland JM, Altman DG. Statistical methods for assessing
agreement between two methods of clinical measurement.
Lancet. 1986;8(1):307‑10. http://dx.doi.org/10.1016/S0140‑
6736(86)90837‑8. PMid:2868172.
29. Tulppo MP, Makikallio TH, Seppanen T, Laukkanen RT,
Heikki V, Huikuri HV. Vagal modulation of heart rate
during exercise: effects of age and physical fitness. Am J
Physiol. 1998;274:H424‑H429.
30. Mitchell JHJB. Wolffe memorial lecture. Neural control
of the circulation during exercise. Med Sci Sports Exerc.
1990;22(2):141‑54. PMid:2192221.
31. Souza NM, Magosso RF, Pereira GB, Sousa MV, Vieira A,
Marine DA, et al. Acute cardiorrespiratory and metabolic
responses during resistance exercise in the lactate threshold
intensity. Int J Sports Med. 2012;33(2):108‑13. http://dx.doi.
org/10.1055/s‑0031‑1286315. PMid:22127560.
Correspondence
Milena Pelosi Rizk Sperling
Universidade Federal de São Carlos
Departamento de Fisioterapia
Laboratório de Fisioterapia Cardiopulmonar
Rodovia Washington Luis, Km 235
CEP 13565‑905, São Carlos, SP, Brazil
e‑mail: [email protected]
| Is heart rate variability a feasible method to determine anaerobic threshold in progressive resistance exercise in coronary artery disease? | 06-16-2016 | Sperling, Milena P R,Simões, Rodrigo P,Caruso, Flávia C R,Mendes, Renata G,Arena, Ross,Borghi-Silva, Audrey | eng |
PMC9518107 | Citation: Prieto-González, P.;
Sedlacek, J. Effects of
Running-Specific Strength Training,
Endurance Training, and Concurrent
Training on Recreational Endurance
Athletes’ Performance and Selected
Anthropometric Parameters. Int. J.
Environ. Res. Public Health 2022, 19,
10773. https://doi.org/10.3390/
ijerph191710773
Academic Editors: Jesús Siquier Coll,
Ignacio Bartolomé and María
Concepción Robles-Gil
Received: 29 June 2022
Accepted: 26 August 2022
Published: 29 August 2022
Publisher’s Note: MDPI stays neutral
with regard to jurisdictional claims in
published maps and institutional affil-
iations.
Copyright:
© 2022 by the authors.
Licensee MDPI, Basel, Switzerland.
This article is an open access article
distributed
under
the
terms
and
conditions of the Creative Commons
Attribution (CC BY) license (https://
creativecommons.org/licenses/by/
4.0/).
International Journal of
Environmental Research
and Public Health
Article
Effects of Running-Specific Strength Training, Endurance
Training, and Concurrent Training on Recreational Endurance
Athletes’ Performance and Selected Anthropometric Parameters
Pablo Prieto-González 1,*
and Jaromir Sedlacek 2
1
Health and Physical Education Department, Prince Sultan University, Riyadh 11586, Saudi Arabia
2
Department of Sport Kinanthropology, Faculty of Sports, University of Prešov, 080 01 Prešov, Slovakia
*
Correspondence: [email protected]
Abstract: Objective: The present study aimed to verify the effects of running-specific strength
training alone, endurance training alone, and concurrent training on recreational endurance athletes’
performance and selected anthropometric parameters. Method: Thirty male recreational endurance
runners were randomly assigned using a blocking technique to either a running-specific strength
training group (RSSTG), an endurance training group (ETG), or a concurrent training group (CTG).
RSSTG performed three strength-training sessions per week orientated to running, ETG underwent
three endurance sessions per week, and CTG underwent a 3-day-per-week concurrent training
program performed on non-consecutive days, alternating the strength and endurance training
sessions applied to RSSTG and ETG. The training protocol lasted 12 weeks and was designed using
the ATR (Accumulation, Transmutation, Realization) block periodization system. The following
assessments were conducted before and after the training protocol: body mass (BM), body mass
index (BMI), body fat percentage (BFP), lean mass (LM), countermovement jump (CMJ), 1RM (one-
repetition maximum) squat, running economy at 12 and 14 km/h (RE12 and RE14), maximum oxygen
consumption (VO2max), and anaerobic threshold (AnT). Results: RSSTG significantly improved the
results in CMJ, 1RM squat, RE12, and RE14. ETG significantly improved in RE12, RE14, VO2max,
and AnT. Finally, CTG, obtained significant improvements in BFP, LM, CMJ, 1RM squat, RE12,
RE14, VO2max, and AnT. RSSTG obtained improvements significantly higher than ETG in CMJ,
1RM squat, and RE14. ETG results were significantly better than those attained by RSSTG in AnT.
Moreover, CTG marks were significantly higher than those obtained by ETG in CMJ and RE14.
Conclusion: Performing a 12-week concurrent training program integrated into the ATR periodization
system effectively improves body composition and performance variables that can be obtained with
exclusive running-specific strength and endurance training in recreational runners aged 30 to 40.
Running-specific strength training enhances maximum and explosive strength and RE, whereas
exclusive endurance training improves VO2max, AnT, and RE. Performing concurrent training on
non-consecutive days effectively prevents the strength and endurance adaptations attained with
single-mode exercise from being attenuated. The ATR periodization system is useful in improving
recreational endurance athletes’ performance parameters, especially when performing concurrent
training programs.
Keywords: concurrent training; endurance training; running-specific strength training; periodization;
recreational runner
1. Introduction
Several sports require an adequate levels of strength and endurance to perform at
optimum level in competitive events. However, successfully combining endurance and
strength training represents the highest complexity in exercise prescription [1]. It has often
Int. J. Environ. Res. Public Health 2022, 19, 10773. https://doi.org/10.3390/ijerph191710773
https://www.mdpi.com/journal/ijerph
Int. J. Environ. Res. Public Health 2022, 19, 10773
2 of 17
been speculated that concurrent training does not generate the same adaptations as single-
mode exercise [1]. Even so, the possible mechanisms whereby concurrent training of both
fitness components can attenuate strength and endurance adaptations remain unclear [2].
In endurance sports, it has traditionally been thought that cardiovascular capacity is
the main limiting factor in sports performance [3]. Therefore, maximum oxygen consump-
tion (VO2max) and anaerobic threshold (AnT) have been considered the best indicators to
predict athletes’ performance [4]. Nevertheless, in reality, endurance athletes with similar
VO2max may perform differently in sports competitions. Hence, VO2max could not be the
best indicator to predict their racing performance. Nowadays, running economy (RE) and
the evaluations that imply assessing the muscular power exerted or the speed reached by
an athlete during the VO2max are considered better sports performance indicators [5–7]. In
this way, specific scientific evidence indicates that combining endurance and strength train-
ing generates additional benefits in terms of athletic performance improvement and injury
prevention [8]. These improvements could be related to the following mechanisms [8–13]:
(a)
Musculotendinous factors: Improved muscle-tendinous stiffness and stretch-shortening
cycle properties, conversion of fast-twitch type IIx into more fatigue-resistant type IIa
fibers, and delayed activation of less-efficient type II fibers.
(b)
Neuromuscular factors: Improved neuromuscular function and efficiency, improved
intramuscular coordination, motor unit recruitment, and firing frequency.
(c)
Physical fitness components: Improved levels of strength. This allows athletes to
apply a lower relative percentage of force, which reduces the contribution of the
anaerobic energy system and results in reduced fatigue, maintenance of the required
application of strength over a longer period, or appliance of more strength per unit
of time. Additionally, peak velocity, speed, maximum aerobic speed, and anaerobic
capacity are enhanced.
All these enhancements would result in an improved RE.
Nevertheless, there are arguments against using concurrent training programs in
endurance runners. Vikmoen et al. (2016) and Berryman et al. (2018) [14,15] state that
sports requiring high strength levels are opposite in nature to endurance sports in terms of
energy metabolism and effort duration. Therefore, developing both fitness components
simultaneously would result in potential combative adaptations [14–16]. Actually, the main
adaptations produced by endurance and strength training are not only different but also
opposed: Endurance training adaptations include oxidative enzyme activity increment,
mitochondrial and capillary density increment, maintenance or reduction of fiber size, and
possibly also fiber type transformations (Type II into I), modifying the model of recruitment
and reducing the muscle contractile capacity [17]. In contrast, strength training is associated
with reduced capillary density, oxidative enzymes, and mitochondrial density, reducing
the oxidative muscle capacity [10,15].
In this regard, some studies have revealed a certain degree of incompatibility between
endurance and strength training [9,18,19]. Thus, maximum voluntary contraction, rate of
force development and some adaptations such as maximal strength and VO2max can be
attenuated [20]. Similarly, time to exhaustion and mitochondrial density can be reduced [14].
This phenomenon is known as the interference effect or concurrent training effect [11], and
the potential interferences can be chronic and acute [2,11].
The reasons why the interference effect occurs are [1,2,11] the mechanism of muscle
fiber recruitment used, the transformation of fast twitch muscle fibers into slow twitch
muscle fibers, and the functioning of the endocrine system. From the molecular point of
view, the simultaneous activation of cellular biomarkers that elicit optimal anabolic and
endurance responses is also not possible [11]. Moreover, muscle hypertrophy implies the
cross-sectional area increment of muscle fibers, which may increase the distance between
the capillaries inside the muscle and could negatively impact performance. Even so, in
untrained individuals, there is an increase in (or at least a maintenance of) the number
of capillaries surrounding each muscle fiber, and the capillaries per fiber area do not
experience modifications. However, since performing concurrent strength and endurance
Int. J. Environ. Res. Public Health 2022, 19, 10773
3 of 17
training can mitigate the hypertrophic response, and endurance-trained athletes have a
greater number of capillaries than untrained athletes, these findings may not apply to
experienced endurance athletes [14].
In line with the existence of arguments for and against the use of concurrent train-
ing, some studies have found improvements in sports performance after using strength
training in endurance athletes [6,11], whereas no beneficial effect was observed in other
research [14,21]. The discrepancies between studies may be due to the development of
different types of strength, external variables not directly related to the intervention, the
use of different training methods [12,22], or the application of training methods lacking
scientific rigor [23].
As a result, it is essential to continue searching for better strategies to improve athletic
performance by implementing strength training sessions on endurance training programs
and preventing the interference between both fitness components [11,16]. Future studies
should focus on minimizing the interference effect when concurrent training is applied,
and training load organization, type, order, and optimal application are crucial to attain
this goal. Further research is also needed to understand better the relationship between
strength training, anaerobic metabolism, and endurance sports performance. New studies
must be longer in duration, since the greatest increases in RE occur after implementing
training protocols of more than 24 sessions, and most of the existing studies are shorter [15].
Moreover, valid strength assessments through a range of different velocities must be used,
implementing adequate strength training programs over a long-term intervention period,
and using multi-joint strength, explosive-strength, or reactive-strength exercises due to
their superior functionally [6].
Future studies must also integrate the intervention design into a suitable periodization
system and apply sports training principles to synchronize all training contents [20]. In this
respect, to the best of our knowledge, at present, only one investigation has been conducted
to verify the effect of strength training on endurance runners by using a periodization
system [21]. Additionally, athletes’ training programs must be adapted to their personal
needs and abilities. Thus, according to the law of diminishing returns, individuals’ ability
to attain specific adaptations will depend on their training level. Less-trained subjects are
likely to obtain greater adaptations since they have a greater adaptation reserve. In contrast,
well-trained individuals need more demanding training stimuli to attain improvements
throughout the training process [22].
Importantly, most existing studies related to concurrent training have focused on
verifying the compatibility of simultaneous strength and endurance training. However,
very few of them have examined the influence of strength, and particularly of running-
specific strength training, on endurance performance [24]. Most studies examined the
interference of endurance on maximal strength and hypertrophy, but not the opposite [11].
In this context, the utility of strength training remains to be clarified for endurance athletes,
and research findings are often inconclusive [22]. For this reason, it is necessary to continue
investigating this topic [17]. In fact, sports scientists have studied new ways to enhance
biomechanics, technique, energy production, sports equipment, injury prevention, and
recovery. However, concurrent training and the effect of strength training in endurance
athletes is a very complex phenomenon that requires new research.
2. Objective
The main objective of the present study was to verify the effect of running-specific
strength training, endurance training alone, and concurrent training on physiological
performance and selected anthropometric variables in recreational runners. We also aimed
to ascertain if 3-day-per-week concurrent training performed on non-consecutive days
attenuates the strength, endurance, and anthropometric adaptations compared to strength
and endurance training in isolation.
Int. J. Environ. Res. Public Health 2022, 19, 10773
4 of 17
3. Materials and Methods
A quasi-experimental randomized study was conducted.
3.1. Subjects
Thirty male recreational endurance runners participated in the present research. They
were assigned into three different groups: a running-specific strength training group
(RSSTG) [age: 34.7 (2.36); height: 1.77 (0.04); weight: 67.05 (5.37); BMI: 21.27 (1.87)], an
endurance training group (ETG) [age: 35.1 (2.77); height: 1.78 (0.05); weight: 65.80 (4.38);
BMI: 20.68 (1.31)], or a concurrent training group (CTG) [age: 34.3 (2.37); height: 1.76 (0.03);
weight: 66.96 (4.48); BMI: 21.48 (1.03)]. RSSTG performed thrice-per-week strength training
program orientated to running on non-consecutive days. ETG performed thrice-per-week
endurance training on non-consecutive days. Finally, CTG underwent 3-days-per-week
concurrent training performed on non-consecutive days. CTG alternated the running-
specific strength training sessions performed by RSSTG and the endurance training sessions
underwent by ETG. The inclusion criteria were (a) be an active recreational runner; (b) able
to run one km in less than 4:30 min; (c) have practiced running at a recreational level for at
least the past five years before participating in the current research; (d) perform endurance
training regularly, with a weekly frequency of not less than three times a week, but no more
than five times a week; (e) have not performed endurance or strength systematic training
for the last year leading up to the research’s commencement; (f) non-smoker; (g) do not use
nutritional supplementation; (h) do not suffer from chronic diseases or ongoing injuries;
(i) aged between 30 and 40 years old.
Participants were asked not to modify their dietary habits or lifestyle during the
intervention process, and attendance was recorded. Subjects were required to attend
at least 90% of the training sessions to be included in the study. Similarly, participants
were informed that they could voluntarily withdraw from the study at any time. The
research was conducted according to the ethical principles of the Declaration of Helsinki. It
was approved by the Ethics Commission of Prešov University (Prešov, Slovakia) (ethical
clearance number: 2/2021). Similarly, all subjects who participated in this study were
required to submit written informed consent. Previously, they received a verbal and
written explanation about the experimental design and the potential risks and benefits of
participating in this research.
3.2. Randomization
A blocking design was used to avoid any possible bias when the subjects were allocated
into three different experimental groups and to ensure the trial’s proper randomization.
The blocking factor was the VO2max value obtained through the incremental load test.
According to the results obtained in this test, participants were allocated to one of the ten
blocks created. Athletes who obtained the best three marks were assigned to block one.
Athletes who obtained the fourth, fifth, and sixth-best marks were assigned to block two,
and so on. Afterward, each block’s three members were randomly assigned to one of the
three different experimental groups. Therefore, there was one member of each block in
each experimental group.
3.3. Training Protocol
The intervention lasted 12 weeks, and study participants performed three sessions
per week on non-consecutive days. The training protocol was designed using the ATR
block periodization system. The training intervention was divided into three mesocycles:
accumulation (first six weeks), transmutation (from week seven to week 10), and realization
(weeks 11 and 12). The duration of the training protocol was the minimum necessary to
attain the planned adaptations in accordance with the abilities developed in each mesocycle,
the sports discipline, and the characteristics of the athletes. The training methods applied
to RSSTG are shown in Table 1, and the training methods that ETG underwent are shown
in Table 2.
Int. J. Environ. Res. Public Health 2022, 19, 10773
5 of 17
Table 1. Training methodology that will be used with RSSTG.
Week
Training Parameters
1
I: 64% 1RM; S: 4; R: 14; RT: 2”; Ex: Squat, leg curl, calf raise
2
I: 69% 1RM; S: 4; R: 12; RT: 2′; Ex: Squat, leg curl, calf raise
3
I: 69% 1RM; S: 5; R: 12; RTS: 2′; Ex: Squat, leg curl, calf raise
4
I: 69%–69%–75%–75%–80% 1RM; S: 5; R: 12-12-10-10-8; RTS: 3′; Ex: Squat, leg curl,
calf raise
5
I: 69%–75%–80%–85% 1RM; S: 4; R: 12-10-8-6; RTS: 3′; Ex: Squat, leg curl, calf raise
6
I: 80% 1RM; S: 4; R: 8; RTS: 3′; Ex: Squat, leg curl, calf raise
7
6S of: Squat (6R at 80% 1RM) + hurdle hops (10R) + 2′30′′ running at 100% of MAS;
RTS: 5′
8
6S of: Squat (5R at 82% 1RM) + hurdle hops (10R) + 2′30′′ running at 100% of MAS;
RTS: 5′
9
6S of: Squat (4R at 84% 1RM) + extended bounds (cover 50 m alternating legs by doing
the lowest possible number of strides) + 2′15′′ running at 105% of MAS; RTS: 5′
10
6S of: = Squat (3R at 86% 1RM) + extended bounds (cover 50 m alternating legs by
doing the lowest possible number of strides) + 2′ running at 110% of MAS; RTS: 5′
11
Uphill running. Di: 200 m; I: 115% of MAS; Incl: 6%; S: 3; R: 5; RTR: 3′; RTS: 10′
12
Uphill running. Di: 200 m; I: 120% of MAS; Incl: 6%; S: 2; R: 5; RTR: 3′; RTS: 10′
I: intensity; S: sets; R: repetitions; RTS: resting time between sets; RTR: resting time between reps; 1RM: one-
repetition maximum; MAS: maximum aerobic speed; Di: distance; Incl: inclination.
Table 2. Training methodology that will be used with ETG.
Week
Training Parameters
1
Fartlek training. Du: 50′; I: 117–162 b.p.m.
2
Fartlek training. Du: 55′; I: 117–162 b.p.m.
3
Fartlek training. Du: 60′; I: 117–162 b.p.m.
4
Continuous training; Du: 55′; I: 135–139 b.p.m.
5
Continuous training; Du: 50′; I: 139–144 b.p.m.
6
Continuous training; Du: 45′; I: 144–149 b.p.m.
7
Extensive interval training (long intervals); I: 159–162 b.p.m.; 10S of 3′; RT: 2′
8
Extensive interval training (long intervals); I: 162–165 b.p.m.; 10S of 2′30′′; RT: 2′
9
Extensive interval training (medium intervals). I: 165–168 b.p.m.; 14S of 1′30′′; RT: 2′
10
Extensive interval training (medium intervals). I: 168–171 b.p.m.; 16S of 1′; RT: 2′
11
Repetition training. I: 180 b.p.m.; 5R of 3′. RT: 8′
12
Competition method. I: 100% of competition running pace; 1R; Di: 3.5 km
Du: duration; I: intensity; R: repetitions; RT: resting time; b.p.m.: beats per minute; Di: distance.
Likewise, CTG underwent 3-day-per-week concurrent training performed on non-
consecutive days, alternating the strength and endurance sessions carried out by RSSTG
and ETG. All training sessions were supervised by the same researcher: a One Physical
Education Bachelor’s Degree holder, expert in sports training, and with more than 20 years
of working experience in the sports field. Likewise, all training sessions were conducted in
the same fitness center to minimize external variables’ influence avoid compromising the
results’ validity and.
Int. J. Environ. Res. Public Health 2022, 19, 10773
6 of 17
3.4. Assessments
Physiological and selected anthropometric parameters were measured before (pre-
test) and after (post-test) applying the 12-week training protocol. The assessments were
conducted after a rest period of 48 h, between 5:00 p.m. and 7:00 p.m. The study participants
were asked to refrain from ingesting food or beverages three hours before testing. To avoid
the learning effect, a theoretical–practical training session was conducted one week before
the pre-test. The main researcher explained the testing protocols in detail. Thereupon, the
participants practiced the proper technique of execution of all tests to verify the correct
functioning of the equipment and procedures used in the assessment. The following
two-phase warm-up was performed before conducting the physical and physiological tests:
General warm-up: 10 min running at 60% of their theoretical maximum heart rate plus five
minutes of joint mobilization exercises. Specific warm-up: 2 × 10 vertical jumps, three sets
of squats (10 reps at 50% of their estimated 1RM, five reps at their estimated 70% 1RM, and
three reps at their estimated 80% 1RM), and one set of 20 m acceleration. The theoretical
1RM was estimated based on the information collected during the theoretical–practical
training session conducted one week before the pre-test. In addition, the subject’s BM, age,
and strength training experience were also taken into account.
Assessments included in the pre-test and post-test are detailed below:
•
Body mass (BM) and body mass index (BMI): Both were measured using a Seca digital
column scale, model 769 (Hamburg, Germany). Height was measured to the nearest
0.1 cm and body mass to the nearest 0.1 kg. Body mass and body mass index were
assessed by the same investigator, and the subjects were in bare feet.
•
Body fat percentage (BFP): BFP was obtained through the following equation [25]:
BFP = [(Σ of abdominal, subscapular, triceps, suprailiac, abdominal, thigh, calf)0.143]
+ 4.56. The plicometer used to measure the fat folds was a Harpenden Skinfold Caliper,
model FG1056 (Sussex, UK).
•
Lean mass (LM): LM was calculated by using the following formula: LM = Body Mass (kg)
− (Fat Mass (Kg) × BFP).
•
Countermovement jump (CMJ): CMJ was used to assess jumping ability and lower
body power due to its high reliability and validity [26]. The test was performed in
one indoor gymnasium on a dry and non-slippery surface. The device used was an
Optojump-next (Bolzano, Italy) connected to one laptop via USB, and the Microgate
software (Optojump software, version 3.01.0001) was also utilized. Participants were
not allowed to use the arm swing. Instead, they were required to keep their hands on
their hips while performing the test. The test started with a quick countermovement
action. When the knees were bent 90◦, athletes initiated the take-off and flight. During
the flight, they had to maintain their hips, knees, and ankles extended. Participants
were also required to jump vertically. Movements forward, backward, or sideways
were not allowed [27]. Each subject had two attempts, with three minutes rest between
each jump.
•
One-repetition maximum (1RM) squat: This test was used to measure lower body
maximum strength due to its validity, reliability, and applicability [28,29]. To conduct
the test properly, participants kept their trunks naturally upright. The bar was grasped
firmly with both hands, and it was also supported on the shoulders. The test started
with knees bent at 90◦ until the performer’s thighs were parallel with the ground.
Then, subjects regained the upright position, with the legs fully extended. The number
of attempts required to determine the 1RM of each subject was between two and four.
Only the attempts correctly performed were registered. The resting time between
attempts was three minutes.
•
Incremental load test: This laboratory test was used to assess the Anaerobic Threshold
(AnT), maximal oxygen consumption (VO2max), and running economy (RE). The
following protocols were implemented:
–
Protocol I: The objective was to determine the RE by using one treadmill (Cy-
bex 625T, Rosemont, IL, USA), one metabolic gas analyzer (Cosmed Srl, Albano
Int. J. Environ. Res. Public Health 2022, 19, 10773
7 of 17
Laziale, Rome, Italy), and one heart rate monitor (Polar H9 BLE Kempele, Finland).
The protocol comprised an 8 min submaximal test consisting of two, four-minute
stages. That is to say, each stage was four minutes in length to allow for oxy-
gen consumption and steady-state heart rate. The first stage was performed at
12 km/h, and the second at 14 km/h. These two speeds are within the intensity
range established by Barnes and Kilding (2015) to measure RE in recreational
athletes [30]. Then, to determine the RE values, the VO2 mean value of the last
minute of each stage was recorded [31].
–
Protocol II: This test was used to estimate the VO2max and AnT. The treadmill’s
initial velocity was set at 2 km/h slower than the subjects’ estimated 4 mmol
when the test started. Then, the speed was increased by 0.5 km/h every 30 s until
exhaustion [32].
3.5. Statistical Analysis
Data is presented using the format mean SD (standard deviation). The Shapiro–Wilk
test was used to contrast the normality of the variables and Levene’s test to verify the
homogeneity of variances. Sphericity assumptions were assessed using Mauchly’s test.
When those sphericity assumptions were violated, the Greenhouse Gessier correction was
applied. To determine the concordance between the pre-test and post-test measurements,
the interclass correlation coefficient (ICC) was calculated for all the assessed anthropometric
and performance parameters. ICC values were interpreted as follows: ICC ≤ 0.49, poor;
0.50 ≤ ICC < 0.75, moderate; 0.75 ≤ ICC < 0.9, good; ICC ≥ 0.9, excellent [33]. To verify
whether there were differences between groups in the baseline, a one-way ANOVA (analysis
of variance) test was conducted. To assess the training effects between groups (ETG vs.
RSSTG vs. CTG) and within groups (pre-test vs. post-test) on the anthropometric and
performance variables, a two-way repeated-measure ANOVA was performed. When
statistically significant p values were found (group by time interaction effect or significant
main effects of time or group), a post hoc pairwise comparison was conducted with
Bonferroni correction to identify those differences. The effect size was calculated using
Cohen’s d. Values of d < 0.2, d = 0.2, d = 0.5, and d = 0.8 were considered as trivial, small,
medium, and large effect sizes, respectively [34]. The level of significance established was
p < 0.05. The statistical analysis of the data was performed using the program IBM SPSS
V.26® computing (IBM Corp., Armonk, NY, USA).
4. Results
Once the normality and homoscedasticity of the data were verified, the one-way
ANOVA confirmed the absence of significant differences between the three experimental
groups at the baseline for all the anthropometric and performance variables. Likewise, the
ICC values obtained between the pre-test and the post-test in all assessed parameters were
higher than 0.9 for the three groups, which indicates an excellent reliability. Then, the two-
way repeated-measure ANOVA showed that there was no interaction effect, main effect of
time, or group for BM and BMI (see Table 3). Furthermore, the two-way repeated-measure
ANOVA revealed the existence of a group-by-time interaction effect for CMJ, 1RM squat,
RE12, RE14, VO2max, and AnT. A main effect of time was observed for the BFP, LM, CMJ,
1RM squat, RE12, RE14, VO2max, and AnT. Finally, a main effect of group was found for
CMJ and RE14 (see Table 3).
Subsequently, the Bonferroni post hoc comparison showed that ST obtained im-
provements significantly higher than ET enhancements in the following variables: CMJ
(p = 0.003; CI95 = 1.81–7.51), 1RM squat (p = 0.035; CI95 = 1.33–9.86), and RE14 (p = 0.046;
CI95 = 46.69–5036.38). ET results were significantly better than those obtained by ST in AnT
(p = 0.04; CI95 = 0.04–1.95). Finally, the improvements obtained by CT were significantly
higher than those attained by ET in CMJ (p = 0.002; CI95 = 1.21–4.26), and RE14 (p = 0.046;
IC95 = 47.47–5035.51).
Int. J. Environ. Res. Public Health 2022, 19, 10773
8 of 17
Table 3. Between-subjects comparisons of all the variables assessed: main effect of time, main effect
of group, and interaction effect.
Variable
Main Effect of Time
Main Effect of Group
Group per Time
Interaction Effect
F (1–9)
p
F (2–18)
p
F (2–18)
p
BM
0.31
0.591
0.24
0.785
0.55
0.584
BMI
0.30
0.596
0.86
0.440
0.47
0.632
BFP
13.24
0.005 *
0.42
0.662
3.92
0.055
LM
4.35
0.006 *
0.25
0.780
0.92
0.413
CMJ
42.93
<0.001*
7.48
0.004 *
18.62
<0.001 *
1RMsquat
216.38
<0.001 *
1.92
0.175
27.62
<0.001 *
RE12
194.51
<0.001 *
0.127
0.882
23.08
<0.001 *
RE14
85.14
<0.001 *
20.86
<0.001 *
23.95
<0.001 *
VO2max
52.59
<0.001 *
0.12
0.891
8,72
0.002 *
AnT
109.84
<0.001 *
1.39
0.275
37.31
<0.001 *
Legend: BM: Body mass; BMI: Body mass index; BFP: Body fat percentage; LM: Lean mass; CMJ: Counter-
movement jump; 1RM squat: One-repetition maximum squat; AnT: Anaerobic threshold; VO2max: Maximum
oxygen consumption; RE: Running economy; p: Level of statistical significance; F: Variation between sample
means/variation within the samples; *: Significant improvement between the pre-test and post-test.
As for the within-subject comparisons (see Table 4), RSSTG significantly improved
between the pre- and post-tests in CMJ (p < 0.001; IC95 = 3.42–4.13), 1RM squat (p < 0.001;
IC = 5.66–8.73), RE12 (p < 0.001; 5.66–8.73) and RE14 (p = 0.007; F: 0.23–1.11). The effect
size of these improvements was small in the case of RE12 and RE14, and large for CMJ and
1RM squat. ETG significantly improved between the pre- and post-tests in the following
parameters: RE12 (p < 0.001; IC95 = 1.93–2.59), RE14 (p = 0.015; F = 620.65–4464.61), VO2max
(p < 0.001; CI95 = 0.51–0.77), and AnT (p < 0.001; CI95 = 262–738). The effect size of these
improvements was small in the case of RE14, medium for VO2max, and large for RE12
and AnT. Additionally, CTG significantly improved its results between the pre- and the
post-tests in the following variables: BFP (p < 0.001; CI95 = 0.354–0.590), LM (p = 0.035;
CI95 = 0.5–1.12), CMJ (p < 0.001; CI95 = 1.72–2.47), 1RM squat (p < 0.001; CI95 = 3.10–4.10),
RE12 (p = 0.035; CI95 = 1.62–2.56), RE14 (p < 0.001; CI95 = 0.80–1.84), VO2max (p < 0.001;
CI95 = 1.57–3.02), and AnT (p < 0.001; 0.84–1.15). The effect size of these improvements was
small in the case of RE14, medium for BFP and RE12, and large for the CMJ, 1RM squat,
and AnT variables.
Table 4. Results obtained by the three experimental groups in the pre- and post-test in all the
variables assessed.
Variable
Group
Pre-Test
Post-Test
Cohen’s d
p
X
SD
X
SD
BM
RSSTG
67.11
5.37
67.49
5.32
0.076
0.159
ETG
65.80
4.38
65.62
4.30
0.037
0.330
CTG
66.96
4.48
67.10
4.31
0.066
0.172
BMI
RSSTG
21.27
1.87
21.38
1.76
0.066
0.190
ETG
20.68
1.31
20.63
1.34
0.051
0.645
CTG
21.47
1.03
21.52
1.08
0.098
0.171
BFP
RSSTG
15.28
1.03
15.04
1.16
0.199
0.152
ETG
15.09
0.95
14.86
0.87
0.233
0.051
CTG
15.18
1.07
14.32
1.15
0.721
<0.001 *
LM
RSSTG
56.77
4.58
57.31
4.71
0.115
0.110
ETG
55.84
3.59
55.81
3.73
0.006
0.108
CTG
56.73
4.29
57.50
4.21
0.141
0.018 *
Int. J. Environ. Res. Public Health 2022, 19, 10773
9 of 17
Table 4. Cont.
Variable
Group
Pre-Test
Post-Test
Cohen’s d
p
X
SD
X
SD
CMJ
RSSTG
33.29
1.71
37.07
1.81
2.141
<0.001 *
ETG
32.86
1.73
32.42
2.33
0.210
0.446
CTG
33.06
1.41
35.16
1.29
1.551
<0.001 *
1RM squat
RSSTG
83.10
4.51
90.30
4.57
1.571
<0.001 *
ETG
83.61
2.27
84.71
3.37
0.348
0.168
CTG
82.70
2.75
86.30
2.66
1.321
<0.001 *
RE12
RSSTG
41.63
3.07
42.41
3.01
0.272
<0.001 *
ETG
40.49
2.65
42.76
2.75
0.838
<0.001 *
CTG
41.61
3.08
42.55
2.71
0.783
<0.001 *
RE14
RSSTG
49.44
3.41
50.28
3.17
0.253
<0.001 *
ETG
48.92
3.13
50.08
3.33
0.346
<0.001 *
CTG
48.58
3.12
50.18
3.01
0.431
<0.001 *
VO2max
RSSTG
60.29
4.18
60.91
4.51
0.138
0.097
ETG
59.02
3.01
60.91
4.27
0.581
<0.001 *
CTG
58.67
3.25
60.98
3.36
0.207
<0.001 *
AnT
RSSTG
14.95
0.68
15.05
0.83
0.131
0.172
ETG
14.90
0.51
16.05
0.83
0.845
<0.001 *
CTG
14.65
0.58
15.70
0.42
2.749
<0.001 *
Legend: BM: body mass; BMI: body mass index; BFP: body fat percentage; LM: lean mass; CMJ: countermovement
jump; 1RM squat: one-repetition maximum squat; AnT: anaerobic threshold; VO2max: maximum oxygen
consumption; RE: running economy; p: level of statistical significance; *: significant improvement between the
pre-test and post-test.
5. Discussion
5.1. Anthropometric Parameters
None of the three experimental groups presented modifications to their BM in the
present study. Furthermore, the effect sizes of the modifications produced in this parameter
between the baseline and the post-test were trivial in all three cases. These results are
consistent with previous recent studies [12,32,35]. Thus, on the one hand, it can be expected
that the exclusive practice of endurance training may promote muscular catabolism and
increase mitochondrial density and activity. Therefore, these adaptations could reduce body
mass and body fat percentage [36]. On the other hand, strength training can potentially
increase lean tissue by increasing the release of anabolic hormones such as testosterone and
growth hormone [37,38]. However, in the present study, we understand that the absence of
significant modifications in BM may be due to the following reasons: Firstly, the duration
of the intervention period (12 weeks) could not be long enough to generate significant
variations in this parameter. Secondly, in RSSTG and CTG, the potential or expected
decrease in BM of the participants derived from the slight reduction in fat mass has been
hindered by the slight increase in LM. Thus, the net result would be a non-significant
alteration of athletes’ BM. Likewise, since the study participants were recreational athletes
who perform endurance training sessions regularly, to be able to attain significant decreases
in their body weight, it is plausible that they require not only more extended intervention
periods, but also significant increases in their weekly training frequency, volume, density,
and intensity.
Regarding ETG, the absence of a significant weight reduction might be related to the
participants’ regular practice of endurance training. Therefore, this group may need more
intense or prolonged workouts to significantly decreases BM. Importantly, there was no
increase in lean tissue in ETG but a slight reduction. This means that the slight decrease in
lean tissue of the participants included in this group, together with their slight reduction in
BFP, was of such small magnitude that it did not cause a significant reduction in BM. This
is in line with the trivial effect sizes observed in ETG in its reduction of BFP and LM.
Int. J. Environ. Res. Public Health 2022, 19, 10773
10 of 17
As for the BMI, there were no significant variations in any of the three experimental
groups. In the case of RSSTG and CTG, a certain increase in their BMI might be expected
due to the strength training practice and its potential anabolic effect. However, the effect
size of the BMI increase in both groups was trivial. These results are somewhat surprising,
since none of the study participants had previous experience in strength training. Moreover,
it must be taken into account that, although the objective of the strength training protocol
used in the present research was not designed to produce muscle hypertrophy (see Table 1),
strength training is likely to generate certain amount of hypertrophy, particularly in subjects
without previous strength training experience [39]. These results are more surprising in the
case of RSSTG, since they did not perform endurance training. Thus, the trivial increase
in RSSTG suggests that for strength training to generate significant increases in BMI, it is
necessary to use training methods specifically aimed at achieving this goal and, probably, a
longer intervention period. In ETG, as expected, the BMI did not increase but decreased.
However, the reduction was trivial. We understand this slight decrease is in line with
the training principle of specificity since endurance training is more likely to generate
reductions in BM and BMI rather than increases [36].
As far as the BFP is concerned, significant reductions were only observed in CTG.
These results coincide with the study carried out by Eklund et al. (2016) [40]. In con-
trast, the BFP remained unaffected in some studies after applying a concurrent training
protocol [12,32]. Furthermore, Blagrove et al. (2018c) conducted a systematic review to ana-
lyze the effects of adding strength training to the endurance training programs of medium-
and long-distance athletes. They observed that BFP is commonly unaffected [41]. In the
present study, we understand that only CTG significantly improved its BFP due to the
strength and endurance training combination. This could be because strength training can
increase basal metabolism [39,42], and endurance training may produce a significant caloric
expenditure [39,42,43]. Regarding ETG, despite the fact that endurance training effectively
reduced BFP in some previous studies [38,44], we understand that subjects included in this
group did not improve their BFP due to the absence of strength training, which implied
that they could not benefit from its potential capacity to increase basal metabolism. On
the contrary, we interpret that RSSTG could not significantly decrease their BFP since
they did not practice endurance training and could not benefit from the significant caloric
expenditure that endurance training produces.
Concerning the LM, only CTG significantly increased this parameter after undertaking
the 12-week-training protocol. This increase could be related to the following reasons. First,
a greater training variability was applied to this group. Second, only CTG significantly
decreased BFP. Therefore, even though CTG and RSSTG reduced their LM in absolute terms
at similar levels and with similar effect sizes, the increase in CTG in relative terms was
higher due to its greater reduction in BFP. Third, according to Coffey and Hawley (2017) and
Fyfe and Loenneke (2018), the practice of divergent exercise (i.e., strength and endurance) by
untrained or recreationally active individuals induced similar increased anabolic signaling
in skeletal muscle during the first weeks of training [1,22]. Not surprisingly, ETG did not
increase the LM. Unlike CTG and RSSTG, this group decreased its LM, but not significantly.
We consider that this could be related to the potential catabolic effect of aerobic exercise [36].
The LM results of the present study are consistent with those attained by Eklund et al. (2016)
and Vikmoen et al. (2020) [40,45]. However, our results only partially coincide with those
obtained by Vikmoen et al. (2017) since they observed that both concurrent strength–
endurance training and running-specific strength training are useful for increasing LM [46].
Furthermore, Beattie et al. (2017) did not observe any changes in LM in competitive distance
runners, probably because it is more difficult for well-trained subjects to attain adaptations
due to their lower reserve of adaptation [32].
5.2. Performance Variables
As expected, ETG did not improve the results in CMJ. This confirms that endurance
training does not significantly modify the marks obtained by recreational endurance ath-
Int. J. Environ. Res. Public Health 2022, 19, 10773
11 of 17
letes in CMJ. In contrast, RSSTG and CTG significantly improved their performance in CMJ.
In the first case, it was expected since RSSTG only performed running-specific strength
training sessions, and the subjects included in the present study had no previous strength
training experience. As for CTG, it has been verified that their improvements in CMJ
produced by the strength training performed were not attenuated despite the concomitant
strength and endurance training. Additionally, RSSTG and CTG obtained significantly
better results than ETG in CMJ. This proves that the interference effect does not occur with
the weekly training frequency used in the training protocol. The results of the present study
are consistent with the findings obtained by Fyfe, Bishop and Stepto (2014) [47]. These
authors state that there is no evidence supporting the interference effect theory. In this
regard, Coffey and Hawley (2017) add that despite chronic studies indicating that there is
robust evidence supporting that endurance training attenuates strength adaptations when
concurrent training protocols are applied, the underlying mechanisms of the mentioned in-
terference are unknown. Nevertheless, some studies have verified that endurance training
attenuates improvements in power, specifically in CMJ when concurrent training protocols
are implemented [45,48].
As for 1RM squat, as expected, RSSTG and CTG improved their marks between pre-
and post-test, probably because the study participants had no previous strength training
experience. Only the marks obtained by RSSTG were significantly higher than those
attained by ETG in the post-test. This could indicate that a higher frequency of strength
training provides additional benefits, since the number of strength sessions performed by
RSSTG was higher than the sessions performed by CTG. However, we understand that
the interference effect did not occur in CTG, because no significant differences between
RSSTG and CTG were observed in the post-test in 1RM squat. The present research results
are consistent with those obtained by Vikmoen et al. (2016), Vikmoen et al. (2017), and
Sousa et al. (2017) [14,46,49]. In all three cases, the utility of concurrent training to improve
the 1RM squat was verified.
As far as RE is concerned, the three groups significantly improved this parameter at 12
and 14 km/h. The improvements obtained in RE14 by CTG and RSSTG were significantly
higher than those achieved by ETG. In RSSTG, the improvements in RE could be related
to the attainment of certain adaptations [9,10,32,50]: (a) improved musculotendinous
stiffness of the lower extremities; (b) improved motor unit recruitment and synchronization
patterns; (c) improved intermuscular coordination and neural inhibition; (d) delayed
activation of less-efficient type II muscle fibers; (e) conversion of type IIx fibers into fatigue-
resistant IIa fibers; (f) facilitation of the optimal application of strength throughout the
entire training or competition; (g) reduction of the relative intensity that each particular
cycle of effort or sports technique represents for one athlete when overcoming the same
resistance; (g) improved ability to perform the same effort with lower oxygen consumption;
(h) improved ability to apply the same strength with less muscle mass; (i) improved reuse
of elastic energy in each stride. Therefore, attaining all of these physiological adaptations
could be the reason why RSSTG obtained significant improvements over ETG in RE14.
Regarding the results obtained by ETG in RE12 and RE14, we interpreted that the
improvements were the result of attaining certain adaptations [51]: (a) improved oxidative
capacity, which in turn is associated with better mitochondrial functioning, and leads to
a reduction in the use of the oxygen required to perform submaximal intensity efforts;
(b) improved buffering capacity of the skeletal muscles and hematological system. As for
CTG, we considered they simultaneously benefited from the adaptations that both strength
and endurance training provide to improve RE. This circumstance would explain why
CTG obtained significantly better results than ETG in RE14. Likewise, it is also under-
standable that the results obtained by ETG were significantly lower than those attained
by RSSTG and CTG in RE14 but not in RE12, since runners show greater RE values at
race pace [12]. In this regard, considering the characteristics of the athletes included in the
present research and their results in AnT (see Table 3), it is understood that their compe-
tition race velocity is close to 14 km/h. The results of the present research are consistent
Int. J. Environ. Res. Public Health 2022, 19, 10773
12 of 17
with the studies of Beattie et al. (2017), Blagrove et al. (2018b), Giovanelli et al. (2017), and
Li et al. (2019) [12,31,32,52]. In all four cases, the practice of concurrent training was effec-
tive in improving RE. Additionally, recent systematic reviews and meta-analyses confirmed
the efficacy of strength training in improving RE [6,15,41]. In contrast, some studies verified
that the implementation of concurrent training programs was not effective in enhancing
RE [14,53].
Regarding the VO2max, although the trainability of this variable could be conditioned
by genetic factors [54], ETG and CTG obtained significant improvements. Therefore, it
can be assumed that these improvements are training-specific adaptations. Likewise,
the fact that the improvements attained by ETG were not significantly higher than those
achieved by CTG suggests that no interference effect has occurred. The present study
results coincide with the research conducted by Patoz et al. (2021) [53]. However, in a
systematic review, Blagrove et al. (2018c) verified that VO2max is typically unaffected
after the application of concurrent training programs [41]. Therefore, the discrepancies
between studies could occur because the possibility of improving VO2max is genetically
conditioned [54]. Moreover, as expected, RSSTG did not improve the VO2max, probably
because this group did not perform endurance training sessions. In fact, few studies found
significant improvements in VO2max after the exclusive practice of strength training. In
this regard, Ozaki et al. (2013) conducted a review study to verify the effects of strength
training on increasing VO2max, and in only three out of the 17 studies analyzed were
significant improvements in VO2max registered. They also ascertained that the higher the
training level, the more difficult it is to improve the VO2max [55].
Finally, regarding AnT, both ETG and CTG improved this variable. We understand
that this improvement resulted from the training methods specifically designed to enhance
the AnT (see Table 2). Likewise, the results attained by ETG were significantly better than
those achieved by RSSTG. However, the absence of significant differences between ETG
and CTG reveals that no interference effect has occurred in CTG. Furthermore, as expected,
RSSTG did not significantly improve the AnT. In this case, we consider that the absence of
endurance training explains the non-achievement of significant improvements. Moreover,
few studies have examined the effects on AnT. Ferrauti et al. (2010) verified the absence of
significant differences in AnT between a concurrent and endurance training program in
isolation [35]. Likewise, Cragnulini (2016), after conducting one review article, concluded
that adding strength training to endurance athletes’ training programs does not have a
negative impact on AnT [17].
5.3. Overall Interpretation of the Results
The improvements attained by the three experimental groups are specific exercise
mode adaptations. Thus, RSSTG improved all strength parameters, ETG all the endurance
parameters, and CTG strength and endurance parameters, and only the concurrent training
program effectively improved body composition in 12 weeks. Additionally, the interference
effect did not occur for the strength, endurance, or anthropometric variables. This suggests
that the weekly frequency used in the training protocol of the present study prevents the
attenuation of adaptations in concurrent training protocols with respect to single-mode
strength or endurance exercise. In this regard, Pattison et al. (2020) point out that CMJ
is useful for analyzing the interference effect on neuromuscular improvements when
performing concurrent training programs [48], and the results obtained in the present
research by RSSTG in CMJ were not significantly better than those attained by CTG.
Furthermore, based on the improvements attained by RSSTG and CTG in 1RM squat
and CMJ, and also considering the large effect sizes obtained by both groups, it can be
inferred that athletes without previous strength training experience can obtain significant
improvements in key performance parameters due to their greater reserve of adaptation.
In this sense, Fyfe and Loenneke (2018) consider that untrained individuals have a greater
capacity to adapt to training stimuli than trained individuals, although their individual
genetic potential could also limit the possibility of obtaining improvements [22]. For this
Int. J. Environ. Res. Public Health 2022, 19, 10773
13 of 17
reason, Beattie et al. (2014) indicate that, for endurance athletes with lower levels of strength,
a general strength training program may be sufficient to improve their maximum strength,
explosive strength, and reactive strength [6]. However, athletes with higher strength
levels should perform explosive and reactive strength training programs to improve their
performance [6,56].
It is also noteworthy that, based on the improvements achieved by the three experi-
mental groups after the 12-week training program, the ATR periodization system seems to
be adequate to improve the performance of recreational endurance runners, mainly when
concurrent training programs are applied. However, it is also possible that intervention
periods longer than 12 weeks can be required to attain significant improvements in body
composition, mainly when a single-mode exercise is used. Regrettably, few studies used
concurrent training protocols integrated into periodization systems such as the ATR block
periodization system. In this regard, García-Manso et al. (2017) conducted one research
with recreational college-age subjects, using one block periodization for nine weeks. They
verified that concurrent and exclusive endurance training effectively improves sports
performance. However, no significant differences were found between both training proto-
cols [21].
Importantly, we must mention future lines of research. Despite several studies exam-
ining concurrent training programs’ effects on endurance athletes, many aspects remain
unclear. This circumstance is further aggravated by the important methodological differ-
ences that exist between studies and their limitations. Thus, future studies might consider
the following aspects:
(a)
The use of concurrent training programs might be oriented to ensure that neuro-
muscular adaptations positively impact athlete’s biomechanical and performance
parameters [57].
(b)
It could be interesting to investigate the effect of different running volumes combined
with explosive strength training [58].
(c)
Since an eight-week concurrent training program could not be long enough to improve
certain performance variables [35], long-term intervention periods might be applied.
Li et al. (2019) propose the use of training protocols longer than 16 weeks [31], and
Beattie et al. (2014) longer than six months [6].
(d)
Although Berryman et al. (2018) found that the beneficial effects of strength training
on endurance performance occur regardless of athletes’ level [15], some authors
propose conducting studies with different populations. Low et al. (2019) recommend
conducting research with women and people with different training status [59], and
Li et al. (2019) with groups of senior citizens and women [31].
(e)
Future training protocols might be integrated into periodization systems such as the
ATR model. In this vein, Berryman et al. (2018) point out that the use of periodization
strategies could help clarify the optimal moment to implement strength training
activities within the annual training plan [15].
(f)
Future studies might include training protocols combining different endurance train-
ing methods (i.e., fartlek, continuous training, interval training) [23].
(g)
Appropriate training methods and tests to develop and assess strength levels must be
used [6].
(h)
The number of strength training sessions per week might be between two and
three [41].
(i)
The training protocols used might be designed in accordance with the sports
training principles.
Finally, it is necessary to mention the study’s strengths and limitations. As for the
strengths, there are two noteworthy aspects. First, as well as including one concurrent
and one single-mode endurance training group, one running-specific strength training
group was incorporated. This circumstance (which is not usual in studies conducted
with endurance athletes) was useful to verify the adaptations that endurance runners
can attain with running-specific strength training and determine the possible existence of
Int. J. Environ. Res. Public Health 2022, 19, 10773
14 of 17
interference effects in CTG. Second, the number of weekly training sessions applied to the
three experimental groups was equated. In this regard, it should be noted that in several
previous studies, the concurrent group performed two or three additional weekly strength
sessions with respect to the endurance training group, which implies using a distinctive
number of sessions.
As for the limitations, first, the sample size was small. A larger sample would have
ensured greater representativity. Second, no time-trial test was conducted in the present
research since the study participants were not specialized in any specific distance. Finally,
we consider that if the intervention period had been longer, it would have been possible
that the three experimental groups had obtained additional improvements, particularly
in body composition. Furthermore, significant differences between groups could have
occurred in more anthropometric and performance variables at the post-test.
6. Conclusions
A concurrent training program of 12 weeks integrated into the ATR periodization
system is effective in enhancing body composition and selected sports performance param-
eters associated with the exclusive practice of strength training (maximum and explosive
strength) and endurance training (VO2max and AnT), in addition to RE in recreational
runners aged 30–40. The exclusive practice of running-specific strength training during
the same period also using the ATR design improves maximum and explosive strength
and RE, while the exclusive practice of endurance training using the ATR model improves
endurance parameters (VO2max and AnT) and RE. Thus, concurrent training is the most
time-efficient method to attain anthropometric and performance adaptations. Addition-
ally, a concurrent training program performed on non-consecutive days did not attenuate
the endurance and strength adaptations that can be attained with single-mode exercise.
However, it cannot be ruled out that a higher weekly training frequency may generate
interference effects.
The ATR periodization system improves the performance parameters of recreational
endurance athletes, especially when performing concurrent training. Likewise, in recre-
ational athletes without previous strength training experience, due to their greater reserve
of adaptation, concomitant running-specific strength and endurance training—in addition
to enhancing their body composition and relevant performance variables—produce large
improvements in maximum and explosive strength and further enhance RE.
7. Practical Applications
•
Performing exclusive strength or endurance training allows athletes only to attain
specific exercise mode adaptations.
•
Undertaking concurrent training programs allows athletes to obtain strength and
endurance adaptations. However, to this end, separating the training sessions by at
least nine hours is necessary to avoid significant interferences, or 24 h to guarantee to
a greater extent that the adaptations will not be attenuated [60].
•
Concurrent training programs should be integrated into a periodization model to attain
greater effectiveness. In this sense, the ATR block periodization system effectively im-
proves anthropometric and performance variables in recreational endurance athletes.
•
Long-term interventions of more than 12 weeks might be used. Intervention periods of
12 weeks are insufficient to attain improvements in anthropometric parameters when
single-mode exercise training is used.
•
Regarding the training load and the type of strength that must be developed, it is nec-
essary to adapt the strength training program to the athlete’s objectives, training level,
and previous training experience. Thus, a general strength training program might
improve sports performance in subjects without previous strength training experience.
However, in subjects who did not previously develop their maximum strength, it can
be leapfrogging using inappropriate workloads or developing types of strength with
lower residual effect (i.e., explosive strength, muscular endurance, reactive strength).
Int. J. Environ. Res. Public Health 2022, 19, 10773
15 of 17
It may also reduce their adaptation reserve unnecessarily and limit future improve-
ments. By contrast, endurance athletes with previous strength training experience
should develop explosive and reactive strength to enhance their performance.
Author Contributions: Conceptualization, P.P.-G.; methodology, P.P.-G.; validation, P.P.-G. and J.S.;
formal analysis, P.P.-G.; investigation, P.P.-G.; resources, P.P.-G.; data curation, P.P.-G.; writing—
original draft preparation, P.P.-G.; writing—review and editing, P.P.-G. and J.S.; supervision, J.S. All
authors have read and agreed to the published version of the manuscript.
Funding: The authors would like to thank Prince Sultan University, Riyadh, Saudi Arabia, for
supporting the article processing charges.
Institutional Review Board Statement: The research was conducted according to the ethical princi-
ples of the Declaration of Helsinki. It was approved by the Ethics Commission of Prešov University
(Slovakia) (ethical clearance number: 2/2021).
Informed Consent Statement: Informed consent was obtained from all subjects involved in the study.
Data Availability Statement: The data presented in this study are available on request from the
corresponding author.
Acknowledgments: The authors would like to thank Prince Sultan University, Riyadh, Saudi Arabia,
for their support in research.
Conflicts of Interest: The authors declare no conflict of interest.
References
1.
Coffey, V.G.; Hawley, J.A. Concurrent exercise training: Do opposites distract? J. Physiol. 2017, 595, 2883–2896. [CrossRef]
[PubMed]
2.
Leveritt, M.; Abernethy, P.J.; Barry, B.K.; Logan, P.A. Concurrent strength and endurance training. A review. Sports Med. 1999, 28,
413–427. [CrossRef] [PubMed]
3.
Bassett, D.R., Jr.; Howley, E.T. Limiting factors for maximum oxygen uptake and determinants of endurance performance. Med.
Sci. Sports Exerc. 2000, 32, 70–84. [CrossRef] [PubMed]
4.
Kravitz, L.; Dalleck, L.C. Physiological factors limiting endurance exercise capacity. IDEA Health Fit. Source 2002, 20, 40–49.
5.
Saunders, P.U.; Pyne, D.B.; Telford, R.D.; Hawley, J.A. Factors affecting running economy in trained distance runners. Sports Med.
2004, 34, 465–485. [CrossRef]
6.
Beattie, K.; Kenny, I.C.; Lyons, M.; Carson, B.P. The effect of strength training on performance in endurance athletes. Sports Med.
2014, 44, 845–865. [CrossRef]
7.
Paavolainen, L.M.; Nummela, A.T.; Rusko, H. Muscle power factors and VO2max as determinants of horizontal and uphill
running performance. Scand. J. Med. Sci. Sports 2000, 10, 286–291. [CrossRef]
8.
Gäbler, M.; Prieske, O.; Hortobágyi, T.; Granacher, U. The Effects of Concurrent Strength and Endurance Training on Physical
Fitness and Athletic Performance in Youth: A Systematic Review and Meta-Analysis. Front. Physiol. 2018, 9, 1057. [CrossRef]
9.
Rønnestad, B.R.; Mujika, I. Optimizing strength training for running and cycling endurance performance: A review. Scand. J. Med.
Sci. Sports 2014, 24, 603–612. [CrossRef]
10.
Blagrove, R.C.; Howe, L.P.; Howatson, G.; Hayes, P.R. Strength and Conditioning for Adolescent Endurance Runners. Strength
Cond. J. 2018, 1, 2–11. [CrossRef]
11.
Flores-Zamora, A.C.; Rodríguez, M.; Rodríguez-Blanco, Y. Physiologicals adaptations to concurrent strength and endurance
training. Review. Olimpia 2017, 14, 119–129.
12.
Blagrove, R.C.; Howe, L.P.; Emily, J.C.; Spence, A.; Howatson, G.; Pedlar, C.R.; Hayes, P.R. Effects of Strength Training on
Postpubertal Adolescent Distance Runners. Med. Sci. Sports Exerc. 2018, 50, 1224–1232. [CrossRef] [PubMed]
13.
Verkhoshansky, Y.; Verkhoshansky, N. Special Strength Training. Manual for Coaches; Verkhoshansky SSTM: Rome, Italy, 2011.
14.
Vikmoen, O.; Raastad, T.; Seynnes, O.; Bergstrøm, K.; Ellefsen, S.; Rønnestad, B.R. Effects of Heavy Strength Training on Running
Performance and Determinants of Running Performance in Female Endurance Athletes. PLoS ONE 2016, 11, e0150799. [CrossRef]
15.
Berryman, N.; Mujika, I.; Arvisais, D.; Roubeix, M.; Laurent, C.B. Strength Training for Middle- and Long-Distance Performance:
A Meta-Analysis. Int. J. Sports Physiol. Perform. 2018, 13, 57–63. [CrossRef] [PubMed]
16.
Cantrell, G.S.; Schilling, B.K.; Paquette, M.R.; Murlasits, Z. Maximal strength, power, and aerobic endurance adaptations to
concurrent strength and sprint interval training. Eur. J. Appl. Physiol. 2014, 114, 763–771. [CrossRef] [PubMed]
17.
Cragnulini, F.E. Entrenamiento de la fuerza en deportes de resistencia: ¿más certezas que dudas o más dudas que certezas?
Perspectivas en Educación Física [Strength training in endurance sports: More certainties than doubts or more doubts than
certainties? Perspectives on Physical Education]. Doc. Res. Notes 2016, 3. Available online: http://www.memoria.fahce.unlp.edu.
ar/art_revistas/pr.7306/pr.7306.pdf (accessed on 17 November 2021).
Int. J. Environ. Res. Public Health 2022, 19, 10773
16 of 17
18.
Aagaard, P.; Andersen, J.L. Effects of strength training on endurance capacity in top-level endurance athletes. Scand. J. Med. Sci.
Sports 2010, 20 (Suppl. 20), 39–47. [CrossRef]
19.
Izquierdo, M.; Häkkinen, K.; Ibáñez, J.; Kraemer, W.J.; Gorostiaga, E.M. Effects of combined resistance and cardiovascular training
on strength, power, muscle cross-sectional area, and endurance markers in middle-aged men. Eur. J. Appl. Physiol. 2005, 94, 70–75.
[CrossRef]
20.
Conceição, M.; Cadore, E.L.; González-Izal, M.; Izquierdo, M.; Liedtke, G.V.; Wilhelm, E.N.; Pinto, R.S.; Goltz, F.R.; Schneider, C.D.;
Ferrari, R.; et al. Strength Training Prior to Endurance Exercise: Impact on the Neuromuscular System, Endurance Performance
and Cardiorespiratory Responses. J. Hum. Kinet. 2014, 44, 171–181. [CrossRef]
21.
García-Manso, J.M.; Arriaza-Ardiles, E.; Valverde, T.; Moya-Vergara, F.; Mardones-Tare, C. Effects of concurrent strength and
endurance training on middle distance races. Cult. Sci. Sport 2017, 12, 221–227. [CrossRef]
22.
Fyfe, J.J.; Loenneke, J.P. Interpreting Adaptation to Concurrent Compared with Single-Mode Exercise Training: Some Method-
ological Considerations. Sports Med. 2018, 48, 289–297. [CrossRef]
23.
Boullosa, D.; Esteve-Lanao, J.; Casado, A.; Peyré-Tartaruga, L.A.; Gomes da Rosa, R.; Del Cose, J. Factors Affecting Training and
Physical Performance in Recreational Endurance Runners. Sports 2020, 1, 35. [CrossRef] [PubMed]
24.
Esteve-Lanao, J.; Rhea, M.R.; Fleck, S.J.; Lucia, A. Running-Specific, Periodized Strength Training Attenuates Loss of Stride Length
During Intense Endurance Running. J. Strength Cond. Res. 2008, 22, 1176–1183. [CrossRef] [PubMed]
25.
González-Gallego, J.; Sánchez-Collado, P.; Mataix, J. Nutrition in Sports and Ergogenic Aids and Doping; Editorial Díaz de Santos:
Madrid, Spain, 2006.
26.
Markovic, G.; Dizdar, D.; Jukic, I.; Cardinale, M. Reliability and factorial validity of squat and countermovement jump tests. J.
Strength Cond. Res. 2004, 18, 551–555.
27.
Heishman, A.; Brown, B.; Daub, B.; Miller, R.; Freitas, E.; Bemben, M. The influence of countermovement jump protocol on
reactive strength index modified and flight time: Contraction time in Collegiate Basketball players. Sports 2019, 7, 37. [CrossRef]
[PubMed]
28.
Levinger, I.; Goodman, C.; Hare, D.L.; Jerums, G.; Toia, D.; Selig, S. The reliability of the 1RM strength test for untrained
middle-aged individuals. J. Sci. Med. Sport 2009, 12, 310–316. [CrossRef] [PubMed]
29.
Verdijk, L.B.; Gleeson, B.G.; JonkerS, R.A.M.; Meijer, K.; Savelberg, H.H.C.M.; Dendale, P.; Van Loon, L.J.C. Skeletal muscle
hypertrophy following resistance training is accompanied by a fiber type-specific increase in satellite cell content in elderly men.
J. Gerontol. Ser. A Biol. Sci. Med. Sci. 2009, 64, 332–339. [CrossRef] [PubMed]
30.
Barnes, K.R.; Kilding, A.E. Running economy: Measurement, norms, and determining factors. Sports Med. Open 2015, 1, 8.
[CrossRef] [PubMed]
31.
Li, F.; Wang, R.; Newton, R.U.; Sutton, D.; Haiyong, Y.S. Effects of complex training versus heavy resistance training on
neuromuscular adaptation, running economy and 5-km performance in well-trained distance runners. PeerJ 2019, 7, e6787.
[CrossRef]
32.
Beattie, K.; Carson, B.P.; Lyons, M.; Rossiter, A.; Kenny, I.C. The Effect of Strength Training on Performance Indicators in Distance
Runners. J. Strength Cond. Res. 2017, 31, 9–23. [CrossRef]
33.
Koo, T.K.; Li, M.Y. A Guideline of Selecting and Reporting Intraclass Correlation Coefficients for Reliability Research. J. Chiropr.
Med. 2016, 15, 155–163. [CrossRef] [PubMed]
34.
Cohen, J. Statistical Power Analysis for the Behavioral Sciences, 2nd ed.; NJ Erlbaum Associates: Hillsdale, MI, USA, 1988.
35.
Ferrauti, A.; Bergermann, M.; Fernandez-Fernandez, J. Effects of a concurrent strength and endurance training on running
performance and running economy in recreational marathon runners. J. Strength Cond. Res. 2010, 24, 2770–2778. [CrossRef]
[PubMed]
36.
Herrera, J.M.C. Aerobic Exercise Effects On Body Composition, Cardiovascular Endurance, Circadian Cycle, Polar T3 Syndrome
in Colombia’s First Mission to Antarctica. Rev. Cienc. Poder Aéreo 2016, 12, 72–90.
37.
Suárez-Rodríguez, D. Testosterone and Growth Hormone: Systems Strength Training. Rev. Entren. Deport. 2016, 30, 9–19.
38.
Willis, L.H.; Slentz, C.A.; Bateman, L.A.; Shields, A.T.; Piner, L.W.; Bales, C.W.; Houmard, J.A.; Kraus, W.E. Effects of aerobic
and/or resistance training on body mass and fat mass in overweight or obese adults. J. Appl. Physiol. 2012, 113, 1831–1837.
[CrossRef] [PubMed]
39.
Haff, G.G.; Triplett, N.T. Essentials of Strength Training and Conditioning, 4th ed.; Human Kinetics: Champaign, IL, USA, 2016.
40.
Eklund, D.; Häkkinen, A.; Laukkanen, J.A.; Balandzic, M.; Nyman, K.; Häkkinen, K. Fitness, body composition and blood
lipids following 3 concurrent strength and endurance training modes. Appl. Physiol. Nutr. Metab. 2016, 41, 767–774. [CrossRef]
[PubMed]
41.
Blagrove, R.C.; Howatson, G.; Hayes, P.R. Effects of Strength Training on the Physiological Determinants of Middle- and
Long-Distance Running Performance: A Systematic Review. Sports Med. 2018, 48, 1117–1149. [CrossRef]
42.
Hughes, D.C.; Ellefsen, S.; Baar, K. Adaptations to Endurance and Strength Training. Cold Spring Harb. Perspect. Med. 2018,
8, a029769. [CrossRef]
43.
Hellsten, Y.; Nyberg, M. Cardiovascular Adaptations to Exercise Training. Compr. Physiol. 2015, 6, 1–32.
44.
Pérez-Gómez, J.; Vicente-Rodríguez, G.; Ara-Royo, I.; Martínez-Redondo, D.; Puzo-Foncillas, J.; Moreno, L.A.; Díez-Sánchez, C.;
Casajús, J.A. Effect of endurance and resistance training on regional fat mass and lipid profile. Nutr. Hosp. 2013, 28, 340–346.
Int. J. Environ. Res. Public Health 2022, 19, 10773
17 of 17
45.
Vikmoen, O.; Raastad, T.; Ellefsen, S.; Rønnestad, B.R. Adaptations to strength training differ between endurance-trained and
untrained women. Eur. J. Appl. Physiol. 2020, 120, 1541–1549. [CrossRef] [PubMed]
46.
Vikmoen, O.; Rønnestad, B.R.; Ellefsen, S.; Raastad, T. Heavy strength training improves running and cycling performance
following prolonged submaximal work in well-trained female athletes. Physiol. Rep. 2017, 5, e13149. [CrossRef] [PubMed]
47.
Fyfe, J.J.; Bishop, D.J.; Stepto, N.K. Interference between concurrent resistance and endurance exercise: Molecular bases and the
role of individual training variables. Sports Med. 2014, 44, 743–762. [CrossRef] [PubMed]
48.
Pattison, K.J.; Drinkwater, E.J.; Bishop, D.J.; Stepto, N.K.; Fyfe, J.J. Modulation of countermovement jump-derived markers of
neuromuscular function with concurrent vs. single-mode resistance training. J. Strength Cond. Res. 2020, 34, 1497–1502. [CrossRef]
49.
Sousa, A.C.; Marinho, D.A.; Gil, M.H.; Izquierdo, M.; Rodríguez-Rosell, D.; Neiva, H.P.; Marques, M.C. Concurrent training
followed by detraining: Does the resistance training intensity matter? J. Strength Cond. Res. 2017, 32, 632–642. [CrossRef]
50.
Mújika, I.; Rønnestad, B.R.; Martin, D.T. Effects of Increased Muscle Strength and Muscle Mass on Endurance-Cycling Performance.
Int. J. Sports Physiol. Perform. 2016, 11, 283–289. [CrossRef]
51.
Mayoralas, F.G.M.; Díaz, J.F.J.; Santos-García, D.J.; Castellanos, R.B.; Yustres, I.; González-Ravé, J.M. Running economy and
performance. High and low intensity efforts during training and warm-up. A bibliographic review. Arch. Sports Med. 2018, 35,
108–116.
52.
Giovanelli, N.; Taboga, P.; Rejc, E.; Lazzer, S. Effects of strength, explosive and plyometric training on energy cost of running in
ultra-endurance athletes. Eur. J. Sport Sci. 2017, 17, 805–813. [CrossRef]
53.
Patoz, A.; Breine, B.; Thouvenot, A.; Mourot, L.; Gindre, C.; Lussiana, T. Does Characterizing Global Running Pattern Help to
Prescribe Individualized Strength Training in Recreational Runners? Front. Physiol. 2021, 12, 631637. [CrossRef]
54.
Williams, C.J.; Williams, M.G.; Eynon, N.; Ashton, K.J.; Little, J.P.; Wisloff, U.; Coombes, J.S. Genes to predict VO2max trainability:
A systematic review. BMC Genomics 2017, 18, 831. [CrossRef]
55.
Ozaki, H.; Loenneke, J.P.; Thiebaud, R.S.; Abe, T. Resistance training induced increase in VO2max in young and older subjects.
Eur. Rev. Aging Phys. 2013, 10, 107–116. [CrossRef]
56.
Williams, J. Effect of explosive-based training on musculotendinous stiffness and running economy in highly-trained distance
runners. J. Aust. Strength Cond. 2020, 28, 86–92.
57.
Trowell, D.; Vicenzino, B.; Saunders, N.; Fox, A.; Bonacci, J. Effect of strength training on biomechanical and neuromuscular
variables in distance runners: A systematic review and meta-analysis. Sports Med. 2020, 50, 133–150. [CrossRef] [PubMed]
58.
Lum, D.; Tan, F.; Pang, J.; Barbosa, T.M. Effects of intermittent sprint and plyometric training on endurance running performance.
J. Sport Health Sci. 2019, 8, 471–477. [CrossRef] [PubMed]
59.
Low, J.L.; Ahmadi, H.; Kelly, L.P.; Willardson, J.; Boullosa, D.; Behm, D.G. Prior Band-Resisted Squat Jumps Improves Running
and Neuromuscular Performance in Middle-Distance Runners. J. Sports Sci. Med. 2019, 18, 301–315.
60.
Baldwin, K.M.; Badenhorst, C.E.; Cripps, A.J.; Landers, G.J.; Merrells, R.J.; Bulsara, M.K.; Hoyne, G.F. Strength Training for
Long-Distance Triathletes. Strength Cond. J. 2022, 44, 1–14. [CrossRef]
| Effects of Running-Specific Strength Training, Endurance Training, and Concurrent Training on Recreational Endurance Athletes' Performance and Selected Anthropometric Parameters. | 08-29-2022 | Prieto-González, Pablo,Sedlacek, Jaromir | eng |
PMC5408286 | ORIGINAL RESEARCH
Using a novel data resource to explore heart rate during
mountain and road running
Andrew Best1
& Barry Braun2
1 Department of Anthropology, University of Massachusetts, Amherst, Massachusetts
2 Department of Health and Exercise Science, Colorado State University, Fort Collins, Colorado
Keywords
Altitude, cardiac drift, hypoxia, Strava.
Correspondence
Andrew Best, Department of Anthropology,
University of Massachusetts, 217 Machmer
Hall, 240 Hicks Way, Amherst, MA.
Tel: +1 (860) 882-3678
Fax: +1 413-577-4217
E-mail: [email protected]
Funding information
No funding information provided.
Received: 2 March 2017; Revised: 23 March
2017; Accepted: 23 March 2017
doi: 10.14814/phy2.13256
Physiol Rep, 5 (8), 2017, e13256,
doi: 10.14814/phy2.13256
Abstract
Online, accessible performance and heart rate data from running competitions
are posted publicly or semi-publicly to social media. We tested the efficacy of
one such data resource- Strava- as a tool in exercise physiology investigations
by exploring heart rate differences in mountain racing and road racing run-
ning events. Heart rate and GPS pace data were gathered from Strava activities
posted by 111 males aged 21–49, from two mountain races (Mt. Washington
Road Race and Pike’s Peak Ascent) and two road race distances (half mara-
thon and marathon). Variables of interest included race finish time, average
heart rate, time to complete the first half (by distance) of the race, time to
complete the second half, average heart rate for both the first and second half,
estimated maximal heart rate, and competitiveness (finish time as percentage
of winning time). Mountain runners on average showed no change in heart
rate in the second versus first half of the event, while road racers at the half
marathon and marathon distances showed increased second-half heart rate.
Mountain runners slowed considerably more in the second half than road
runners. Heart rate increases in road races were likely reflective of cardiac
drift. Altitude and other demands specific to mountain racing may explain
why this was not observed in mountain races. Strava presents enormous
untapped opportunity for exercise physiology research, enabling initial inquiry
into physiological questions that may then be followed by targeted laboratory
studies.
Introduction
Heart rate and GPS recording devices have become a
common training tool for endurance athletes. Thousands
post running and cycling activities on social media ser-
vices
such
as
Strava
(www.strava.com),
Movescount
(www.movescount.com), and Training Peaks (www.train
ingpeaks.com). Of these, Strava has the largest cache of
public data (registration not required) and semipublic
data (free registration required). Strava enables access to
in situ data from thousands of athletic competitions that
would require significant time and effort to collect
through traditional research approaches. Standard data
posted in Strava activities include pace and elevation, and
sometimes heart rate and age. Some limitations are
inherent: potentially relevant information, such as body
mass, aerobic capacity, and training history are inaccessi-
ble without direct communication with individual ath-
letes, which violates Strava’s terms of use. Despite these
constraints
Strava
represents
an
untapped
“big-data”
source for exercise physiology research. Here, we explore
the efficacy of this novel investigative approach through a
comparative study of heart rate profiles of road running
and mountain running competitions, events for which
sufficient data is available on Strava.
Mountain running may be broadly defined as running
or run/hiking over mountainous terrain and differs from
traditional road racing in grade, altitude and terrain. The
Mt. Washington Road Race (Gorham, NH) and the Pike’s
Peak Ascent (Manitou Springs, CO) are analogous in
ª 2017 The Authors. Physiological Reports published by Wiley Periodicals, Inc. on behalf of
The Physiological Society and the American Physiological Society.
This is an open access article under the terms of the Creative Commons Attribution License,
which permits use, distribution and reproduction in any medium, provided the original work is properly cited.
2017 | Vol. 5 | Iss. 8 | e13256
Page 1
Physiological Reports ISSN 2051-817X
duration (but not distance) to the half marathon and
marathon, respectively, traditional distances for road run-
ning competitions. Differences in heart rate and perfor-
mance data between these races, gathered from Strava
activities, may provide clues to the physiological demands
unique to these two disciplines. Thus, a secondary goal of
this study is to infer physiological differences between
mountain and road running from studying heart rate
profiles during these events. If identified, effects of differ-
ential biomechanics, cardiovascular physiology, and/or
hypoxia could be studied directly in more targeted experi-
mental studies.
Methods
Mountain races sampled include the Mt. Washington
Road Race (MWRR, 2009–2016) and the Pike’s Peak
Ascent or the ascent split from the Pike’s Peak Marathon
(PP, 2012–2016), held the same weekend over the same
course but with a return downhill run. Candidate road
races were identified by searching Strava for races with
large samples of posted data. Chosen races included the
Hartford Half Marathon (2011–2015), the Philadelphia
Half Marathon (2011–2015), the Boston Athletic Associa-
tion Half Marathon (2015), and the New York City Mara-
thon (2015). It was necessary to sample from several half
marathons over multiple years to achieve a sufficient sam-
ple size, and each of these races was chosen because these
race courses are not excessively hilly nor is the second
half of the race substantially more difficult than the first
half. For any race, data were excluded for years where the
ambient temperature was excessively warm. Also excluded
were Strava activities that showed unrealistic HR profiles,
such as precipitous changes, long periods of total stasis,
or rapid fluctuations, all of which suggest heart rate mon-
itor malfunction.
Most data were accessed through fully public or
semipublic Strava pages (those requiring free site member-
ship), but several runners submitted GPX data files directly
after responding to Facebook requests and completing an
informed consent document. Per Strava’s request, Strava
users were not contacted. Data were anonymized and par-
ticipants were assigned study subject ID’s. Heart rate and
GPS data from 111 males aged 21–49 (mean 34.5 6.4)
were included in this study (see Table 1). Age, race finish
time, winning time, and status of altitude residence (de-
fined here as living at 3500’ or higher) were determined by
accessing race results posted on race websites. Variables of
interest included: race finish time, age and residence (ob-
tained from official race results); average heart rate (HR)
throughout the race; time to complete the first half (by
distance) of the race; time to complete the second half;
and average HR for both the first and second half. HRmax
was estimated using the Tanaka et al. (2001) equation for
endurance trained men, 205—(0.6 x age). From these data
other measures were calculated, including percentage of
winning time (a measure of competitiveness), time differ-
ence to complete the first versus second half, heart rate
difference in beats per minute (bpm) for the first versus
second half, heart rate difference as a percentage of HRmax
for the first versus second half, and overall heart rate as a
percentage of HRmax. Data were analyzed using SPSS sta-
tistical software v. 22 (IBM). T-tests were performed for
duration-matched races and ANOVAs were used to exam-
ine differences between all test groups. Linear regressions
were used to explore relationships between variables of
interest. Finally, three additional MWRR competitors for
whom heart rate data were not available were included to
create a subgroup of four altitude-acclimatized MWRR
runners. All study protocol were reviewed and endorsed
by the University of Massachusetts Human Subjects
Review Board.
Table 1. Characteristics of participants in each race, mean SD.
All
MWRR
½ Marathon
Pikes Peak
Marathon
MWRR altitude
subgroup
N
111
19
21
21
50
4
Age (years)
34.5 6.4
35.0 5.4
32.9 6.7
37.0 7.5
33.9 5.9
31.8 5.2
Race duration
(hr:min:sec)
–
1:23:45 0:08:20
1:25:42 0:04:52
2:59:33 0:11:47
2:54:43 0:07:20
1:05:31 0:10:073
% of winning
time
136.1 9.5
142.8 13.81,2
138.2 10.1
133.1 8.81
133.9 5.62
111.3 16.63
1Significant difference (P < 0.01) between MWRR and Pikes Peak.
2Significant difference (P < 0.01) between MWRR and the Marathon. Runners in the shorter races, considered together, were slightly less com-
petitive than those in longer races (P < 0.01). MWRR and the ½ marathon were significantly shorter in duration than Pikes Peak and the mara-
thon (P < 0.0001).
3MWRR altitude subgroup had significantly lower % of winning time and race duration than all other groups (P < 0.01).
2017 | Vol. 5 | Iss. 8 | e13256
Page 2
ª 2017 The Authors. Physiological Reports published by Wiley Periodicals, Inc. on behalf of
The Physiological Society and the American Physiological Society.
Heart Rate during Mountain Running
A. Best et al.
There were no significant differences in age between test
groups nor significant differences in duration for matched
races- MWRR (1:23:45 0:08:20) versus the half mara-
thon (1:25:42 0:04:52), and PP (2:59:33 0:11:47) ver-
sus the marathon (2:54:43 0:07:20). Only one MWRR
participant of 19 lived at altitude (in the initial sample)
while only three of 21 PP runners did not. Runners’ finish
times ranged from 112% to 174% of the respective race
winning time (mean 136 9.5) and this measure was not
different between duration and matched races. MWRR
runners, however, were less competitive by this measure
than PP and marathon runners (P < 0.01) and sampled
runners in the shorter races considered together (MWRR
and the ½ marathon) were slightly less competitive than
the longer races (PP and the marathon; P < 0.01). Still,
overall the runners in this study can be described as recre-
ationally competitive. For example, most of the marathon-
ers finished in under 3 h, a common benchmark of
competitiveness, and the fastest ran 2 h 34 min, a perfor-
mance that would earn a top-10 finish in most American
marathons outside of the major city marathons (Boston,
New York, and Chicago). The subgroup of four MWRR
runners from altitude was markedly more competitive than
the larger study groups: two were race winners and one fin-
ished in second place, giving this subgroup an average fin-
ish time relative to the winner of 111.3%, significantly
faster than any test group (P < 0.01).
Results
Heart rate increased over the second half of both road
events: half marathoners saw a 3.7 bpm increase (P < 0.01)
and marathoners a 1.8 bpm increase (P < 0.05). There was
no change in heart rate in the second versus first half of
MWRR and HR decreased by 4.4 bpm at PP, though not
quite significantly (P = 0.056; see Table 2 and Fig. 1).
These differences are also reflected in percentage estimated
HRmax. Heart rate change was highly variable in the PP
sample (SD=9.9 bpm) and there was a dramatic outlier
whose HR dropped 34 bpm over the second half while
slowing about the same as the average PP runner (see
Fig. 2). When both mountain races were compared against
both road races, the former were found to have signifi-
cantly greater slowdown in the second half (P < 0.001) and
a significantly different second half HR change (P < 0.01;
see Table 3). This was true in duration-matched pair com-
parisons as well: MWRR had greater second-half slowing
and less of a second-half HR increase than the ½ marathon
(P < 0.05), while PP had greater second-half slowing and a
drop in HR over the second half, as compared with the
marathon where HR increased (P < 0.01). The four accli-
matized runners comprising the MWRR altitude subgroup
had similar second half slowing to other MWRR runners
(15.4% 1.0 vs. 11.5% 4.8), significantly more than
half
marathoners
and
marathoners
(P < 0.05)
and
significantly less than PP runners (P < 0.001). Runners in
the
shorter
races
(MWRR
and
the
half
marathon)
displayed higher overall HR (bpm and as percentage
estimated HRmax; P < 0.05) and less second-half slowing
(P < 0.01).
Age was positively, though very weakly, correlated with
slowing in the second half when all races were analyzed
together (r2=0.054; P < 0.05). No significant relationship
was found in individual races. Percentage of winning time
was positively and weakly correlated with HR for the
marathon (r2=0.085; P < 0.05) and inversely correlated
with HR change in the second half of MWRR, both in
bpm and percent estimated HRmax (r2=0.207; P = 0.05);
that is, less competitive runners had a smaller HR
increase or larger decrease in the second half at these
Table 2. HR and pace results, mean SD.
MWRR
½ Marathon
Pikes Peak
Marathon
MWRR altitude
subgroup
HR (bpm)
168.9 8.3
171.1 7.7
164.1 9.8
166.5 9.3
–
HR % estimated
max
91.8 4.7
92.4 4.2
89.2 4.7
90.2 5.1
–
Second half %
slower
11.5 4.81
4.3 8.4
51.2 8.82
5.7 5.4
15.4 1.03
HR change bpm
0.4 4.3
3.7 5.44
4.4 9.95
1.8 6.14
–
Change % est. HRmax
0.2 2.3
2.0 2.94
2.4 5.45
1.0 3.3
–
1MWRR slowed more than ½ marathoners (P < 0.01) and marathoners (P = 0.01).
2PP runners slowed more than all other groups (P < 0.0001).
3MWRR altitude runners slowed more than ½ marathoners and marathoners (P < 0.05).
4½ Marathoners’ (P < 0.01) and marathoners’ (P < 0.05) HR increased in the 2nd half.
5PP runners’ HR decrease was significantly different from the marathoners’ (P < 0.01) and ½ marathoners’ (P = 0.001) HR increase.
ª 2017 The Authors. Physiological Reports published by Wiley Periodicals, Inc. on behalf of
The Physiological Society and the American Physiological Society.
2017 | Vol. 5 | Iss. 8 | e13256
Page 3
A. Best et al.
Heart Rate during Mountain Running
races. Considered all together, runners who slowed more
in the second half of their race had less of a HR increase
or a greater decrease both in bpm and percent estimated
HRmax (r2=0.187 and 0.189, respectively; P < 0.001). This
relationship was significant only for MWRR runners
(r2=0.222 and 0.221, P < 0.05) and marathon runners
(r2=0.371 and 0.373, P < 0.001); there was almost no cor-
relation at PP or the half marathon.
Discussion
Our primary objective was to pilot the use of Strava’s
publicly and semi-publicly available data in an exercise
physiology investigation. The primary strength of this
approach is that large quantities of data are available from
a variety of competitions and training activities, enabling
initial inquiry into questions that would otherwise be dif-
ficult to test. Questions dealing with relative performance
will likely be more amenable to this approach than inves-
tigations of the underlying physiology as performance
data are abundant but physiological data from Strava
activities are limited. Indeed, heart rate is the only physi-
ological measurement available; without knowledge of
each subject’s maximal heart rate, VO2-max, blood lactate
values and oxygen consumption during the event, physio-
logical differences can only be inferred. Predictive equa-
tions based on age are routinely used to estimate HRmax,
but even the best of these (specific to endurance-trained
subjects and as used in this study) explains only 53% of
the variation in HRmax between subjects (Tanaka et al.
2001). We could not control for effects of training history
and runners unaccustomed to the specific demands of
prolonged uphill running will surely respond and perform
differently than well-prepared competitors, a point that
will be discussed further.
There are several ways in which Strava’s utility to a
researcher could be improved. First, recruiting users to
record and submit heart rate and GPS data specifically
for study purposes, which is currently against Strava’s
terms of use, could increase sample size and permit col-
lection of additional information, including training char-
acteristics.
However,
this
would
require
extensive
recruitment,
obviating
the
primary
strength
of
the
approach piloted here (easy data access). Second, Strava
or other athletic social media services may choose to
incorporate a running power feature, allowing uploading
of data from running power meters such as Stryd (www.
stryd.com), a relatively new device which estimates power
in watts; if accuracy is validated this could be a useful
measure. Other services dedicated to tracking and analyz-
ing athlete data—such as Movescount or Training Peaks,
mentioned previously-—may provide additional metrics,
but at present these services do not host a public or
semipublic data cache as large as Strava’s. Finally, poten-
tial
physiological
differences
identified
through
this
approach could be explored further with targeted labora-
tory-based studies where variables such as training his-
tory, climate, and terrain could be controlled and direct
physiological measurements could be collected.
This study is also, to our knowledge, the first to demon-
strate heart rate differences between mountain running and
road running events. Compared to duration-matched road
HR difference (bpm)
20
10
0
–10
–20
–30
–40
MWRR
1/2 marathon
Pikes peak
Marathon
Figure 1. HR difference in the second versus first half of each race
in beat per minute (bpm). Boxplots show first, second, and third
quartiles, minimum and maximum, and outliers.
Second half % slower
80.0
60.0
40.0
20.0
.0
–20.0
HR change (bpm)
20
10
0
–10
–20
–30
–40
Marathon
PP
1/2 marathon
MWRR
Figure 2. Relationship between slowing in the second half and HR
change in the second half. Significant negative correlation for
MWRR (r2=0.224, P < 0.05) and the marathon (r2=0.370,
P < 0.0001).
2017 | Vol. 5 | Iss. 8 | e13256
Page 4
ª 2017 The Authors. Physiological Reports published by Wiley Periodicals, Inc. on behalf of
The Physiological Society and the American Physiological Society.
Heart Rate during Mountain Running
A. Best et al.
races, participants in mountain races experienced no
increase (MWRR) or a reduction (PP) in HR (both
absolute and as percentage estimated HRmax) and slowing
pace over the second half of the race. The observed
increases in HR in the second versus first half of the half
marathon and marathon are not surprising: cardiac drift,
an increase in HR without a concomitant increase in work
output (i.e., running speed), has been demonstrated in
one-hour (Ekelund 1967; Mognoni et al. 1990) and 4-hour
(Dawson et al. 2005) exercise tests and is likely resultant
from lower stroke volume concomitant with reduced
plasma volume (Hamilton et al. 1991) and lower diastolic
function (Dawson et al. 2005). There is little reason to
speculate that these factors do not affect runners at MWRR
and PP— races of similar duration to the half marathon
and marathon, respectively— so maintenance of heart rate
in those events must be attributable to a factor sufficiently
robust that it masks the effect of drift. Three potential
explanatory factors, not mutually exclusive, warrant further
consideration.
Altitude
The PP Ascent begins at 2382 m (7814 feet) and finishes
at 4302 m (14 115 feet) while the Mt. Washington Road
Race starts at 465 m (1526 feet) and finishes at 1917 m
(6289 feet). The marathon and half marathons included
in this study are held near sea level. Performance and car-
diac function are undoubtedly impacted by the hypoxia
encountered throughout the PP race: acute altitude expo-
sure at simulated 4000 m (13 123 feet) increases heart
rate and cardiac output during submaximal exercise to
compensate for reduced arterial partial pressure of oxygen
(PaO2), but maximal HR is reduced slightly, possibly due
to reduced oxygen delivery to cardiac tissue (Stenberg
et al. 1966) or increased production of epinephrine,
which may have additional performance effects as it
speeds uptake of glucose into muscle cells (Richardson
et al. 1998). Wehrlin and Hallen (2006) found a 1.9 bpm
decrease in maximal HR per 1000 m–in keeping with
Stenberg et al.’s (1966) result at 4000 m– beginning at
least as low as 1000 m, encompassing most of the alti-
tudes encountered at MWRR. Thus, perhaps the lack of
HR increase observed in mountain races does not reflect
a reduction in percent HRmax (reduced aerobic effort),
but rather lower HRmax resultant from hypoxia opposes
cardiac drift and attenuates (MWRR) or completely
negates (PP) a rise in heart rate in the second half of the
race. Importantly, most participants in the PP Ascent live
at and are ostensibly acclimatized to altitude, so many of
their physiological responses during the race are not
directly comparable with nonacclimatized runners; how-
ever, some HR effects persist even after acclimatization
(Vogel et al. 1967).
Acclimatization also affords these runners a buffer
against altitude-induced performance decrements and thus
a performance advantage relative to sea-level runners
(Mahe et al. 1974; Fulco et al. 2000). Most participants in
the Mt. Washington Road Race are sea-level residents and
so any hypoxia experienced during the event is novel and
acute. Little work has been done to evaluate HR responses
to altitudes below 4000 m, but Wehrlin and Hallen’s
results (2006) suggest that Mt. Washington’s altitude is
sufficient to decrease maximal HR, as has been observed at
altitudes equivalent to PP. Regardless, Mt. Washington’s
elevation should certainly incur a performance penalty,
especially for non-acclimatized runners. Reductions in
VO2-max have been observed relative to sea level values
beginning at low altitudes: 580 m (Gore et al. 1996, 1997)
and even right from sea level (Wehrlin and Hallen 2006).
Thus, VO2-max decreases linearly up to 3000 m, an
impairment that appears to be more severe for endurance-
trained individuals (Lawler et al. 1988; Koistinen et al.
1995). This effect is not resultant from reduced maximal
exercise intensity achieved in altitude tests: Wehrlin and
Hallen (2006) found that performance, measured as time
to exhaustion in running tests at simulated altitudes with
speed kept constant, followed an observed 6% VO2-max
decrease per 1000 m. As the second half of MWRR
ascends to an altitude 730 meters higher than the first half,
Table 3. HR and pace for mountain versus road races and shorter versus longer races.
Mountain races
Road races
Shorter races
Longer races
HR (bpm)
166.4 9.3
167.9 9.0
170.0 8.03
165.8 9.4
HR % estimated max
90.5 4.8
90.9 4.9
92.1 4.43
90.0 4.9
Second half % slower
32.4 21.31
5.2 6.4
7.7 7.83
19.1 21.9
HR change bpm
2.1 8.02
2.4 5.9
2.1 5.1
0.0 7.9
Change % est. HRmax
1.1 4.42
1.3 3.2
1.2 2.8
0.0 4.3
1Mountain races had greater second half slowing than road races (P < 0.0001).
2HR change in bpm and as % est. HRmax was significantly different in mountain versus road races (P < 0.001).
3Shorter races were characterized by higher HR (bpm and % est. HRmax; P < 0.05) and less second half slowing (P < 0.01) than longer races.
ª 2017 The Authors. Physiological Reports published by Wiley Periodicals, Inc. on behalf of
The Physiological Society and the American Physiological Society.
2017 | Vol. 5 | Iss. 8 | e13256
Page 5
A. Best et al.
Heart Rate during Mountain Running
we may predict a 4.3% VO2-max penalty imposed by alti-
tude in the second half, explaining part of the observed
11.5% slowing.
It is also possible that increasing altitude throughout
both mountain races led to a slowing of pace while effort
remained relatively unchanged. Enormous slowing at PP
(51.2% 8.8) may be explained in part by the more
extreme altitude and the increasing difficulty of the footing
on the racecourse. However, as slowing was not associated
with a decrease in HR for PP (r2=0.001), increasing techni-
cal difficulty alone is unlikely to explain the vast discrep-
ancy between PP slowing and marathon slowing (5.7%).
MWRR is run on pavement, eliminating terrain as a com-
plicating variable, so a direct comparison with the half
marathon is easier- and indeed MWRR runners slowed far
more than half marathoners (11.5% vs. 4.3%). Slowing
pace at MWRR explains only 22% of the variability in HR
decrease, so other factors must be operative.
Pacing and psychological factors
An alternative or additional explanation for lack of HR
increase together with slowing pace during mountain
races is that runners reduce their effort as the race pro-
gresses due to psychological factors, or perhaps runners
are not as adept at pacing themselves evenly in these
events. This interpretation is supported by a weak but sig-
nificant correlation between second-half slowing and HR
change at MWRR (r2=0.202; P < 0.001): runners who slo-
wed more had a greater decrease, or smaller increase, in
HR compared with those who slowed less. However, 80%
of the variation in HR change is not explained by slowing
pace, and there is almost no correlation at all between
these variables for PP (r2=0.001). Additionally, the four
altitude-acclimatized MWRR athletes- three of whom are
world class mountain runners (one is a world champion)-
slowed just as much as their fellow MWRR competitors.
This may suggest that altitude is not responsible for the
observed slowing (and perhaps HR) effects as altitude
acclimatization afforded no buffering against second-half
slowing. Alternatively, the fact that highly trained and
experienced mountain runners slowed just as much as
everyone else may suggest that pacing and psychological
factors alone cannot account for slowing and HR changes,
as these runners should be expected to be expertly pre-
pared for the physical and psychological demands of
mountain racing. Also, as acclimatization simply mitigates
but does not eliminate altitude-incurred diminishments
in aerobic capacity, and VO2-max declines linearly begin-
ning from sea level, we may not expect acclimatized run-
ners to slow less over the second half but rather to simply
experience less of a performance declination overall rela-
tive to un-acclimatized athletes.
Muscle recruitment and biomechanical
factors
The road races included in this study climb and descend
no more than several hundred feet in total, while MWRR
ascends about 4600’ at an average grade of 12% and the
PP Ascent climbs about 7800’, also averaging 12% in
grade. The biomechanics of uphill running differ signifi-
cantly from level running: less eccentric work is per-
formed by muscles and tendons, and none above 15%
grade (Minetti et al. 1994), contributing to a higher
energy cost. At grades steeper than 15% (which are briefly
encountered at MWRR and PP) slopes of cost of trans-
port for walking and running converge (Minetti et al.
1994), and above 28% grade, walking is more efficient
than running (Giovanelli et al. 2016). Of course, in a
race, efficiency is second to the primary goal of covering
the course as quickly as possible. Many participants in
mountain racing events, especially the slower racers,
employ a mix of running and walking. So different are
uphill biomechanics that the traditional definition of run-
ning gait may need to be modified to encompass locomo-
tion lacking a true flight phase, but characterized by a
bouncing gait instead of the inverted pendulum motion
of walking. Such a gait has been described as “Groucho
running” in a study of bent-knee running on a level
treadmill (McMahon et al. 1987) and “grounded run-
ning” in a study of ostriches (Rubenson et al. 2004), but
these terms aptly describe the slower ranges of uphill
human running (Giovanelli et al. 2016).
How do the incline-specific biomechanics encountered
in mountain racing affect physiology? Balducci et al.
(2016) found that ten elite French mountain runners each
achieved the same VO2-max, blood lactate concentrations
and heart rate in maximal tests on level ground, 12.5%
slope, and 25% slope. Further, incline running perfor-
mance was poorly predicted by level running perfor-
mance,
and
there
was
significant
inter-individual
variation in energy cost increase from level to uphill run-
ning: moving from 0% to 12.5% incline increased energy
cost 50% for some subjects and 104% for others. These
results inform the present study by suggesting that, in
uphill-trained subjects, (1) uphill racing absent altitude
effects should not be characterized by different heart rate
profiles and aerobic capacities; and (2) uphill running
imposes unique challenges and specific training may have
a strong effect on performance. However, this was a short
test (<16 min) and the physiology of 1- to 3-h mountain
racing may be different; and importantly, the two moun-
tain races we sampled do present altitude challenges. The
observation that even highly trained mountain runners
showed tremendous variation in uphill energy cost sug-
gests that this effect may be even stronger in many of the
2017 | Vol. 5 | Iss. 8 | e13256
Page 6
ª 2017 The Authors. Physiological Reports published by Wiley Periodicals, Inc. on behalf of
The Physiological Society and the American Physiological Society.
Heart Rate during Mountain Running
A. Best et al.
athletes included in this study. A runner unaccustomed to
the muscle recruitment patterns specific to mountain rac-
ing may prematurely exhaust particular muscle groups
resulting in reduced performance relative to what he or
she can achieve in a traditional road race (for which they
are ostensibly better trained). For example, uphill running
activates the vastus muscle group and the soleus to a
greater extent than level running while other muscle
groups are activated less (Gostill et al. 1974; Sloniger
et al. 1997). It is possible that a runner unaccustomed to
the demands of uphill running may exhaust the vastus
and soleus muscles prematurely and be forced to reduce
pace and aerobic effort as the race progresses. However,
substantial second-half slowing observed in this study’s
elite and altitude-acclimatized subgroup—three athletes
trained for the specific demands of mountain running—
suggests that this effect may be minimal. A larger sample
of elite runners could clarify this point.
Heart rate change in the second half of the marathon is
more strongly
correlated (r2=0.370) with second-half
slowing than at MWRR (again, this correlation was
inconsequential for the half marathon and PP). Here, per-
haps is a good candidate application for the hypothesis of
reduced HR due to reduced aerobic output. The eccentric
muscle damage and diminishing glycogen stores incurred
during a marathon may cause a slowing that is concur-
rent with, and causative of, a reduction in aerobic output.
A moderate and significant correlation between slowing
pace and HR decrease supports this explanation. Glyco-
gen depletion should also be expected to pose a challenge
for PP runners, but as HR change was not associated with
changing pace, glycogen depletion can only minimally
explain HR and pace changes.
Other observed significant correlations are most likely
specious. Less competitive MWRR runners displayed a
greater reduction or smaller increase in HR over the sec-
ond half, but competitiveness was not correlated with sec-
ond-half slowing at this or any race. The very weak
correlation between age and second-half slowing disap-
peared when examined for individual races. Less competi-
tive marathoners had higher HR, but an r-squared value
of 0.085 and a lack of correlation between competitive-
ness and any other variable suggest this is a false positive.
Finally, the higher HR (as bpm and percentage estimated
HRmax) in the shorter-duration events confirms long-
established observations that relative intensity is inversely
related to exercise duration.
In summary, uphill mountain racing does not appear
to be characterized by the continually increasing heart
rate seen in the half marathon and marathon. Our three
hypotheses for this phenomenon could be investigated
with a laboratory based study with subjects completing
race effort runs on flat and uphill grades. This would
control for terrain, altitude, and variability in interindi-
vidual responses (each subject could complete both flat
and uphill race efforts), and would permit collection of
physiological data (VO2, RER, blood lactate, etc.) and
training history.
Conclusions
Strava’s performance and heart rate data are a useful
and novel resource for exercise science investigations
provided that research questions are carefully articulated
in consideration of the strengths and limitations of this
approach. Competitors in mountain races slowed more
than their counterparts in duration-matched road races.
Mountain racing is characterized by a maintained or
decreased heart rate in the second versus first half of the
event, while road racing at the half marathon and mara-
thon distances is characterized by an increasing heart
rate. It is unclear whether or how altitude or demands
specific to uphill running explain this difference. This
study demonstrates how Strava data can be used in an
inquiry into a physiological or performance question;
results may then be used to inform a targeted labora-
tory-based study.
Acknowledgment
The authors wish to thank the athletes who contributed
data directly or indirectly to make this study possible.
Conflict of Interest
The authors declare no conflicts of interest.
References
Balducci, P., M. Clemencon, B. Morel, G. Quiniou, D. Saboul,
and C. A. Hautier. 2016. Comparison of level and graded
treadmill tests to evaluate endurance mountain runners. J.
Sports Sci. Med. 15:239–246.
Dawson, E. A., R. Shave, K. George, G. Whyte, D. Ball, D.
Gaze, et al. 2005. Cardiac drift during prolonged exercise
with echocardiographic evidence of reduced diastolic
function of the heart. Eur. J. Appl. Physiol. 94:305–309.
Ekelund, L. G. 1967. Circulatory and respiratory adaptation
during prolonged exercise. Acta Physiol. Scand. Suppl.
292:1.
Fulco, C. S., P. Rock, and A. Cymerman. 2000. Improving
athletic performance: is altitude residence or
altitude training helpful? Aviat. Space Environ. Med.
71:162–171.
Giovanelli, N., A. L. R. Ortiz, K. Henninger, and R. Kram.
2016. Energetics of vertical kilometer foot races; is steeper
cheaper? J. Appl. Physiol. 120:370–375.
ª 2017 The Authors. Physiological Reports published by Wiley Periodicals, Inc. on behalf of
The Physiological Society and the American Physiological Society.
2017 | Vol. 5 | Iss. 8 | e13256
Page 7
A. Best et al.
Heart Rate during Mountain Running
Gore, C. J., A. G. Hahn, G. C. Scroop, D. B. Watson, K. I.
Norton, R. J. Wood, et al. 1996. Increased arterial
desaturation in trained cyclists during maximal exercise at
580 m altitude. J. Appl. Physiol. 80:2204–2210.
Gore, C. J., S. C. Little, A. G. Hahn, G. C. Scroop, K. I.
Norton, P. C. Bourdon, et al. 1997. Reduced performance of
male and female athletes at 580 m altitude. Eur. J. Appl.
Physiol. Occup. Physiol. 75:136–143.
Gostill, D. L., E. Jansson, P. D. Gollnick, and B. Saltin. 1974.
Glycogen utilization in leg muscles of men during level and
uphill running. Acta Physiol. Scand. 91:475–481.
Hamilton, M. T., J. Gonzalez-Alonso, S. J. Montain, and E. F.
Coyle. 1991. Fluid replacement and glucose infusion during
exercise prevent cardiovascular drift. J. Appl. Physiol.
71:871–877.
Koistinen, P., T. Takala, V. Martikkala, and J. Lepp€aluoto.
1995. Aerobic fitness influences the response of maximal
oxygen uptake and lactate threshold in acute hypobaric
hypoxia. Int. J. Sports Med. 16:78–81.
Lawler, J., S. K. Powers, and D. Thompson. 1988. Linear
relationship between VO2-max and VO2-max decrement during
exposure to acute hypoxia. J. Appl. Physiol. 64:1486–1492.
Mahe, J. T., L. G. Jones, and L. H. Hartley. 1974. Effects of
high-altitude exposure on submaximal endurance capacity
of men. J. Appl. Physiol. 37:895–898.
McMahon, T. A., G. Valian, and E. C. Frederick. 1987.
Groucho running. J. Appl. Physiol. 62:2326–2337.
Minetti, A. E., L. P. Ardigo, and F. Saibene. 1994. Mechanical
determinants of the minimum energy cost of gradient
running in humans. J. Exp. Biol. 195:211–225.
Mognoni, P., M. D. Sirtori, F. Lorenzelli, and P. Cerretelli.
1990. Physiological responses during prolonged exercise at
the power output corresponding to the blood
lactate threshold. Eur. J. Appl. Physiol. Occup. Physiol.
60:239–243.
Richardson, R. S., E. A. Noyszewski, J. S. Leigh, and P. D.
Wagner. 1998. Lactate efflux from exercising human
skeletal muscle: role of intracellular. J. Appl. Physiol.
85:627–634.
Rubenson, J., D. Heliams, D. G. Lloyd, and P. A. Fournier.
2004. Gait selection in the ostrich: mechanical and
metabolic characteristics of walking and running with and
without an aerial phase. Proc. R. Soc. Lond. B Biol. Sci.
271:1091–1099.
Sloniger, M. A., K. J. Cureto, B. M. Prior, and E. M. Evans.
1997. Lower extremity muscle activation during horizontal
and uphill running. J. Appl. Physiol. 83:2073–2079.
Stenberg, J., B. Ekblom, and R. Messin. 1966. Hemodynamic
response to work at simulated altitude, 4000 m. J. Appl.
Physiol. 21:1589–1594.
Tanaka, H., K. D. Monahan, and D. R. Seals. 2001. Age-
predicted maximal heart rate revisited. J. Am. Coll. Cardiol.
37:153–156.
Vogel, J. A., J. E. Hansen, and C. W. Harris.
1967. Cardiovascular responses in man during exhaustive
work at sea level and high altitude. J. Appl. Physiol. 23:531–
539.
Wehrlin, J. P., and J. Hallen. 2006. Linear decrease in VO2-
max and performance with increasing altitude in endurance
athletes. Eur. J. Appl. Physiol. 96:404–412.
2017 | Vol. 5 | Iss. 8 | e13256
Page 8
ª 2017 The Authors. Physiological Reports published by Wiley Periodicals, Inc. on behalf of
The Physiological Society and the American Physiological Society.
Heart Rate during Mountain Running
A. Best et al.
| Using a novel data resource to explore heart rate during mountain and road running. | [] | Best, Andrew,Braun, Barry | eng |
PMC10329575 | Vol.:(0123456789)
1 3
Journal of Muscle Research and Cell Motility (2023) 44:115–122
https://doi.org/10.1007/s10974-022-09633-1
REVIEW
Participation and performance characteristics in half‑marathon run:
a brief narrative review
Pantelis Theodoros Nikolaidis1 · Beat Knechtle2
Received: 13 April 2022 / Accepted: 17 October 2022 / Published online: 3 November 2022
© The Author(s) 2022
Abstract
Half-marathon (HM) is a running sport of increasing popularity in both sexes and in all age groups worldwide during the
last years. Many studies have examined several aspects of HM, such as performance and participation trends, sex and age
differences, physiological correlates, and training; however, no comprehensive review has ever been contacted to summa-
rize the recently accumulated knowledge. Therefore, the aim of the present study was to review all previous research in this
sport, focusing on participation and performance aspects. It was shown that HM runners had similar anthropometric and
physiological characteristics as full-marathon runners which should be attributed to the affinity of these two races in terms
of metabolic demands. Performance in HM was related with superior scores in aerobic capacity (maximal oxygen uptake,
anaerobic threshold and running economy) and training characteristics (sport experience, weekly distance, training speed,
frequency of sessions and long single endurance run distance), and lower scores in adiposity-related scores (e.g. body mass,
body mass index, body fat percentage and skinfold thickness). Considering the popularity of HM race and the lack of many
original studies (compared to FM race), this is an exciting field for scientific research with a large potential for practical
applications, since the majority of HM runners are amateur runners in need of sex-, age- and performance-tailored exercise
prescription.
Keywords Aerobic capacity · Anaerobic threshold · Endurance · Exercise · Nutrition · Participation · Running economy
Introduction
Half-marathon (HM) is a running event of increasing popu-
larity in both sexes and in all age groups worldwide during
the last years (Bonet et al. 2022). Although full-marathon
(FM) is the most popular endurance running distance, most
runners participate in HM (Cribari et al. 2013). Many studies
have examined several aspects of HM, such as performance
and participation trends, sex and age differences, physio-
logical correlates, and training,however, no comprehensive
review has ever been contacted to summarize the accumu-
lated knowledge. Therefore, the aim of the present study is
to review all previous research covering all aspects such as
epidemiological trends, the role of age and sex, especially
focusing on the physiological aspects.
Participation
Main aspects of participation included the numbers of par-
ticipants in HM and whether these numbers would change
across calendar years, sex differences in participation (typi-
cally examined using the men-to-women ratio) and age
across years, whereas these aspects may vary by national-
ity. In 2016, more finishers and events were observed in HM
than in FM (Fig. 1). Participation trends in HM have been
examined with regards to FM in a single country, Switzer-
land, where those participating in HM are three times more
than those competing in FM races (Anthony et al. 2014).
It has been observed in 226,754 HM and 86,419 FMrun-
ners competing in Switzerland between 2000 and 2010 that
the number of HM increased from 2000 to 2010 for both
men (+ 231%) and women (+ 299%), whereas the number
of male and female FM runners increased until 2005 only
and decreased thereafter (Anthony et al. 2014). A study on
* Pantelis Theodoros Nikolaidis
[email protected]
1
School of Health and Caring Sciences, University of West
Attica, Ag. Spyridonos, 122 43 Egaleo, Athens, Greece
2
Institute of Primary Care, University of Zurich, Zurich,
Switzerland
116
Journal of Muscle Research and Cell Motility (2023) 44:115–122
1 3
508,108 runners (125,894 female and 328,430 male HM and
10,205 female and 43,489 male FM) competing between
1999 and 2014 in all flat HM and FM held in Switzerland
showed that the number of women and men increased across
years in both HM and FM, and there were 12.3 times more
female HM than female FM and 7.5 times more male HM
than male FM (Knechtle et al. 2016a). These different pro-
portions of women and men competing in HM and FM races
indicated that HM was a sport where relatively more women
participated compared to FM. The abovementioned prelimi-
nary studies (Anthony et al. 2014; Knechtle et al. 2016a)
highlighted a larger participation in HM than in FM races,
and this trend was more striking in women than in men. An
explanation of this discrepancy between sexes might be that
women could be considered as more ‘novice’ runners than
men, and consequently, should run more HM before enter-
ing FM races.
The participation and performance in HM may vary by
nationality, and it would be interesting to focus on trends
concerning East African runners who were considered as
experts in long distance running events (Knechtle et al.
2016b).Actually, a study of ~ half million HM and FM run-
ners originating from 126 countries and competing between
1999 and 2014 in all road-based HM and FM held in Swit-
zerland reported that, in HM, 48 women (0.038%) and 63
men (0.019%) were from Ethiopia and 80 women (0.063%)
and 134 men (0.040%) from Kenya, whereas in FM, three
women (0.029%) and 15 men (0.034%) were from Ethiopia
and two women (0.019%) and 33 men (0.075%) from Kenya
(Knechtle et al. 2016b). These findings suggested that the
largest percentage of participants in HM is of local origin.
In both women and men, the best performance in HM and
FM held in Switzerland was achieved by East African run-
ners with Ethiopian and Kenyan runners being the youngest
in both sexes and formats of race (Knechtle et al. 2016b).
These findings showed that East-African runners were the
fastest in both HM and FM although they represented the
smallest percentage of participants (Knechtle et al. 2016b).
This observation was in agreement with an analysis of the
world’s best HM runners during 1999–2015, where it was
shown that most of them were Kenyans (30% in women
and 57% in men) (Nikolaidis et al. 2017). According to
the abovementioned effect of nationality, the characteris-
tics of HM differed from country to country, e.g. the local
people tended to participate more to races taking place in
their country than foreigners. Furthermore, the participa-
tion and performance may change across years. Actually, a
study examined the changes in participation, performance
and age of East African runners competing in HM and FM
held in Switzerland between 2000 and 2010 indicated that
across time, the number of Kenyan and Ethiopian finishers
remained stable while the number of Non-African finishers
increased for both women and men in both HM and FM
(Cribari et al. 2013). This difference in participation trends
across years by nationality might be due to the increase of
local ‘recreational’ runners, while the number of the most
competitive runners coming from abroad would remain sta-
ble. To sum up, more runners compete in HM than in FM,
and the fastest HM runners are East Africans.
Age of peak performance
Every sport has its own age of peak performance and thus,
it would be important to estimate at which age HM runners
achieve their peak performance in order to set long-term
training goals. The largest part of the finishers in HM and
FM held in Switzerland of both genders was assigned to
age group 40–44 years in HM (19.5% of the total number
of finishers) and FM (22.0% of finishers) (Anthony et al.
2014). For both HM and FM races, most of the female and
male finishers were recorded in age group 40–44 years
(Knechtle et al. 2016a; Knechtle and Nikolaidis 2018).
In HM, women (41.4 years) were at the same age as men
(41.3 years),in FM, women (42.2 years) were at the same
age than men (42.1 years),however, women and men FM
runners were older than their counterpart HM runners (Kne-
chtle et al. 2016a). With regards to the age of peak perfor-
mance, it may differ depending on the performance level,
i.e. whether all or the top finishers were considered. For
instance, in the world’s largest HM race—the Göteborgs-
Varvet—U40 was the fastest age group when all finishers
were analyzed, whereas U35 were the fastest when the top
10 were considered (Knechtle and Nikolaidis 2018). Moreo-
ver, an analysis of the world’s best HM runners indicated
Fig. 1 Finishers and events in
USA in 2016. Source: http://
www. runni ngusa. org (accessed
on 16/9/2017)
117
Journal of Muscle Research and Cell Motility (2023) 44:115–122
1 3
an age of 26–27 years, which was younger than in FM
and 100 km ultra-marathon races (Nikolaidis et al. 2017)
(Fig. 2). This observation was confirmed by a study using
non-linear regression on world records in HM, where the age
peak performance was ~ 27 years (Nikolaidis et al. 2018).
The effect of age on HM performance differed from other
endurance sports (Sterken 2005). It has been shown that
age-related losses in endurance performance did not occur
before the age of 50 years with mean FM and HM race times
being identical for the age groups 20–49 years, whereas age-
related performance decreases of the 50–69-year-old sub-
jects were only in the range of 2.6–4.4% per decade (Leyk
et al. 2007). These results suggested that the majority of
older athletes were able to maintain a high degree of physi-
cal plasticity supporting the hypothesis that lifestyle factors
had considerably stronger influences on functional capacity
than the factor age (Leyk et al. 2007). No significant age-
related decline in performance appeared before the age of
55 years, whereas only a moderate decline is seen thereafter
(Leyk et al. 2010). Performance losses in middle age were
mainly due to a sedentary lifestyle, rather than biological
aging (Leyk et al. 2010). In summary, the average age of HM
runners was 40–44 years, and the age of peak performance
was younger than 35–40 years.
Performance trends
Performance in HM might be examined using either race
time or average speed. In HM held in Switzerland, women
(10.29 ± 3.03 km/h) were faster than men (10.22 ± 3.06 km/h)
as well as in FM, women (14.77 ± 4.13 km/h) were faster
than men (14.48 ± 4.07 km/h) (Knechtle et al. 2016a).
Slower HM race time by 13% (10% in FM) in women than
in men was observed in a study of all and top 10 finishers
aged 20–79 years (Leyk et al. 2007). Moreover, a sex dif-
ference of ~ 14% was noted in the world best HM runners
(Nikolaidis et al. 2017). Sex differences in performance may
also be attributed to the different adaptation to long-term
exercise between women and men. A study on 16 males and
16 females preparing for a HM revealed a larger increase in
the average daily metabolic rate in men than in women sug-
gesting exercise stimulates more habitual physical activity
and diet-induced thermogenesis in men than in women (Mei-
jer et al. 1991). The variation of performance from a race to
race seems to depend on competitive experience and attitude
toward competing and was found 4.2% for the fastest quartile
of men runners in HM with men, slower and younger run-
ners presenting more variation (Hopkins and Hewson 2001).
Physiological, anthropometric and training correlates of
performance were examined in following sections,however,
they might differ depending on performance level, consider-
ing that in the elite level other factors (e.g. shoe technology)
would play an important role (Goss et al. 2022).
Physiological correlates of performance
Physical fitness is classified as health- (consisting of body
composition, aerobic capacity, muscular strength, muscular
endurance and flexibility, i.e. components related directly
to health) or sport-related (consisting of those components
related to sport performance such as speed or reaction time).
Based on the relatively long duration of a HM, it is reason-
able to assume that performance in this sport relates to maxi-
mal oxygen uptake (VO2max), since it relies mostly on the
aerobic energy transfer system. One of the oldest studies on
HM (Williams and Nute 1983) already identified VO2max
and anaerobic threshold as correlates of race time (r = −0.81
and r = −0.88, respectively). In addition, in male recreational
runners, HM race time correlated with VO2max (r = −0.64),
speed at VO2max (r = −0.84) and anaerobic threshold
(r = −0.79) (Alvero-Cruz et al. 2019). These relationships
were confirmed in HM runners with asthma, too,actually,
the HM pace of asthmatic runners correlated largely with
VO2max (r = 0.86) and almost perfectly with running speed
Fig. 2 Race time and age of the best runners by race distance and sex. Based on Nikolaidis and colleagues Nikolaidis et al. 2017. Error bars rep-
resent standard deviations. HM = half-marathon, FM = full-marathon
118
Journal of Muscle Research and Cell Motility (2023) 44:115–122
1 3
at a blood lactate concentration of 2 mmol.L−1 (r = 0.97)
(Freeman et al. 1990). In amateur runners (nine men, age
36 years), HM time was almost perfectly correlated with
VO2max (r = 0.91) and speed corresponding to VO2max
(r = 0.90) (Santos et al. 2012). Furthermore, a comparison
among 400–800 m, 1500–3000 m and HM women runners
showed that HM runners had the highest VO2max (Nurme-
kivi et al. 1998). In runners, the pace of HM was comparable
to the maximal lactate steady state velocity (Legaz-Arrese
et al. 2011).
Considering the physiological relevance of HM and FM
races, previous research examined the relationship of per-
formances in these two races (Salinero et al. 2017; Karp
2007; Coyle 2007). A research on 84 male amateur FM run-
ners (aged 41.0 years, finish time 226.0 min) showed a very
large correlation between FM and HM race time (r = 0.81)
(Salinero et al. 2017). In 2004 U.S. Olympic Marathon Tri-
als qualifiers (104 men, 151 women), FM performance cor-
related to HM performance (Karp 2007). Maintaining the
world record pace for the HM in the FM would lead to run
a FM in 1:58 h:min (Coyle 2007).
In addition to the abovementioned correlation studies, an
approach to study determinants of performance in HM is to
develop prediction equations of race time based on corre-
lates (Pérez et al. 2012; Gómez-Molina et al. 2017), which
include usually two steps, first, the development of an equa-
tion in a sample of runners, and second, the validation of this
equation in another sample. For instance, a study on male
runners considered training-related and anthropometric vari-
ables, and laboratory data from a graded exercise test (GXT)
on a treadmill (VO2max, speed at the anaerobic threshold,
peak speed) and biomechanical variables (contact and flight
times, step length and step rate) (Gómez-Molina et al. 2017).
This study found that HM race time could be predicted to
90.3% by variables related to training and anthropometry,
94.9% by physiological variables, 93.7% by biomechanical
parameters and 96.2% by a general equation, and using these
equations, the predicted time was significantly correlated
with performance (r = 0.78, 0.92, 0.90 and 0.95, respec-
tively) (Gómez-Molina et al. 2017). Moreover, HM race
speed could be predicted as V21k (km/h) = (V2*1.085) + (−
0.282*bLA2) −0.131, r2 = 0.97, ETE = 0.414 km/h, where
V2 was the speed during a test in track at constant pace
over 2400 m slightly higher than the competition expected
pace and bLA2 blood lactate (Pérez et al. 2012). A research
examined the ratio between running speed and heart rate
(HR) as predictor for aerobic capacity (based on the assump-
tion that lower HR at a given speed is expected for more fit
individuals), and subsequently race time in 10 km, HM and
FM (Altini et al. 2017). This study showed that the speed
to HR ratio provides higher accuracy in aerobic capacity
estimation compared to resting HR or no-physiological data,
and large correlations between aerobic capacity and race
time (r = 0.56–0.61) (Altini et al. 2017). In addition to the
comparison with laboratory exercise testing, HM perfor-
mance has been investigated with field tests such as Cooper
test (Alvero-Cruz et al. 2019; Alvero-Cruz et al. 2020). For
instance, Alvero-Cruz and colleagues (Alvero-Cruz et al.
2019; Alvero-Cruz et al. 2020) observed an almost perfect
correlation between HM race time and Cooper test distance,
and high predictive validity of this test.
Another methodological approach that may provide indi-
rect information for the identification of correlates of per-
formance is to examine the acute physiological responses
to a HM by comparing values pre- and post-race (Dressen-
dorfer 1991). For instance, runners were tested in a graded
exercise test before and after a HM,time to exhaustion (6.0
vs 4.1 min), VO2max (60.0 vs 56.3 ml.kg−1.min−1), peak
respiratory exchange ratio (RER; 1.18 vs 1.06), and peak
La (9.7 vs 7.8 mmol.L−1) decreased after the HM (Dres-
sendorfer 1991). Other studies focus on muscular acute
responses to a HM (Boccia et al. 2017 Boccia et al. 2017).
For instance, intermittent isometric rapid contractions of the
knee extensor muscles were performed the day before and
immediately after a HM, where it was observed that HM had
a greater negative effect on repeated, rather than on single,
attempts of maximal force production (Boccia et al. 2017).
In another study of this research group that examined pre-
and post-HM race maximum voluntary contractions of knee
extensor muscles, it was found that post-race knee extensors
showed a decreased strength (−13.9%) and a reduction in
EMG amplitude (−13.10%) and in CV (−6.8%, p = 0.032)
(Boccia et al. 2017). Moreover, compared to FM, a HM race
induced smaller vertical jump height reduction and less self-
reported muscle pain suggesting less muscular fatigue in
HM than in FM race (Coso et al. 2017). The abovemen-
tioned studies (Boccia et al. 2017; Boccia et al. 2017; Coso
et al. 2017) highlighted the importance of muscular fitness
in addition to aerobic capacity for HM race. In line with the
findings of the correlation studies presented in this section,
a comparative study among different performance groups
highlighted the role of physiological characteristics (Ogueta-
Alday et al. 2018). Particularly, Oguea-Alday and colleagues
(Ogueta-Alday et al. 2018) found that HM runners with race
time 70 min had superior VO2max and running economy
than those with race time 80 min, 90 min and 105 min. In
summary, performance on HM race would depend mostly on
VO2max and other indices of aerobic capacity.
Anthropometric and training correlates
of performance
Performance in HM is not related only to physiological char-
acteristics, but also to anthropometry and training habits
(Campbell 1985; Friedrich et al. 2014; Knechtle et al. 2012;
Zillmann et al. 2013). For instance, a study in university HM
119
Journal of Muscle Research and Cell Motility (2023) 44:115–122
1 3
runners observed that distance run per week and number of
weeks training for the event, and body mass index (BMI) were
predictors of HM race time (Campbell 1985). In female and
male recreational HM runners, HM race time was related
to body fat percentage (BF), running speed during training,
and BMI was predictor of performance only in men (Frie-
drich et al. 2014), whereas elsewhere race time was more
strongly associated with anthropometry in women than men
(Knechtle et al. 2010). For male HM runners, BMI, BF and
speed in running during training were related to race time
(Zillmann et al. 2013). Furthermore, in female finishers of
the ‘Half Marathon Basel’, race time was largely related to
body mass, BMI, BF (r = 0.48–0.60), and could be predicted
by the formula ‘166.7 + 1.7*midaxilla skinfold – 6.4*speed
in training’(R2 = 0.71) (Knechtle et al. 2011). In HM, FM and
UM master runners (> 35 years old), BF and training charac-
teristics, not skeletal muscle mass, were associated with run-
ning times (Knechtle et al. 2012). A comparison between HM
and FM men runners showed that HM runners were heavier,
had longer legs, thicker upper arms, a thicker thigh, a higher
sum of skinfold thicknesses, a higher body fat percentage and
a higher skeletal muscle mass, fewer years of experience, com-
pleted fewer weekly training kilometers, and fewer weekly run-
ning hours (Zillmann et al. 2013). The relationship of HM with
training characteristics might be explained by exercise-induced
chronic cardiometabolic adaptations resulting in improvements
in VO2max, a major predictor of performance (Meyler et al.
2021).
A comparative study between HM and FM races showed
that a fast race time was associated with high weekly train-
ing volume (> 32 km) and a long training single distance
(> 21 km) in HM, and high weekly training volume (> 65 km)
in FM runners (Fokkema et al. 2020). Furthermore, a research
among groups different performance groups confirmed the role
of anthropometric and training characteristics on HM perfor-
mance (Ogueta-Alday et al. 2018), where runners with race
time 70 min had more sport experience and weekly running
distance, and less body mass, BMI and skinfold thickness than
those with race time 80 min, 90 min and 105 min. In summary,
performance on HM race would depend mostly on VO2max
and other indices of aerobic capacity. To sum up, HM per-
formance depended on weekly running distance, number of
training units, training running speed, BMI and BF (Campbell
1985; Friedrich et al. 2014; Knechtle et al. 2012; Zillmann
et al. 2013). Therefore, a practical advice for recreational run-
ners wishing to improve their race time might be to increase
training (distance, running speed and frequency) and decrease
BMI and BF.
Pacing
Pacing in sport refers to the conscious or subconscious
regulation of performance (e.g. speed during a race) in
order to achieve a goal (Thiel et al. 2018; Micklewright
2019). Although pacing is related to performance in endur-
ance sports (the less variable the pacing, the faster the
race time), little information exist with regards to pac-
ing strategies of HM runners (Hanley 2015). The relevant
literature has been developed recently (Nikolaidis et al.
2019 Cuk et al. 2019; Cuk et al. 2020; Nikolaidis et al.
2019). A methodological approach to analyze pacing in
HM has been to consider speed for intermediate segments,
e.g. every 5 km (0–5, 5–10, 10–15 and 15–20 km) and
end 1.1 km segments (Hanley 2015). An analysis of all
finishers in Ljubljana 2017 HM showed a positive pacing
with every segment being slower than its precedent one
(Nikolaidis et al. 2019). In elite (finishers in the IAAF
World Half Marathon Championships) HM runners, it was
observed that the fastest finishers maintained split speed
from 5 to 15 km, whereas slower runners decreased speed
after 5 km (Hanley 2015). Compared to FM, an analysis of
recreational HM runners observed a more even pacing in
Ljubljana (Nikolaidis et al. 2019) and Vienna races (Cuk
et al. 2020).A research of the pacing in the Great West
Run HM showed that RPE scales with the proportion of
exercise time that remains inferring that the brain uses
a scalar timing mechanism (Faulkner et al. 2008). With
regards to the variation of pacing by age, a research in
Vienna City HM race revealed a positive pacing strategy,
i.e. speed decreased across race, with an end spurt in all
age groups and larger variation of speed in the younger age
groups (Cuk et al. 2019). Additionally, a more even pacing
was observed to relate with high training volume and long
single training run (Fokkema et al. 2020). In summary,
similarly to other endurance races, pacing was associated
with performance in HM race with faster runners adopting
a more even pacing than their slower counterparts. Thus,
recreational runners would be recommended to maintain
running speed during a HM race.
Practical applications
With regards to the popularity of HM race, the findings of
research on this field would concern an increasing number
of HM runners. The fact that HM race has been a sport
activity recently developed would imply that it could
not be ‘covered’ by scientific and professional knowl-
edge derived from long distance running – such as 5 km
and 10 km – and FM race. This observation highlighted
the need to develop further the research on this field.
120
Journal of Muscle Research and Cell Motility (2023) 44:115–122
1 3
In this context, the present review attempted to address
fundamental questions on the identity of HM runners
(sex, age and nationality), performance trends and cor-
relates, to develop practical information for professionals
working with runners. For instance, the knowledge of the
age of peak performance can aid setting long-term per-
formance goals considering the age of runners. Several
original research studies were identified that examined
correlates of HM performance and enhanced our knowl-
edge on this field. Based on these findings, practitioners
working with HM runners should aim to optimize aerobic
capacity (e.g., VO2max and anaerobic threshold), train-
ing indices (e.g., weekly running distance and running
speed) and adiposity-related parameters. Although these
parameters clearly were related to HM performance in
recreational runners that might be considered as an het-
erogeneous group, a challenge for future studies would be
to examine the variation of their importance depending
on performance level.
Conclusion
In conclusion, performance in HM was related with supe-
rior scores in aerobic capacity (maximal oxygen uptake,
anaerobic threshold and running economy) and training
characteristics (sport experience, weekly distance, training
speed, frequency of sessions and long single endurance
run distance), and lower scores in adiposity-related scores
(e.g. body mass, body mass index, body fat percentage and
skinfold thickness) (Fig. 3). Considering the popularity of
HM race and the lack of many original studies (compared
to FM race), this is an exciting field for scientific research
with a large potential for practical applications.
Acknowledgements Figure 3 is credited to Dr. Celine Dewas.
Author contributions All authors wrote the main manuscript text and
prepared figures. All authors reviewed the manuscript.
Funding Open access funding provided by HEAL-Link Greece.
Fig. 3 Physiological, anthropo-
metric and training characteris-
tics influencing half-marathon
performance
121
Journal of Muscle Research and Cell Motility (2023) 44:115–122
1 3
Data availability All data are available by the corresponding author
upon reasonable request.
Declarations
Conflict of interest The authors declare no conflict of interest.
Open Access This article is licensed under a Creative Commons Attri-
bution 4.0 International License, which permits use, sharing, adapta-
tion, distribution and reproduction in any medium or format, as long
as you give appropriate credit to the original author(s) and the source,
provide a link to the Creative Commons licence, and indicate if changes
were made. The images or other third party material in this article are
included in the article's Creative Commons licence, unless indicated
otherwise in a credit line to the material. If material is not included in
the article's Creative Commons licence and your intended use is not
permitted by statutory regulation or exceeds the permitted use, you will
need to obtain permission directly from the copyright holder. To view a
copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/.
References
Altini M, Van Hoof C, Amft O. 2017 Relation between estimated car-
diorespiratory fitness and running performance in free-living: an
analysis of HRV4Training data C3 - 2017 IEEE EMBS Interna-
tional Conference on Biomedical and Health Informatics, BHI.
2017:249–252.
Alvero-Cruz JR, Carnero EA, Giráldez García MA et al (2019) Cooper
test provides better half-marathon performance prediction in rec-
reational runners than laboratory tests. Front Physiol 10:1349
Alvero-Cruz JR, Standley RA, Giráldez-García MA, Carnero EA
(2020) A simple equation to estimate half-marathon race time
from the cooper test. Int J Sports Physiol Perform 15(5):690–695
Anthony D, Rüst CA, Cribari M, Rosemann T, Lepers R, Knechtle B
(2014) Differences in participation and performance trends in age
group half and full marathoners. Chin J Physiol 57(4):209–219
Boccia G, Dardanello D, Tarperi C et al (2017) Decrease of muscle
fiber conduction velocity correlates with strength loss after an
endurance run. Physiol Meas 38(2):233–240
Boccia G, Dardanello D, Tarperi C et al (2017) Fatigue-induced dis-
sociation between rate of force development and maximal force
across repeated rapid contractions. Hum Mov Sci 54:267–275
Bonet JB, Javierre C, Guimarães JT et al (2022) Benefits on hemato-
logical and biochemical parameters of a high-intensity interval
training program for a half-marathon in recreational middle-aged
women runners. Int J Environ Res Public Health 19(1):498
Campbell MJ (1985) Predicting running speed from a simple question-
naire. Br J Sports Med 19(3):142–144
Coyle EF (2007) Physiological regulation of marathon performance.
Sports Med 37(4–5):306–311
Cribari M, Rüst CA, Rosemann T, Onywera V, Lepers R, Knechtle B
(2013) Participation and performance trends of East-African run-
ners in Swiss half-marathons and marathons held between 2000
and 2010. BMC Sports Sci Med Rehabil 5(1):24
Cuk I, Nikolaidis PT, Markovic S, Knechtle B (2019) Age differences
in pacing in endurance running: comparison between marathon
and half-marathonmen and women. Med (Kaunas, Lithuania)
55(8):479
Cuk I, Nikolaidis PT, Knechtle B (2020) Sex differences in pac-
ing during half-marathon and marathon race. Res Sports Med
28(1):111–120
Del Coso J, Salinero JJ, Lara B, Abián-Vicén J, Gallo-Salazar C, Are-
ces F (2017) A comparison of the physiological demands imposed
by competing in a half-marathon vs. a marathon. J Sports Med
Phys Fit 57(11):1399–1406
Dressendorfer RH (1991) Acute reduction in maximal oxygen uptake
after long-distance running. Int J Sports Med 12(1):30–33
Faulkner J, Parfitt G, Eston R (2008) The rating of perceived exertion
during competitive running scales with time. Psychophysiology
45(6):977–985
Fokkema T, van Damme A, Fornerod MWJ, de Vos RJ, Bierma-
Zeinstra SMA, van Middelkoop M (2020) Training for a (half-)
marathon: training volume and longest endurance run related
to performance and running injuries. Scand J Med Sci Sports
30(9):1692–1704
Freeman W, Williams C, Nute MG (1990) Endurance running perfor-
mance in athletes with asthma. J Sports Sci Summ 8(2):103–117
Friedrich M, Rüst CA, Rosemann T et al (2014) A comparison of
anthropometric and training characteristics between female and
male half-marathoners and the relationship to race time. Asian
J Sports Med 5(1):10–20
Gómez-Molina J, Ogueta-Alday A, Camara J, Stickley C, Rodríguez-
Marroyo JA, García-López J (2017) Predictive variables of
half-marathon performance for male runners. J Sports Sci Med
16(2):187–194
Goss CS, Greenshields JT, Noble TJ, Chapman RF (2022) A narra-
tive analysis of the progression in the top 100 marathon, half-
marathon, and 10-km road race times from 2001 to 2019. Med
Sci Sports Exerc 54(2):345–352
Hanley B (2015) Pacing profiles and pack running at the IAAF world
half marathon championships. J Sports Sci 33(11):1189–1195
Hopkins WG, Hewson DJ (2001) Variability of competitive
performance of distance runners. Med Sci Sports Exerc
33(9):1588–1592
Karp JR (2007) Training characteristics of qualifiers for the
U.S. olympic marathon trials. Int J Sports Physiol Perform
2(1):72–92
Knechtle B, Nikolaidis PT (2018) Sex- and age-related differences in
half-marathon performance and competitiveness in the world’s
largest half-marathon - the GöteborgsVarvet. Res Sports Med
(print) 26(1):75–85
Knechtle B, Knechtle P, Rosemann T, Senn O (2010) Sex differences in
association of race performance, skin-fold thicknesses, and train-
ing variables for recreational half-marathon runners. Percept Mot
Skills 111(3):653–668
Knechtle B, Knechtle P, Barandun U, Rosemann T, Lepers R (2011)
Predictor variables for half marathon race time in recreational
female runners. Clinics (sao Paulo, Brazil) 66(2):287–291
Knechtle B, Rüst CA, Knechtle P, Rosemann T (2012) Does muscle
mass affect running times in male long-distance master runners?
Asian J Sports Med 3(4):247–256
Knechtle B, Nikolaidis PT, Zingg MA, Rosemann T, Rüst CA (2016a)
Half-marathoners are younger and slower than marathoners.
Springerplus 5:76
Knechtle B, Nikolaidis PT, Onywera VO, Zingg MA, Rosemann T,
Rüst CA (2016b) Male and female Ethiopian and Kenyan runners
are the fastest and the youngest in both half and full marathon.
Springerplus 5:223
Legaz-Arrese A, Carranza-García LE, Serrano-Ostáriz E, González-
Ravé JM, Terrados N (2011) The traditional maximal lactate
steady state test versus the 5 × 2000 m test. Int J Sports Med
32(11):845–850
Leyk D, Erley O, Ridder D et al (2007) Age-related changes in
marathon and half-marathon performances. Int J Sports Med
28(6):513–517
Leyk D, Rüther T, Wunderlich M et al (2010) Physical performance in
middle age and old age: good news for our sedentary and aging
society. Deutsch Arztebl Int 107(46):809–816
122
Journal of Muscle Research and Cell Motility (2023) 44:115–122
1 3
Meijer GA, Westerterp KR, Seyts GH, Janssen GM, Saris WH, ten
Hoor F (1991) Body composition and sleeping metabolic rate in
response to a 5-month endurance-training programme in adults.
Eur J Appl Physiol 62(1):18–21
Meyler S, Bottoms L, Muniz-Pumares DJEP (2021) Biological and
methodological factors affecting response variability to endurance
training and the influence of exercise intensity prescription. Exp
Physiol 106(7):1410–1424
Micklewright D (2019) Interventions decision-making, pacing, and
performance in endurance sport. Routledge, London, pp 47–69
Nikolaidis PT, Onywera VO, Knechtle B (2017) Running performance,
nationality, sex, and age in the 10-km, half-marathon, marathon,
and the 100-km ultramarathon IAAF 1999–2015. J Strength Cond
Res 31(8):2189–2207
Nikolaidis PT, Di Gangi S, Knechtle B (2018) World records in
half-marathon running by sex and age. J Aging Phys Act
26(4):629–636
Nikolaidis PT, Cuk I, Rosemann T, Knechtle B (2019) Performance
and Pacing of Age Groups in Half-Marathon and Marathon. Int J
Environ Res Public Health 16(10):1777
Nikolaidis PT, Ćuk I, Knechtle B. 2019 Pacing of women and men in
half-marathon and marathon races. Medicina (Lithuania). 55(1).
Nurmekivi A, Lemberg H, Maaroos J, Lusti J, Jürimäe T (1998) Run-
ning performance and aerobic working capacity in female runners.
Med Sport 51(2):221–225
Ogueta-Alday A, Morante JC, Gómez-Molina J, García-López J (2018)
Similarities and differences among half-marathon runners accord-
ing to their performance level. PLoS ONE 13(1):e0191688
Pérez IM, Pérez DM, González CC, Esteve-Lanao J (2012) Prediction
of race pace in long distance running from blood lactate concen-
tration around race pace. J Hum Sport Exerc 7(4):763–769
Salinero JJ, Soriano ML, Lara B et al (2017) Predicting race time
in male amateur marathon runners. J Sports Med Phys Fitness
57(9):1169–1177
Santos TM, Rodrigues AI, Greco CC, Marques AL, Terra BS, Oliveira
BRR (2012) Estimated VO2max and its corresponding velocity
predict performance of amateur runners. Rev Bras De Cineantro-
pometria e Desempenho Hum 14(2):192–201
Sterken E (2005) A stochastic frontier approach to running perfor-
mance. IMA J Manag Math 16(2):141–149
Thiel C, Pfeifer K, Sudeck G (2018) Pacing and perceived exertion in
endurance performance in exercise therapy and health sports. Ger
J Exerc Sport Res 48(1):136–144
Williams C, Nute ML (1983) Some physiological demands of a
half-marathon race on recreational runners. Br J Sports Med
17(3):152–161
Zillmann T, Knechtle B, Rüst CA, Knechtle P, Rosemann T, Lepers R
(2013) Comparison of training and anthropometric characteris-
tics between recreational male half-marathoners and marathoners.
Chin J Physiol 56(3):138–146
Publisher's Note Springer Nature remains neutral with regard to
jurisdictional claims in published maps and institutional affiliations.
| Participation and performance characteristics in half-marathon run: a brief narrative review. | 11-03-2022 | Nikolaidis, Pantelis Theodoros,Knechtle, Beat | eng |
PMC6040753 | RESEARCH ARTICLE
Russians are the fastest 100-km ultra-
marathoners in the world
Beat Knechtle1,2*, Pantelis Theodoros Nikolaidis3, Fabio Valeri2
1 Medbase St. Gallen Am Vadianplatz, St. Gallen, Switzerland, 2 Institute of Primary Care, University of
Zurich, Zurich, Switzerland, 3 Exercise Physiology Laboratory, Nikaia, Greece
* [email protected]
Abstract
Objectives
A recent study investigating the top 10 100-km ultra-marathoners by nationality showed that
Japanese runners were the fastest worldwide. This selection to top athletes may lead to a
selection bias and the aim of this study was to investigate from where the fastest 100-km
ultra-marathoners originate by considering all finishers in 100-km ultra-marathons since
1959.
Methods
We analysed data from 150,710 athletes who finished a 100-km ultra-marathon between
1959 and 2016. To get precise estimates and stable density plots we selected only those
nationalities with 900 and more finishes resulting in 24 nationalities. Histograms and density
plots were performed to study the distribution of race time. Crude mean, standard deviation,
median, interquartile range (IQR), mode, skewness and excess of time for each nationality
were computed. A linear regression analysis adjusted by sex, age and year was performed
to study the race time between the nationalities. Histograms, density and scatter plots
showed that some races seemed to have a time limit of 14 hours. From the complete dataset
the finishes with more than 14 hours were removed (truncated dataset) and the same
descriptive plots and analysis as for the complete dataset were performed again. In addition
to the linear regression a truncated regression was performed with the truncated dataset to
allow conclusion for the whole sample. To study a potential difference between races at
home and races abroad, an interaction term race site home/abroad with nationality was
included in the model.
Results
Most of the finishes were achieved by runners from Japan, Germany, Switzerland, France,
Italy and USA with more than 260’000 (85%) finishes. Runners from Russia and Hungary
were the fastest and runners from Hong Kong and China were the slowest finishers.
PLOS ONE | https://doi.org/10.1371/journal.pone.0199701
July 11, 2018
1 / 29
a1111111111
a1111111111
a1111111111
a1111111111
a1111111111
OPEN ACCESS
Citation: Knechtle B, Nikolaidis PT, Valeri F (2018)
Russians are the fastest 100-km ultra-marathoners
in the world. PLoS ONE 13(7): e0199701. https://
doi.org/10.1371/journal.pone.0199701
Editor: Luca Paolo Ardigò, Universita degli Studi di
Verona, ITALY
Received: November 18, 2017
Accepted: June 12, 2018
Published: July 11, 2018
Copyright: © 2018 Knechtle et al. This is an open
access article distributed under the terms of the
Creative Commons Attribution License, which
permits unrestricted use, distribution, and
reproduction in any medium, provided the original
author and source are credited.
Data Availability Statement: The data underlying
this study were collected from the ‘Deutsche
Ultramarathon Vereinigung’ (DUV) and are freely
accessible using the following link: http://statistik.
d-u-v.org/geteventlist.php?year=all&dist=
100km&country=all&Submit.x=17&Submit.y=
6&label=&surface=all&sort=1&from=&to=. The
authors used the following search criteria: ‘Year’–
‘all’, ‘Distance’–‘100 km’, and ‘Country’–‘All’. This
search leads to more than 4,500 races; all race
results were manually downloaded by the
investigators. The authors did not have any special
access privileges.
Conclusion
In contrast to existing findings investigating the top 10 by nationality, this analysis showed
that ultra-marathoners from Russia, not Japan, were the fastest 100-km ultra-marathoners
worldwide when considering all races held since 1959.
Introduction
Ultra-marathon running, such as 100-km race, is a sport of increasing popularity [1]. Particu-
larly, the number of finishers in 100-km ultra-marathon running increased exponentially,
both for women and men, from 1998 to 2011 [2]. Performance in this sport depends on physi-
ological, e.g. peak running velocity during a graded exercise test, anaerobic threshold, maximal
oxygen uptake (VO2max) and oxygen uptake (VO2) at 16 km/h [3] and psychological charac-
teristics, e.g. cognitive function [4].
Another performance-related characteristic is pacing as it has been shown that faster run-
ners exhibit smaller decrease in their speed during a race than slower runners [5]. With regards
to their training characteristics, ultra-marathoners have ~8 years of experience in ultra-run-
ning and show higher training volume and lower intensity than marathoners [1]. Training
characteristics, such as weekly training distance and training pace, are predictors of 100-km
ultra-marathon performance [6]. Also, anthropometric characteristics as age, body mass, body
mass index and body fat correlate with performance in this sport; however, they might be less
important than training characteristics [7, 8]. Furthermore, performance in 100-km ultra-mar-
athon running is influenced by nationality and origin of the runners [2].
It is well known that athletes of a certain origin are the fastest in certain sport disciplines.
For example in running, the fastest marathoners originate from East Africa, especially from
Kenya and Ethiopia [9]. In longer running distances such as 100-km ultra-marathon running,
most athletes at world class level originate from Japan [9]. When the top 10n athletes in
100-km ultra-marathon running between 1998 and 2011 were analysed, the 10 best race times
were achieved by Japanese runners for both women and men [2].
However, these results [2] might be biased since only the top athletes were considered and
not the whole number of athletes competing worldwide, and have not been verified by other
studies. The knowledge of the effect of nationality on 100-km ultra-marathon performance
would be of great practical importance for professionals working with ultra-marathoners.
Therefore, we analysed all 100-km ultra-marathon races held worldwide between 1959 and
2016 with the aim to identify the fastest runners by nationality. Based upon previous findings
we hypothesized to confirm that female and male Japanese runners would be the fastest world-
wide also when considering all finishers in 100-km ultra-marathon races held since 1959.
Materials and methods
Ethical approval
All procedures used in the study were approved by the Institutional Review Board of Kanton
St. Gallen, Switzerland with a waiver of the requirement for informed consent of the partici-
pants given the fact that the study involved the analysis of publicly available data.
Methods
We obtained from the website www.ultra-marathon.org/ of ‘Deutsche Ultramarathon Verei-
nigung’ (DUV). DUV has a large data record with race data from all ultra-marathons in
The fastest 100-km ultra-marathoners worldwide
PLOS ONE | https://doi.org/10.1371/journal.pone.0199701
July 11, 2018
2 / 29
Funding: The author(s) received no specific
funding for this work.
Competing interests: The authors have declared
that no competing interests exist.
http://statistik.d-u-v.org/index.php. By using the link http://statistik.d-u-v.org/ each person
can access the publicly available database. We used http://statistik.d-u-v.org/geteventlist.php
and inserted in ‘Year’ the term ‘all’, in ‘Distance’ the term ‘100 km’ and in ‘Country’ the term
‘All’ when using the English version of the website. By clicking on ‘Go’, all 100-km ultra-mar-
athons held worldwide are presented. This search leads to more than 4,500 races; all race
results were manually downloaded by one of the investigators.
The original dataset contains the following variables: name, age at race, year of race, sex of
finisher, nationality and country, and speed in km per hours. We converted running speed to
time in hours by dividing 100 km by speed (km/h). To identify unique finisher we computed
date of birth with year of race minus age at race. After cleaning the variable name we identified
unique finisher if finisher has the same name, date of birth, sex and nationality. Finishes with
missing in age, year, sex and times were removed. Finishes out of the following ranges were
removed: date of birth between 1890 and 2000, age between 15 and 100, year between 1950
and 2016, and number if character in nationality and country is 3. After removing the finishes
from missing data and outliers we selected only finishes which nationality has equal or more
than 900 finishes. This dataset has 24 nationalities and we named it complete dataset. To study
the distribution of time we produced histograms and density plots with Gaussian kernel for
each of the selected nationality. Also we produced normal distribution for each nationality
defined by the crude mean and standard deviation of each nationality to compare with the
empirical distribution. Furthermore we computed crude mean, standard deviation, median,
interquartile range (IQR), mode, skewness and excess kurtosis of time for each nationality.
Excess kurtosis or shortened excess is defined as kurtosis minus 3. An excess of 0 means a
Gaussian-like kurtosis (mesokurtic), a positive excess has a slender form of curve (leptokurtic),
and negative excess has a broader curve (platykurtic).
Due to various kind of distribution of time we decided to cluster the nationalities according
to time density. The range of time (h) was segmented in 0 to 7, 7 to 8, 8 to 9, . . .,22 to 23, 23 to
24 and 24. For each of these 19 segments we computed the area under der density curve.
With that, we performed an agglomerative hierarchical clustering using the group average
clustering to analyse groups of similar distribution of time.
To study the time between the nationalities we performed a linear regression analysis
adjusting by sex, age and year:
time ¼ sex ðyear þ year2Þ þ sex ðage þ age2Þ þ sex nationality
ð1Þ
We included a quadratic term for age and year and also an interaction term between sex
and age and age squared, sex and year and year squared and sex and nationality. This model
(1) is based on visual inspection of scatterplots of time against year and time against age for
each nationality (Figs 1 and 2). We included the following fitting curve to each panel: a b-
spline (solid) of age, respectively, time, and a quadratic term (dashed). Since both curves over-
laps for the large range using a quadratic term seems admissible. The variable age and year
were centered by the median with median year of 2009 and median age of 44. Reference level
of sex was male and reference level of nationality was Australia (AUS).
Histograms, density plots (Fig 3) and scatterplots (Figs 1 and 2) show that some races seem
to have a time limit: finishers who didn’t reach certain time limit were discarded. To account
for this limit we defined a time limit of 14 hours based on the plots and histograms of Japan,
Korea and Taiwan. From the complete dataset we removed the finishes which have more than
14 hours (Fig 4) which we called the truncated dataset and produced the same descriptive plots
and analysis as for the complete dataset (Figs 5 and 6). Additionally to the linear regression we
performed a truncated regression with the truncated dataset to allow conclusion to the whole
The fastest 100-km ultra-marathoners worldwide
PLOS ONE | https://doi.org/10.1371/journal.pone.0199701
July 11, 2018
3 / 29
sample. The estimates from the regressions were used to compute the times of a reference fin-
isher: median of age, median of year and male. These times and resulting ranks were compared
between the various nationalities.
Furthermore, to study if there is a difference between races at home and races on abroad
we included in model (1) an interaction term race site (home/abroad) with nationality and sex
(2):
time ¼ sex ðyear þ year2Þ þ sex ðage þ age2Þ þ sex nationality site
ð2Þ
Since 64.6% of the finisher had only one race and 16.6% had two races we performed
regression analysis without including the cluster effect of finishers.
All data processing and analysis were performed with the statistical software R [10]. Trun-
cated regression was performed with function truncreg from package truncreg.
Results
Between 1959 and 2016, a total of 363,924 athletes finished a 100-km ultra-marathon. The vari-
able with the highest number of missing data is date of birth. There was no missing in variable
sex. Table 1 summarizes the exclusion criteria. Only nationalities with at least 900 finishes
were considered to allow precise estimates and robust histogram. To analyse which country
have the most missing data in variable date of birth nationalities with at least 1,000 finishes
and at least 10% of missing data are listed in Table 2 by descending order of missing data.
Malaysia, Korea, Portugal and Great Britain have the most missing data with 70.1%, 60.6%,
Fig 1. Scatterplots time against race year for each nationality based on the complete dataset. Year has been jittered.
https://doi.org/10.1371/journal.pone.0199701.g001
The fastest 100-km ultra-marathoners worldwide
PLOS ONE | https://doi.org/10.1371/journal.pone.0199701
July 11, 2018
4 / 29
48.1%, 41.5%, respectively. Finally, a total of 150,710 finishers originating from 24 countries
with a total of 307,871 finishes could be considered for data analysis.
Table 3 presents the number of finishes by origin of the athletes. Most of the finishes were
achieved by runners from Japan, Germany, Switzerland, France, Italy and USA with more
than 260’000 (85%) finishes.
A total of 20 nationalities performed more than 50% of their races at their home country with
runners from Japan, Switzerland, Italy and Korea on the top whereas runners from Germany,
Great Britain, Belgium and Austria have performed less than 50% of the races abroad (Fig 7).
Runners from Finland, Germany, Switzerland, Italy, Netherlands Hungary, Belgium, Austria,
France and Russia have an average number of finishes per finisher of more than 2 (Fig 8).
A total of 64.6% of the finishers completed only one 100-km ultra-marathon (Table 4). On
average, the athletes were 43.7±11.1 years old (Table 5). A total of 88% of the finishers are men
and 12% are women (Table 6).
From the agglomerative hierarchical clustering 5 groups can be retrieved:
• Group 1 with China and Hong Kong which show a wide spread distribution of time.
• Group 2 with Russia which has the lowest mode and high excess (3.2).
• Group 3 with Korea, Japan and Taiwan which have a very high excess (7.1, 5.8 and 16.8). All
have a very step right curve at 18, 14 and 14 hours, respectively.
• Group 4 with Czech Republic, Spain, Great Britain, Australia, Switzerland, Italy and USA
with low excess and low skewness.
Fig 2. Scatterplots time against race age for each nationality based on the complete dataset. Age has been jittered.
https://doi.org/10.1371/journal.pone.0199701.g002
The fastest 100-km ultra-marathoners worldwide
PLOS ONE | https://doi.org/10.1371/journal.pone.0199701
July 11, 2018
5 / 29
• Group 5 with Finland, Denmark, Nederland, Belgium, Hungary, Poland, Sweden, France,
Canada, Austria and Germany which have a higher skewness (1) compared to group 4
with skewness 1 (Fig 4).
Table 7 compares the number of finishes before and after truncation of the data set for
each nationality. Hong Kong and China have more than 90% truncated observation, Australia,
Czech Republic, Spain, Switzerland and USA have between 50% and 63% truncated observa-
tion and all others less than 50%. Table 8 presents the mean, SD, median, interquartiles, mode,
skewness and excess of time for each nationality of the complete dataset and Table 9 for the
truncated data set.
Estimates, standard errors and p-values from models (1) and (2) based on complete and
truncated dataset are given in Tables 10–14. These data were used to compute times at the ref-
erence sex = male, year of race = 2009 and age = 44 which are presented in Tables 15 and 16
and Figs 9 and 10.
The upper panel of Fig 9(A)–9(C) display the time for each nationality and confidence
intervals from the multivariable model (1) at reference sex = male, age = 44 and year = 2009.
(A) was computed with the complete data set, (B) with the truncated data set and (C) was per-
formed by truncated regression and truncated data set. The ranks from the results with com-
plete data set show that runners from Russia, Finland, and Hungary were the fastest and
Fig 3. Histograms, density plots and normal distributions based on mean and standard deviation for each country. The diagrams are
positioned according the hierarchical cluster analysis. Graphs are based on the complete dataset.
https://doi.org/10.1371/journal.pone.0199701.g003
The fastest 100-km ultra-marathoners worldwide
PLOS ONE | https://doi.org/10.1371/journal.pone.0199701
July 11, 2018
6 / 29
runners from Hong Kong and China were the slowest finishers. The ranks from the results
from the truncated dataset changed to Russia, Hungary, Spain and Great Britain.
To visualize changes in ranks between these three methods the estimates were ordered by
descending time estimates and the nationality of the estimates were connected with lines. Fig 9
(D) shows the changes between results from linear regression with complete data set and trun-
cated data set and Fig 9(E) shows the results from linear regression with the truncated data set
and truncated regression. Comparing the regression of complete with truncated data set shows
that Russia, Canada, Hong Kong and China hold their position where all other nationalities
change their rank. Japan changed from rank 5 to rank 18. Hungary changed from third to sec-
ond and Finland from second to fourth. The time for Russia changed from 9.4 h (95%-CI: 9.1–
9.6) to 9.0 h (95%-CI: 8.9–9.1), Hungary from 10.7 h (95%-CI 10.4–11.0) to 9.9 h (95%-CI:
9.7–10.0), Japan from 11.1 h (95%-CI: 10.9–11.3) to 11.4 h (95%-CI: 11.3–11.5) and China
from 19.1 h (95%-CI: 18.9–19.3) to 11.9 h (95%-CI: 11.6–12.1).
There are only four nationalities which change ranks when ranks from linear regression are
compared with ranks from truncated regression both based on truncated dataset (Fig 9E). Fig
10(A) and 10(B) shows changes of ranks between running at home and abroad computed with
Fig 4. Scatterplot with excess against skewness. Groups of nation are distinguished by different colours.
https://doi.org/10.1371/journal.pone.0199701.g004
The fastest 100-km ultra-marathoners worldwide
PLOS ONE | https://doi.org/10.1371/journal.pone.0199701
July 11, 2018
7 / 29
complete data set, respectively, truncated data set based on model (2). Both show many
changes in rank position. Again Russia remains at place 1 at home and abroad with 10.0 h
and 8.2 h, respectively, with complete data set and 9.7 h and 7.8 with truncated data set. Japan
changed from rank 10 (11.1 h, at home) to rank 20 (14.9 h, abroad) when complete dataset was
used and from rank 18 (11.4 h) to 3 (9.2 h) when truncated dataset was used (Table 16).
Table 17 shows the mean time of the top 10, 100 and 1000 finishers. Only the fastest finishes
of each finisher was considered to define the top. Japan is first at top 10 and top 100 and sec-
ond at top 1000 whereas Russia is second at top 10 and top 100 and ninth at top 1000. Table 18
shows the mean time of the top fastest finishes. Russia is first at top 10 and 100 with 5 finisher
and 10 finishes and 37 finishers with 100 finishes. Japan is second at top 10 and 100 with 7 fin-
ishers and 10 finishes and 50 finishes with 100 finishes.
Discussion
The aim of this study was to investigate the aspect of nationality of the fastest 100-km ultra-
marathons competing between 1959 and 2016 with the hypothesis that the fastest runners
would originate from Japan as it has been found for the top 10 runners worldwide competing
between 1998 and 2011. However, in contrast to previous findings, athletes from Russia
achieved the fastest race times, not athletes from Japan, when all athletes were considered by
nationality.
A first potential explanation could be the quote of finishes at the origin country. For exam-
ple, Russians have ~37% of the finishes outside the origin but Japanese less than 2%. Most
probably only the fastest Russian runners compete outside Russia on the fastest races (e.g.
Fig 5. Scatterplots time against race year for each nationality based on the truncated dataset. Year has been jittered.
https://doi.org/10.1371/journal.pone.0199701.g005
The fastest 100-km ultra-marathoners worldwide
PLOS ONE | https://doi.org/10.1371/journal.pone.0199701
July 11, 2018
8 / 29
World Championships) or the fastest courses (e.g. completely flat course, track races) world-
wide. In contrast, Japanese runners competed preferably in races held in Japan where the
courses are most probably not fast (i.e. rather hilly courses than flat courses). The present
study is, however, not the first to show that Russian athletes are the fastest in an ultra-endur-
ance sport. Recently, an analysis of the ‘Engadin Ski Marathon’ showed that Russians were the
fastest cross-country skiers [11], which, combined with the findings of the present study, indi-
cated a general trend of excellence of Russians in ultra-endurance sports.
The two strongest factors which seems influence the population speed are the attitude to
participate and rules concerning time limits. Firstly, as the density plots and histograms show
it seems that there are countries where also very slow participants were allowed to compete in
races and who has been also considered in the ranking. Extreme examples of this kind of com-
petitions are athletes from the nationalities China, and Hong Kong, but also Czech Republic,
Great Britain, Spain, and Australia. Athletes from other countries like Denmark, Finland,
Fig 6. Scatterplots time against race age for each nationality based on the truncated dataset. Age has been jittered.
https://doi.org/10.1371/journal.pone.0199701.g006
Table 1. Total number, missing and out of range.
Criteria
N Finishes
N Finisher
N Nationality
% Finishes
% Finisher
% Nationality
[1] Total
363,924
195,983
128
100
100
100
[2] Exclude missing/incorrect hours
363,923
195,982
128
100
100
100
[3] Exclude missing age/date of birth
318,231
157,190
125
87.4
80.2
97.7
[4] Exclude unclear nationality
318,228
157,187
124
87.4
80.2
96.9
[5] Exclude nation < 900 finishes
307,871
150,710
24
84.6
76.9
18.8
https://doi.org/10.1371/journal.pone.0199701.t001
The fastest 100-km ultra-marathoners worldwide
PLOS ONE | https://doi.org/10.1371/journal.pone.0199701
July 11, 2018
9 / 29
Table 2. Missing data in date of birth and/or age according to nationality. Only nationalities with at least 10% miss-
ing are shown.
Nationality
N
Missing
Missing (%)
MAS
1,507
1,057
70.1
KOR
6,539
3,963
60.6
POR
1,442
693
48.1
GBR
5,834
2,419
41.5
NZL
1,119
452
40.4
FIN
2,359
917
38.9
HKG
2,123
791
37.3
CHN
6,165
1,942
31.5
ESP
5,913
1,854
31.4
TPE
3,843
1,065
27.7
JPN
79,011
17,021
21.5
DEN
1,224
263
21.5
AUS
5,103
1,093
21.4
CAN
3,093
417
13.5
NED
2,261
234
10.3
BEL
2,896
294
10.2
https://doi.org/10.1371/journal.pone.0199701.t002
Table 3. Quantity structure of selected nationalities.
Nationality
N Finishes
N Finisher
Race at home (%)
Finishes per finisher
JPN
61’990
41’081
98.8%
1.51
GER
51’313
18’085
39.7%
2.84
SUI
49’596
17’609
98.8%
2.82
FRA
46’553
22’768
89.6%
2.04
ITA
38’177
14’766
96.2%
2.59
USA
14’356
9’627
91.6%
1.49
POL
5’472
3’112
79.4%
1.76
CHN
4’223
3’069
75.2%
1.38
ESP
4’059
2’785
78.8%
1.46
AUS
4’010
2’437
92.0%
1.65
GBR
3’414
2’301
38.7%
1.48
TPE
2’778
1’955
88.4%
1.42
CAN
2’676
1’401
69.6%
1.91
BEL
2’602
1’163
38.5%
2.24
KOR
2’576
1’486
95.9%
1.73
CZE
2’506
1’289
66.6%
1.94
AUT
2’082
969
19.5%
2.15
NED
2’027
805
62.0%
2.52
RUS
1’852
920
62.6%
2.01
FIN
1’442
479
86.8%
3.01
HKG
1’332
1’070
89.9%
1.24
DEN
960
544
70.6%
1.76
HUN
947
419
65.6%
2.26
SWE
928
570
73.0%
1.63
https://doi.org/10.1371/journal.pone.0199701.t003
The fastest 100-km ultra-marathoners worldwide
PLOS ONE | https://doi.org/10.1371/journal.pone.0199701
July 11, 2018
10 / 29
Sweden, and Russia have a high skewness and excess which means that the bunch is concen-
trated over a narrow limit. This may due to attitudes within society (e.g. popularity of sports,
policy of furtherance) or socio-economic backgrounds of the individuals that only fast com-
petitors participates. It has been suggested also that a successful finish in this sport depends on
the motivation to train intensively [7].
Secondly, the density plots of athletes from Japan, Taiwan and Korea show a very steep
slope on the right side of the curve. This may due to time limits given by the organizer. These
time limits may not only performed in Japan or Taiwan but less frequently also in other coun-
tries. This could be the main source of bias which would explain why Japanese were the fastest
[1]. To counteract this bias we truncated the dataset and considered only finishes with lower
or equal 14 hours. This can cause bias like using top 10 finishers if we would conclude to the
whole population. So, we have to consider that using the complete dataset would give bias due
to policy rules and if we use the truncated dataset we have selection bias. We used also trun-
cated regression to allow conclusion to the whole running population but it seems that too
many observations have been truncated which changed the dataset in that way that it changed
completely the shape of the original distribution which does not anymore allow conclusion
on the complete data set but only on the truncated data set. That’s why the linear regression
and truncated regression of the truncated data set gives similar results. Nonetheless, the
Fig 7. Percentage of races which takes place at the origin of the finisher.
https://doi.org/10.1371/journal.pone.0199701.g007
The fastest 100-km ultra-marathoners worldwide
PLOS ONE | https://doi.org/10.1371/journal.pone.0199701
July 11, 2018
11 / 29
assumption that in Japan is a time limit may be supported by the fact that the race time at
home is 11.1 h and on abroad is 14.9 h using model (2) and complete data set. This is an
increase of 3.1 hours. For athletes from Russia, the times are 10.0 h and 8.2 h, respectively, a
decrease of 1.7 hours. Assuming that only good and the best ultra-marathoners take the effort
Fig 8. Average number of finishes. This figure is based on the complete dataset.
https://doi.org/10.1371/journal.pone.0199701.g008
Table 4. Distribution of number of finishes per finisher.
N Finishes
N Finisher
Frequency in %
1
97,340
64.6
2
25,032
16.6
3
10,588
7
4
5,561
3.7
5
3,558
2.4
5–9
5,365
3.6
10–19
2,708
1.8
20–39
498
0.3
40–59
51
0
60–79
4
0
80–99
1
0
100–149
2
0
https://doi.org/10.1371/journal.pone.0199701.t004
The fastest 100-km ultra-marathoners worldwide
PLOS ONE | https://doi.org/10.1371/journal.pone.0199701
July 11, 2018
12 / 29
Table 5. Baseline of continuous variables.
Variable
Mean
SD
Median
IQR
Min
Max
Age
43.7
11.1
44
36–51
15
92
Date of birth
1959
15.3
1960
1949–1970
1891
2000
Year
2003
14
2009
1993–2014
1959
2016
Time
13.7
3.82
12.9
11–15.7
6.17
46.8
https://doi.org/10.1371/journal.pone.0199701.t005
Table 6. Distribution of finishing according to categorical variables.
Variable
Level
N
Percent (%)
sex
Male
271,224
88
Female
36,647
12
nat
JPN
61,990
20
GER
51,313
17
SUI
49,596
16
FRA
46,553
15
ITA
38,177
12
USA
14,356
4.7
POL
5,472
1.8
CHN
4,223
1.4
ESP
4,059
1.3
AUS
4,010
1.3
GBR
3,414
1.1
TPE
2,778
0.9
CAN
2,676
0.87
BEL
2,602
0.85
KOR
2,576
0.84
CZE
2,506
0.81
AUT
2,082
0.68
NED
2,027
0.66
RUS
1,852
0.6
FIN
1,442
0.47
HKG
1,332
0.43
DEN
960
0.31
HUN
947
0.31
SWE
928
0.3
country
SUI
84,856
28
JPN
61,702
20
FRA
43,751
14
ITA
40,736
13
GER
21,943
7.1
USA
13,770
4.5
POL
4,404
1.4
AUS
4,223
1.4
ESP
3,922
1.3
NED
3,342
1.1
other (49)
28,564
9.3
https://doi.org/10.1371/journal.pone.0199701.t006
The fastest 100-km ultra-marathoners worldwide
PLOS ONE | https://doi.org/10.1371/journal.pone.0199701
July 11, 2018
13 / 29
Table 7. Numbers of finishes before and after truncation and percentage of removed finishes.
Nationality
N finishes
before truncation
N finishes
after truncation
Removed (%)
AUS
4,010
1,702
57.6
AUT
2,082
1,417
31.9
BEL
2,602
1,810
30.4
CAN
2,676
1,728
35.4
CHN
4,223
287
93.2
CZE
2,506
940
62.5
DEN
960
826
14
ESP
4,059
1,965
51.6
FIN
1,442
1,324
8.2
FRA
46,553
29,815
36
GBR
3,414
1,871
45.2
GER
51,313
37,412
27.1
HKG
1,332
109
91.8
HUN
947
774
18.3
ITA
38,177
20,021
47.6
JPN
61,990
56,777
8.4
KOR
2,576
1,483
42.4
NED
2,027
1,612
20.5
POL
5,472
2,970
45.7
RUS
1,852
1,648
11
SUI
49,596
21,981
55.7
SWE
928
774
16.6
TPE
2,778
2,286
17.7
USA
14,356
5,819
59.5
https://doi.org/10.1371/journal.pone.0199701.t007
Table 8. Mean, SD, median, interquartiles, mode, skewness and excess of time for each nationality of the complete dataset.
Nationality
Number of finishes
Mean (SD)
Median (IQ)
Minimum
Maximum
Mode
Skewness
Excess
AUS
4,010
15.2 (4.17)
15 (12.2–17.8)
6.62
37.9
13.6
0.62
1.256
AUT
2,082
13 (3.69)
12.2 (10.3–15.2)
7.11
27.6
10.8
0.884
0.29
BEL
2,602
12.7 (4.21)
11.8 (9.48–14.8)
6.26
31.7
9.58
1.035
0.88
CAN
2,676
13.8 (4.39)
12.7 (10.7–16)
6.68
35.6
11.6
1.22
1.535
CHN
4,223
20.7 (4.4)
20.9 (17.4–23.9)
6.31
32.3
22.9
-0.086
-0.545
CZE
2,506
16 (5)
16 (11.7–20.3)
6.3
38.2
15.7
0.02
-0.879
DEN
960
11.6 (3.36)
10.7 (9.69–12.3)
6.96
29.8
10.2
2.225
6.055
ESP
4,059
14.8 (5.3)
14.3 (9.96–19.1)
6.33
33.1
9.42
0.377
-0.864
FIN
1,442
11 (2.52)
10.7 (9.49–12.1)
6.51
32.8
10.1
2.068
8.999
FRA
46,553
13.6 (3.88)
12.8 (10.9–15.4)
6.39
36.6
11.7
1.288
2.475
GBR
3,414
13.9 (4.92)
13.4 (9.8–16.8)
6.17
36.4
8.56
0.713
0.222
GER
51,313
12.5 (3.47)
11.7 (9.9–14.4)
6.41
33.9
9.78
1.004
0.728
HKG
1,332
20.1 (4.14)
20 (17.4–23)
8.09
33.4
22.6
-0.072
-0.103
HUN
947
11.2 (3.36)
10.4 (8.82–12.7)
6.53
26.6
9.91
1.46
2.744
ITA
38,177
14 (3.04)
13.8 (11.9–16.1)
6.31
35.7
13.4
0.412
0.489
JPN
61,990
12.4 (2.13)
12.5 (11.2–13.4)
6.23
31.7
12.8
1.262
5.774
KOR
2,576
13.8 (2.76)
13.7 (12.4–14.7)
7.2
28.2
14.5
1.905
7.052
NED
2,027
12 (3.43)
11.2 (9.72–13.4)
6.64
34.2
10.2
1.55
3.787
POL
5,472
14.9 (5.41)
13.4 (10.8–17.3)
6.3
32.4
10.8
1.028
0.292
RUS
1,852
9.92 (3.23)
8.94 (7.56–11.3)
6.31
28.1
7.47
1.644
3.245
SUI
49,596
15.1 (4.03)
14.8 (11.8–18.2)
6.63
33.8
11.8
0.187
-0.89
SWE
928
12 (3.61)
11.4 (9.85–13)
6.38
44.2
11.2
2.226
9.605
TPE
2,778
13.1 (2.7)
12.9 (11.7–13.8)
7.62
46.8
13.6
2.903
16.754
USA
14,356
15.1 (3.96)
14.8 (12.6–17)
6.46
41.8
14.7
1.019
2.522
https://doi.org/10.1371/journal.pone.0199701.t008
The fastest 100-km ultra-marathoners worldwide
PLOS ONE | https://doi.org/10.1371/journal.pone.0199701
July 11, 2018
14 / 29
to go abroad the mean time should decrease which is not the case for Japanese runners. Never-
theless, both analyses with the complete and the truncated dataset show that Russian runners
were the fastest and athletes from China and Hong Kong were the slowest. All other nationali-
ties change their rankings reflecting the distribution of the running time.
If we look at the top 10, 100 and 1000 of the fastest finishers, Japanese ultra-marathoners
take the first place and the second place, respectively. The rank shifts to the rear the more par-
ticipants are included in the data set. It seems that there some very fast Japanese but as the
number of participants growths the mean time increase more than in other nationalities. A
look at the top 10, 100 of the finishes shows that Russian ultra-marathoners take the first
places. This is due to five runners with 10 finishes and 37 runners with 100 finishes. In this
case it seems that Russian ultra-marathoners take high ranks since some few runners achieved
some very fast finishes. This could be a limit of the linear regression if finishers are not consid-
ered in a multilevel regression as random variable. We also performed a linear regression with
finisher as random variable and we got similar results as in the linear regression with complete
and truncated data set (data not shown).
The role of nationality on 100-km ultra-marathon race times highlighted in the present
study was in agreement with previous research that identified sports as the most powerful
form of national performance [12]. An attempt to use sport to build a sense of national identity
has been reported [13]. Either biological or cultural heredity has been identified as a factor
associated with the dominance of a nationality in a sport [14]. For instance, certain genes have
been identified to relate to endurance performance [15]. In addition, an explanation of the
Table 9. Mean, SD, median, interquartiles, mode, skewness and excess of time for each nationality of the truncated dataset.
Nationality
N finishes
Mean (SD)
Median (IQ)
Minimum
Maximum
Mode
Skewness
Excess
AUS
1,702
11.4 (1.87)
11.7 (9.93–13)
6.62
14
13.4
-0.492
-0.782
AUT
1,417
10.9 (1.68)
10.9 (9.62–12.2)
7.11
14
10.5
-0.094
-0.896
BEL
1,810
10.4 (1.98)
10.3 (8.79–12.1)
6.26
14
9.58
0.015
-1.025
CAN
1,728
11.1 (1.71)
11.3 (9.84–12.6)
6.68
14
11.6
-0.249
-0.848
CHN
287
12.4 (1.29)
12.8 (11.6–13.4)
6.31
14
13.4
-1.152
1.482
CZE
940
10.6 (2)
10.6 (9.01–12.3)
6.3
14
9.56
-0.054
-1.053
DEN
826
10.5 (1.46)
10.4 (9.53–11.4)
6.96
14
10.1
0.119
-0.473
ESP
1,965
10.1 (1.88)
9.88 (8.66–11.5)
6.33
14
9.59
0.296
-0.801
FIN
1,324
10.5 (1.63)
10.5 (9.38–11.7)
6.51
14
10
0.009
-0.692
FRA
29,815
11.3 (1.64)
11.4 (10.1–12.6)
6.39
14
11.7
-0.352
-0.581
GBR
1,871
10.3 (2.14)
10.2 (8.41–12.1)
6.17
14
8.34
0.086
-1.231
GER
37,412
10.8 (1.69)
10.8 (9.48–12.1)
6.41
14
9.74
-0.027
-0.869
HKG
109
12.4 (1.39)
12.6 (11.4–13.6)
8.09
14
13.4
-0.882
0.019
HUN
774
9.91 (1.79)
9.84 (8.54–11.1)
6.53
14
10.1
0.244
-0.675
ITA
20,021
11.7 (1.6)
12 (10.6–13)
6.31
14
13.5
-0.643
-0.32
JPN
56,777
11.9 (1.5)
12.3 (11–13)
6.23
14
12.8
-0.846
0.155
KOR
1,483
12.3 (1.43)
12.7 (11.5–13.4)
7.2
14
13.5
-1.098
0.72
NED
1,612
10.6 (1.72)
10.5 (9.42–11.8)
6.64
14
10
-0.023
-0.696
POL
2,970
11 (1.64)
10.9 (10–12.2)
6.3
14
10.5
-0.312
-0.127
RUS
1,648
9.04 (1.93)
8.53 (7.39–10.5)
6.31
14
7.23
0.623
-0.647
SUI
21,981
11.3 (1.71)
11.5 (9.95–12.7)
6.63
14
11.7
-0.323
-0.814
SWE
774
10.7 (1.76)
10.9 (9.53–12)
6.38
14
11.3
-0.31
-0.627
TPE
2,286
12.3 (1.31)
12.6 (11.4–13.4)
7.62
14
13.6
-0.74
-0.135
USA
5,819
11.7 (1.79)
12 (10.5–13.1)
6.46
14
13.5
-0.776
-0.245
https://doi.org/10.1371/journal.pone.0199701.t009
The fastest 100-km ultra-marathoners worldwide
PLOS ONE | https://doi.org/10.1371/journal.pone.0199701
July 11, 2018
15 / 29
Table 10. Results from linear regression with complete dataset time = sex×(year+year2)+sex×(age+age2)+sex×nationality and referenced to male, age 44, year 2009
and nationality Australia.
Coefficient
Standard error
P-Value
Intercept
13.879
0.0600
0.000
Sex (female)
0.892
0.1254
0.000
Age
0.012
0.0006
0.000
Age squared
0.0033
0.0000
0.000
Year
0.156
0.0013
0.000
Year squared
0.0062
0.0000
0.000
Female×Age
0.022
0.0019
0.000
Female×Age squared
-0.0006
0.0001
0.000
Female×Year
0.014
0.0038
0.000
Female×Year squared
0.0016
0.0001
0.000
AUT
-1.922
0.0976
0.000
BEL
-1.713
0.0905
0.000
CAN
-0.976
0.0967
0.000
CHN
5.199
0.0808
0.000
CZE
1.104
0.0927
0.000
DEN
-2.806
0.1301
0.000
ESP
-0.069
0.0801
0.389
FIN
-3.634
0.1126
0.000
FRA
-0.898
0.0621
0.000
GBR
-0.597
0.0867
0.000
GER
-2.075
0.0634
0.000
HKG
4.708
0.1152
0.000
HUN
-3.176
0.1334
0.000
ITA
-0.378
0.0629
0.000
JPN
-2.764
0.0613
0.000
KOR
-1.425
0.0895
0.000
NED
-2.503
0.0985
0.000
POL
0.220
0.0758
0.004
RUS
-4.524
0.1064
0.000
SUI
-0.320
0.0642
0.000
SWE
-2.678
0.1339
0.000
TPE
-1.956
0.0881
0.000
USA
0.079
0.0676
0.244
AUT×Female
-0.705
0.2708
0.009
BEL×Female
-0.245
0.2660
0.357
CAN×Female
0.330
0.1868
0.077
CHN×Female
-0.070
0.1939
0.718
CZE×Female
0.629
0.2280
0.006
DEN×Female
-0.638
0.3222
0.048
ESP×Female
1.444
0.2405
0.000
FIN×Female
0.149
0.2622
0.569
FRA×Female
-0.038
0.1341
0.778
GBR×Female
-1.098
0.1905
0.000
GER×Female
-0.441
0.1372
0.001
HKG×Female
0.476
0.2756
0.084
HUN×Female
-0.856
0.2987
0.004
ITA×Female
-0.418
0.1374
0.002
JPN×Female
-0.661
0.1293
0.000
KOR×Female
0.304
0.3260
0.351
NED×Female
-0.548
0.2645
0.038
POL×Female
0.473
0.2011
0.019
RUS×Female
-0.358
0.2195
0.102
SUI×Female
0.583
0.1439
0.000
SWE×Female
-1.207
0.3046
0.000
TPE×Female
-0.586
0.2747
0.033
USA×Female
0.028
0.1383
0.840
https://doi.org/10.1371/journal.pone.0199701.t010
The fastest 100-km ultra-marathoners worldwide
PLOS ONE | https://doi.org/10.1371/journal.pone.0199701
July 11, 2018
16 / 29
Table 11. Results from linear regression with truncated dataset time = sex×(year+year2)+sex×(age+age2))+sex×nationality and referenced to male, age 44, year 2009
and nationality Australia.
Coefficient
Standard error
P-Value
Intercept
11.110
0.0427
0.000
Sex (female)
0.047
0.0969
0.628
Age
0.021
0.0004
0.000
Age squared
0.0011
0.0000
0.000
Year
0.070
0.0008
0.000
Year squared
0.0024
0.0000
0.000
Female×Age
0.007
0.0013
0.000
Female×Age squared
-0.0004
0.0001
0.000
Female×Year
0.015
0.0025
0.000
Female×Year squared
0.0000
0.0001
0.857
AUT
-0.319
0.0614
0.000
BEL
-0.705
0.0575
0.000
CAN
-0.094
0.0615
0.128
CHN
0.758
0.1064
0.000
CZE
-0.561
0.0690
0.000
DEN
-0.764
0.0730
0.000
ESP
-1.073
0.0559
0.000
FIN
-0.880
0.0636
0.000
FRA
0.058
0.0438
0.182
GBR
-0.740
0.0589
0.000
GER
-0.316
0.0441
0.000
HKG
0.787
0.1674
0.000
HUN
-1.241
0.0762
0.000
ITA
0.515
0.0443
0.000
JPN
0.261
0.0431
0.000
KOR
0.579
0.0595
0.000
NED
-0.553
0.0596
0.000
POL
-0.137
0.0522
0.009
RUS
-2.124
0.0616
0.000
SUI
0.242
0.0449
0.000
SWE
-0.620
0.0759
0.000
TPE
0.678
0.0545
0.000
USA
0.353
0.0484
0.000
AUT×Female
0.222
0.1713
0.195
BEL×Female
0.148
0.1704
0.386
CAN×Female
0.393
0.1282
0.002
CHN×Female
0.127
0.3268
0.698
CZE×Female
0.450
0.1899
0.018
DEN×Female
0.436
0.1792
0.015
ESP×Female
0.693
0.1996
0.001
FIN×Female
0.754
0.1505
0.000
FRA×Female
0.304
0.1015
0.003
GBR×Female
-0.434
0.1321
0.001
GER×Female
0.521
0.1015
0.000
HKG×Female
-0.137
0.4468
0.759
HUN×Female
0.068
0.1701
0.691
ITA×Female
0.134
0.1040
0.196
JPN×Female
0.164
0.0979
0.094
KOR×Female
0.022
0.2418
0.928
NED×Female
0.385
0.1582
0.015
POL×Female
0.600
0.1476
0.000
RUS×Female
0.534
0.1325
0.000
SUI×Female
0.536
0.1085
0.000
SWE×Female
0.051
0.1715
0.767
TPE×Female
-0.158
0.1617
0.327
USA×Female
0.108
0.1081
0.317
https://doi.org/10.1371/journal.pone.0199701.t011
The fastest 100-km ultra-marathoners worldwide
PLOS ONE | https://doi.org/10.1371/journal.pone.0199701
July 11, 2018
17 / 29
Table 12. Results from truncated regression with truncated dataset time = sex×(year+year2)+sex×(age+age2) + sex×nationality and referenced to male, age 44, year
2009 and nationality Australia.
Coefficient
Standard error
P-Value
Intercept
11.490
0.067
0.000
Sex (female)
0.003
0.153
0.985
Age
0.038
0.001
0.000
Age squared
0.002
0.000
0.000
Year
0.109
0.001
0.000
Year squared
0.004
0.000
0.000
Female×Age
0.017
0.002
0.000
Female×Age squared
-0.000
0.000
0.123
Female×Year
0.035
0.004
0.000
Female×Year squared
0.001
0.000
0.010
AUT
-0.443
0.093
0.000
BEL
-0.912
0.086
0.000
CAN
-0.087
0.095
0.358
CHN
1.672
0.215
0.000
CZE
-0.717
0.102
0.000
DEN
-0.985
0.107
0.000
ESP
-1.343
0.083
0.000
FIN
-1.138
0.094
0.000
FRA
0.109
0.069
0.115
GBR
-0.935
0.088
0.000
GER
-0.405
0.069
0.000
HKG
1.454
0.330
0.000
HUN
-1.538
0.109
0.000
ITA
0.856
0.070
0.000
JPN
0.497
0.068
0.000
KOR
1.226
0.106
0.000
NED
-0.698
0.090
0.000
POL
-0.174
0.081
0.031
RUS
-2.464
0.089
0.000
SUI
0.389
0.071
0.000
SWE
-0.933
0.112
0.000
TPE
1.390
0.096
0.000
USA
0.614
0.078
0.000
AUT×Female
0.377
0.259
0.146
BEL×Female
0.676
0.255
0.008
CAN×Female
0.625
0.201
0.002
CHN×Female
0.245
0.701
0.726
CZE×Female
0.496
0.284
0.081
DEN×Female
0.672
0.268
0.012
ESP×Female
0.871
0.296
0.003
FIN×Female
0.852
0.232
0.000
FRA×Female
0.570
0.161
0.000
GBR×Female
-0.414
0.196
0.034
GER×Female
0.858
0.160
0.000
HKG×Female
-0.442
0.839
0.599
HUN×Female
0.267
0.242
0.270
ITA×Female
0.364
0.167
0.029
JPN×Female
0.368
0.156
0.018
KOR×Female
0.074
0.468
0.875
NED×Female
0.613
0.243
0.011
POL×Female
1.219
0.241
0.000
RUS×Female
0.699
0.194
0.000
SUI×Female
0.950
0.173
0.000
SWE×Female
0.311
0.258
0.228
TPE×Female
-0.190
0.294
0.517
USA×Female
0.218
0.173
0.208
https://doi.org/10.1371/journal.pone.0199701.t012
The fastest 100-km ultra-marathoners worldwide
PLOS ONE | https://doi.org/10.1371/journal.pone.0199701
July 11, 2018
18 / 29
Table 13. Interaction with race site, results from truncated regression with complete data set time = sex×(year+year2)+sex×(age+age2) + sex×nationality×site and
referenced to male, age 44, year 2009 and nationality Austria.
Coefficient
Standard error
P-Value
Intercept
14.043
0.0615
0.000
Sex (female)
1.029
0.1296
0.000
Age
0.013
0.0006
0.000
Age squared
0.0032
0.0000
0.000
Year
0.157
0.0013
0.000
Year squared
0.0062
0.0000
0.000
0.157
0.0013
0.000
Female×Age
0.021
0.0019
0.000
Female×Age squared
-0.0007
0.0001
0.000
Female×Year
0.009
0.0039
0.015
Female×Year squared
0.0016
0.0001
0.000
AUT
-3.621
0.1796
0.000
BEL
-3.480
0.1234
0.000
CAN
-1.300
0.1076
0.000
CHN
5.201
0.0869
0.000
CZE
2.448
0.1055
0.000
DEN
-3.661
0.1475
0.000
ESP
-0.399
0.0851
0.000
FIN
-3.773
0.1179
0.000
FRA
-1.332
0.0637
0.000
GBR
-2.477
0.1188
0.000
GER
-2.968
0.0673
0.000
HKG
4.780
0.1187
0.000
HUN
-3.101
0.1529
0.000
ITA
-0.511
0.0643
0.000
JPN
-2.961
0.0626
0.000
KOR
-1.707
0.0907
0.000
NED
-3.523
0.1151
0.000
POL
0.632
0.0801
0.000
RUS
-4.060
0.1229
0.000
SUI
-0.424
0.0656
0.000
SWE
-3.181
0.1519
0.000
TPE
-2.170
0.0910
0.000
USA
-0.012
0.0693
0.867
AUT×race abroad
3.994
0.2913
0.000
BEL×race abroad
4.684
0.2605
0.000
CAN×race abroad
2.645
0.2770
0.000
CHN×race abroad
1.420
0.2545
0.000
CZE×race abroad
-2.328
0.2665
0.000
DEN×race abroad
4.545
0.3366
0.000
ESP×race abroad
2.878
0.2568
0.000
FIN×race abroad
1.972
0.3570
0.000
FRA×race abroad
4.657
0.2278
0.000
GBR×race abroad
4.787
0.2561
0.000
GER×race abroad
3.308
0.2240
0.000
HKG×race abroad
-0.484
0.4050
0.232
HUN×race abroad
1.293
0.3396
0.000
ITA×race abroad
1.596
0.2406
0.000
JPN×race abroad
5.831
0.2636
0.000
KOR×race abroad
5.142
0.4134
0.000
NED×race abroad
4.362
0.2723
0.000
(Continued)
The fastest 100-km ultra-marathoners worldwide
PLOS ONE | https://doi.org/10.1371/journal.pone.0199701
July 11, 2018
19 / 29
Table 13. (Continued)
Coefficient
Standard error
P-Value
POL×race abroad
-0.624
0.2490
0.012
RUS×race abroad
0.267
0.2873
0.353
SUI×race abroad
2.078
0.2632
0.000
SWE×race abroad
3.296
0.3443
0.000
TPE×race abroad
2.539
0.3055
0.000
USA×race abroad
1.300
0.2495
0.000
AUT×female
-1.319
0.5884
0.025
BEL×female
-0.744
0.4024
0.064
CAN×female
0.792
0.2117
0.000
CHN×female
-0.483
0.2139
0.024
CZE×female
0.357
0.2593
0.169
DEN×female
-0.611
0.3860
0.113
ESP×female
0.849
0.2644
0.001
FIN×female
-0.351
0.2769
0.205
FRA×female
-0.068
0.1385
0.626
GBR×female
-0.251
0.2474
0.310
GER×female
-0.913
0.1526
0.000
HKG×female
0.650
0.2918
0.026
HUN×female
-1.180
0.4018
0.003
ITA×female
-0.491
0.1413
0.001
JPN×female
-0.781
0.1331
0.000
KOR×female
-0.311
0.3419
0.363
NED×female
-0.730
0.3210
0.023
POL×female
0.025
0.2143
0.906
RUS×female
-0.436
0.2716
0.109
SUI×female
0.442
0.1477
0.003
SWE×female
-1.042
0.3398
0.002
TPE×female
-0.522
0.3117
0.094
USA×female
0.092
0.1427
0.518
AUT× race abroad×female
-1.319
0.5884
0.092
BEL× race abroad×female
-0.744
0.4024
0.043
CAN× race abroad×female
0.792
0.2117
0.053
CHN× race abroad×female
-0.483
0.2139
0.000
CZE× race abroad×female
0.357
0.2593
0.291
DEN× race abroad×female
-0.611
0.3860
0.771
ESP× race abroad×female
0.849
0.2644
0.000
FIN× race abroad×female
-0.351
0.2769
0.000
FRA× race abroad×female
-0.068
0.1385
0.556
GBR× race abroad×female
-0.251
0.2474
0.335
GER× race abroad×female
-0.913
0.1526
0.004
HKG× race abroad×female
0.650
0.2918
0.982
HUN× race abroad×female
-1.180
0.4018
0.030
ITA× race abroad×female
-0.491
0.1413
0.356
JPN× race abroad×female
-0.781
0.1331
0.004
KOR× race abroad×female
-0.311
0.3419
0.011
NED× race abroad×female
-0.730
0.3210
0.189
POL× race abroad×female
0.025
0.2143
0.000
RUS× race abroad×female
-0.436
0.2716
0.031
SUI× race abroad×female
0.442
0.1477
0.005
SWE× race abroad×female
-1.042
0.3398
0.816
TPE× race abroad×female
-0.522
0.3117
0.870
USA× race abroad×female
0.092
0.1427
0.005
https://doi.org/10.1371/journal.pone.0199701.t013
The fastest 100-km ultra-marathoners worldwide
PLOS ONE | https://doi.org/10.1371/journal.pone.0199701
July 11, 2018
20 / 29
Table 14. Interaction with race site, results from linear regression with truncated dataset time = sex×year+year2)+sex×(age+age2) + sex×nationality×site and refer-
enced to male, age 44, year 2009, site at home and nationality Australia.
Coefficient
Standard error
P-Value
Intercept
11.319
0.0447
0.000
Sex (female)
0.139
0.1045
0.182
Age
0.020
0.0004
0.000
Age squared
0.0011
0.0000
0.000
Year
0.068
0.0008
0.000
Year squared
0.0024
0.0000
0.000
Female×Age
0.005
0.0013
0.000
Female×Age squared
-0.0004
0.0001
0.000
Female×Year
0.006
0.0025
0.013
Female×Year squared
-0.0002
0.0001
0.152
AUT
-1.092
0.0953
0.000
BEL
-1.228
0.0701
0.000
CAN
-0.128
0.0668
0.056
CHN
0.601
0.1196
0.000
CZE
-0.314
0.0984
0.001
DEN
-0.831
0.0792
0.000
ESP
-1.188
0.0601
0.000
FIN
-0.994
0.0667
0.000
FRA
-0.136
0.0458
0.003
GBR
-1.322
0.0737
0.000
GER
-1.040
0.0472
0.000
HKG
0.841
0.2060
0.000
HUN
-0.986
0.0867
0.000
ITA
0.336
0.0463
0.000
JPN
0.074
0.0451
0.102
KOR
0.432
0.0609
0.000
NED
-0.856
0.0658
0.000
POL
-0.105
0.0567
0.063
RUS
-1.672
0.0706
0.000
SUI
0.001
0.0469
0.986
SWE
-0.661
0.0835
0.000
TPE
0.510
0.0568
0.000
USA
0.273
0.0507
0.000
AUT×race abroad
2.484
0.1627
0.000
BEL×race abroad
2.343
0.1501
0.000
CAN×race abroad
0.957
0.1663
0.000
CHN×race abroad
1.571
0.2566
0.000
CZE×race abroad
1.002
0.1705
0.000
DEN×race abroad
1.025
0.1950
0.000
ESP×race abroad
1.297
0.1558
0.000
FIN×race abroad
0.860
0.1944
0.000
FRA×race abroad
1.484
0.1350
0.000
GBR×race abroad
2.418
0.1524
0.000
GER×race abroad
2.584
0.1308
0.000
HKG×race abroad
1.065
0.3543
0.003
HUN×race abroad
0.197
0.1876
0.295
(Continued)
The fastest 100-km ultra-marathoners worldwide
PLOS ONE | https://doi.org/10.1371/journal.pone.0199701
July 11, 2018
21 / 29
Table 14. (Continued)
Coefficient
Standard error
P-Value
ITA×race abroad
0.735
0.1427
0.000
JPN×race abroad
-0.399
0.1659
0.016
KOR×race abroad
0.012
0.2824
0.965
NED×race abroad
2.018
0.1583
0.000
POL×race abroad
0.864
0.1450
0.000
RUS×race abroad
-0.102
0.1583
0.520
SUI×race abroad
0.884
0.1622
0.000
SWE×race abroad
1.021
0.1956
0.000
TPE×race abroad
1.460
0.1714
0.000
USA×race abroad
0.567
0.1488
0.000
AUT×female
-0.564
0.3079
0.067
BEL×female
0.207
0.2213
0.349
CAN×female
0.324
0.1445
0.025
CHN×female
0.293
0.3774
0.437
CZE×female
-0.191
0.2767
0.489
DEN×female
0.408
0.2037
0.045
ESP×female
0.377
0.2155
0.081
FIN×female
0.567
0.1591
0.000
FRA×female
0.296
0.1089
0.007
GBR×female
-0.032
0.1626
0.845
GER×female
0.371
0.1127
0.001
HKG×female
0.798
0.8050
0.321
HUN×female
-0.139
0.2253
0.536
ITA×female
0.161
0.1113
0.149
JPN×female
0.144
0.1053
0.171
KOR×female
0.088
0.2539
0.729
NED×female
0.278
0.1813
0.125
POL×female
0.191
0.1639
0.244
RUS×female
0.284
0.1617
0.079
SUI×female
0.454
0.1155
0.000
SWE×female
0.024
0.1890
0.897
TPE×female
-0.121
0.1845
0.512
USA×female
0.234
0.1167
0.045
AUT× race abroad×female
0.658
0.4180
0.115
BEL× race abroad×female
-0.506
0.3768
0.179
CAN× race abroad×female
-0.052
0.3147
0.868
CHN× race abroad×female
-0.906
0.7296
0.214
CZE× race abroad×female
0.692
0.4175
0.098
DEN× race abroad×female
-0.199
0.4214
0.637
ESP× race abroad×female
1.056
0.5267
0.045
FIN× race abroad×female
0.958
0.4499
0.033
FRA× race abroad×female
-1.505
0.2853
0.000
GBR× race abroad×female
-1.157
0.3115
0.000
GER× race abroad×female
-0.119
0.2609
0.648
HKG× race abroad×female
-1.186
0.9940
0.233
HUN× race abroad×female
0.672
0.3816
0.078
ITA× race abroad×female
-1.355
0.3009
0.000
(Continued)
The fastest 100-km ultra-marathoners worldwide
PLOS ONE | https://doi.org/10.1371/journal.pone.0199701
July 11, 2018
22 / 29
differences in participation among nationalities might be the differences in their attitudes
towards physical activity [16]. Participation in running might be influenced by economic
and cultural factors, e.g. those without a migration background are more likely to participate
in running than those with a migration background [17]. Another aspect affecting sport
Table 14. (Continued)
Coefficient
Standard error
P-Value
JPN× race abroad×female
-0.214
0.3048
0.482
KOR× race abroad×female
0.116
0.7603
0.878
NED× race abroad×female
-0.144
0.3698
0.697
POL× race abroad×female
0.900
0.3611
0.013
RUS× race abroad×female
0.585
0.3127
0.061
SUI× race abroad×female
-0.592
0.3508
0.091
SWE× race abroad×female
-0.452
0.4260
0.289
TPE× race abroad×female
-0.285
0.3870
0.461
USA× race abroad×female
-0.878
0.2887
0.002
https://doi.org/10.1371/journal.pone.0199701.t014
Table 15. Comparing times in hours with finishes performed. Times were computed based on the model time = sex×(year+year2)+sex×(age+age2) + sex×nationality
and referenced to male, age 44, year 2009 and nationality Australia. In parentheses: 95%-CI.
A
B
% Difference
C
Data: complete
Linear regression with fixed effects
Data: truncated at 15 hours
Linear regression with fixed effects
Between A and B
(B-A)/A
Data: truncated at 15 hours
Truncated linear regression with
fixed effects
AUS
13.9
(13.8–14.0)
11.1
(11.0–11.2)
-20.0%
11.5
(11.4–11.6)
AUT
12.0
(11.7–12.2)
10.8
(10.6–10.9)
-9.8%
11.0
(10.8–11.3)
BEL
12.2
(12.0–12.4)
10.4
(10.3–10.5)
-14.5%
10.6
(10.4–10.8)
CAN
12.9
(12.7–13.1)
11.0
(10.9–11.2)
-14.6%
11.4
(11.2–11.6)
CHN
19.1
(18.9–19.3)
11.9
(11.6–12.1)
-37.8%
13.2
(12.7–13.6)
CZE
15.0
(14.8–15.2)
10.5
(10.4–10.7)
-29.6%
10.8
(10.5–11.0)
DEN
11.1
(10.8–11.4)
10.3
(10.2–10.5)
-6.6%
10.5
(10.3–10.8)
ESP
13.8
(13.6–14.0)
10.0
(9.9–10.2)
-27.3%
10.1
(9.9–10.4)
FIN
10.2
(10.0–10.5)
10.2
(10.1–10.4)
-0.1%
10.4
(10.1–10.6)
FRA
13.0
(12.8–13.2)
11.2
(11.0–11.3)
-14.0%
11.6
(11.4–11.8)
GBR
13.3
(13.1–13.5)
10.4
(10.2–10.5)
-21.9%
10.6
(10.3–10.8)
GER
11.8
(11.6–12.0)
10.8
(10.7–10.9)
-8.6%
11.1
(10.9–11.3)
HKG
18.6
(18.3–18.8)
11.9
(11.6–12.2)
-36.0%
12.9
(12.3–13.6)
HUN
10.7
(10.4–11.0)
9.9
(9.7–10.0)
-7.8%
10.0
(9.7–10.2)
ITA
13.5
(13.3–13.7)
11.6
(11.5–11.7)
-13.9%
12.3
(12.2–12.5)
JPN
11.1
(10.9–11.3)
11.4
(11.3–11.5)
2.3%
12.0
(11.8–12.2)
KOR
12.5
(12.2–12.7)
11.7
(11.5–11.8)
-6.1%
12.7
(12.5–13.0)
NED
11.4
(11.2–11.6)
10.6
(10.4–10.7)
-7.2%
10.8
(10.6–11.0)
POL
14.1
(13.9–14.3)
11.0
(10.8–11.1)
-22.2%
11.3
(11.1–11.5)
RUS
9.4
(9.1–9.6)
9.0
(8.8–9.1)
-3.9%
9.0
(8.8–9.2)
SUI
13.6
(13.4–13.7)
11.4
(11.2–11.5)
-16.3%
11.9
(11.7–12.1)
SWE
11.2
(10.9–11.5)
10.5
(10.3–10.7)
-6.3%
10.6
(10.3–10.8)
TPE
11.9
(11.7–12.1)
11.8
(11.7–11.9)
-1.1%
12.9
(12.7–13.1)
USA
14.0
(13.8–14.1)
11.5
(11.3–11.6)
-17.9%
12.1
(11.9–12.3)
https://doi.org/10.1371/journal.pone.0199701.t015
The fastest 100-km ultra-marathoners worldwide
PLOS ONE | https://doi.org/10.1371/journal.pone.0199701
July 11, 2018
23 / 29
performance is doping, for which no accurate rates exist due to its undisclosed practice; how-
ever, its prevalence has been estimated as 14–39% in adult elite athletes and has been shown to
vary by performance level and nationality [18].
Since it has been shown that the role of nationality might vary by distance in endurance
and ultra-endurance running [9], the findings of the present study should not be generalized
to other distances. On the other hand, strength of the study was its methodological approach
to the research question: first, it used one of the larger samples of 100-km ultra-marathoners
ever studied, second, it considered a probably applied time limit barrier by using a truncated
data set, third, we compared adjusted times of a reference finisher to compare ranks, and
forth, we provide a time distribution classification which helps to interpret the results.
We found that ultra-marathoners from Russia were the fastest in this specific ultra-mara-
thon distance. Unfortunately, this kind of analysis is not able to explain the reason for this
dominance. The role of ethnicity in running is, however, well-known for other running dis-
tances. Running best performances are dominated by a few groups of athletes including run-
ners with West African ancestry for the sprint distances and East African runners for the long
Table 16. Comparing times in hours with finishes performed at home. Times were computed based on the model time = ex×(year+year2)+sex×(age+age2) + sex×natio-
nality×site and referenced to male, age 44, year 2009 and nationality Australia.
A
% Difference between
races abroad/at home
B
% Difference between races
at home/on abroad
C
% Difference between
races abroad/at home
Data: complete
Linear regression
with fixed effects
Data: truncated at 14
hours
Linear regression
with fixed effects
Data: truncated at 14
hours
Truncated linear
regression with fixed
effects
Races at
home
Races
abroad
Races at
home
Races
abroad
Races at
home
Races
abroad
AUS
14.0
12.0
11.3
11.3
9.6
-15.4%
11.7
9.6
-17.8%
AUT
10.4
12.4
10.2
10.2
11.0
7.2%
10.5
11.4
9.3%
BEL
10.6
13.2
10.1
10.1
10.7
5.9%
10.2
10.9
6.5%
CAN
12.7
13.3
11.2
11.2
10.4
-7.1%
11.7
10.6
-9.4%
CHN
19.2
18.6
11.9
11.9
11.7
-1.5%
13.1
13.2
0.7%
CZE
16.5
12.1
11.0
11.0
10.3
-6.8%
11.3
10.3
-8.8%
DEN
10.4
12.9
10.5
10.5
9.8
-6.9%
10.7
9.9
-7.2%
ESP
13.6
14.5
10.1
10.1
9.7
-4.4%
10.3
10.0
-3.2%
FIN
10.3
10.2
10.3
10.3
9.4
-8.6%
10.4
9.3
-10.6%
FRA
12.7
15.3
11.2
11.2
10.9
-2.4%
11.6
11.2
-3.4%
GBR
11.6
14.3
10.0
10.0
10.7
6.7%
10.1
10.9
8.1%
GER
11.1
12.3
10.3
10.3
11.1
8.1%
10.4
11.5
10.6%
HKG
18.8
16.3
12.2
12.2
11.5
-5.6%
13.1
11.2
-14.5%
HUN
10.9
10.2
10.3
10.3
8.8
-15.0%
10.5
8.9
-15.3%
ITA
13.5
13.1
11.7
11.7
10.6
-8.7%
12.4
10.8
-12.9%
JPN
11.1
14.9
11.4
11.4
9.2
-18.8%
12.0
9.4
-21.4%
KOR
12.3
15.4
11.8
11.8
10.0
-14.8%
12.8
10.3
-19.4%
NED
10.5
12.8
10.5
10.5
10.7
2.6%
10.6
11.0
4.0%
POL
14.7
12.0
11.2
11.2
10.3
-7.9%
11.6
10.5
-9.2%
RUS
10.0
8.2
9.6
9.6
7.8
-19.2%
9.5
7.9
-17.2%
SUI
13.6
13.7
11.3
11.3
10.5
-7.6%
11.8
10.4
-12.3%
SWE
10.9
12.1
10.7
10.7
9.9
-6.8%
10.7
9.8
-8.5%
TPE
11.9
12.4
11.8
11.8
11.5
-2.4%
12.9
12.2
-5.8%
USA
14.0
13.3
11.6
11.6
10.4
-10.2%
12.4
10.6
-14.1%
https://doi.org/10.1371/journal.pone.0199701.t016
The fastest 100-km ultra-marathoners worldwide
PLOS ONE | https://doi.org/10.1371/journal.pone.0199701
July 11, 2018
24 / 29
distances [9, 19]. For marathoners, East-African runners from Kenya and Ethiopia dominate
since decades running distances up to the marathon [9, 20–22] while for other running dis-
tances such as the sprint distances, runners from Jamaica are dominating [23].
For elite Kenyan runners, it is well-known that they are from a distinctive environmental
background in terms of geographical distribution, ethnicity and run to school [20]. Interest-
ingly, the same findings were reported for elite Ethiopian runners, who also have a distinct
environmental background in terms of geographical distribution, ethnicity, and also often run
to school [21]. So for both Kenyan and Ethiopian runners both environmental and social fac-
tors are important for their athletic success. These aspects are, however, not only for East Afri-
can distance runners, but also for sprinters from Jamaica of importance. It has been shown
that a higher proportion of middle distance and both jump and throw athletes walked to
school and travelled greater distances to school [23].
Although different anthropometric, physiological, biomechanical and training characteristics
are of importance for the East African running dominance [22, 24, 25], a strong psychological
Fig 9. The upper panel shows the adjusted time for each nationality in ascending order at reference sex = male,
age = 44 and year = 2009. (A) is based on linear regression of the complete dataset, (B) on the truncated dataset and
(C) on the truncated regression of the truncated dataset. The lower panel with figures (D) and (E) shows the changes in
rank from (A) to (B) and (B) to (C).
https://doi.org/10.1371/journal.pone.0199701.g009
The fastest 100-km ultra-marathoners worldwide
PLOS ONE | https://doi.org/10.1371/journal.pone.0199701
July 11, 2018
25 / 29
Fig 10. The rank of nationality computed from model 2 (interaction of nationality with races at home/abroad).
(A) shows rank changes from races at home to races abroad based on linear regression with complete dataset. (B)
shows rank changes from races at home to races abroad based on linear regression with truncated dataset.
https://doi.org/10.1371/journal.pone.0199701.g010
Table 17. Mean time of the top 10, 100 and 1000 of the fastest finishers for each nationality. Only the lowest time of a finishes was considered if one finisher had sev-
eral finishes.
Nationality
N of finishers
Mean time of top 10
Nationality
N of finishers
Mean time of top 100
Nationality
N of finishers
Mean time of top 1000
1
JPN
10
6.41
JPN
100
6.83
GER
1000
7.83
2
RUS
10
6.43
RUS
100
6.86
JPN
1000
7.97
3
FRA
10
6.52
GER
100
7.00
FRA
1000
8.19
4
GBR
10
6.55
FRA
100
7.01
SUI
1000
8.31
5
ESP
10
6.61
SUI
100
7.18
ITA
1000
8.81
6
BEL
10
6.62
GBR
100
7.29
USA
1000
9.46
7
USA
10
6.62
USA
100
7.31
POL
1000
10.24
8
GER
10
6.63
ITA
100
7.38
GBR
1000
10.80
9
POL
10
6.65
ESP
100
7.57
RUS
920
10.82
10
ITA
10
6.67
BEL
100
7.74
FIN
479
10.96
11
SUI
10
6.72
POL
100
7.96
ESP
1000
11.05
12
FIN
10
6.79
NED
100
8.06
TPE
1000
11.42
13
CZE
10
6.81
AUS
100
8.08
AUS
1000
11.60
14
HUN
10
6.81
HUN
100
8.18
DEN
544
11.62
15
NED
10
6.87
AUT
100
8.21
HUN
419
11.62
16
SWE
10
6.89
FIN
100
8.28
CAN
1000
11.86
17
AUS
10
6.95
CZE
100
8.34
NED
805
12.10
18
DEN
10
7.33
CAN
100
8.38
BEL
1000
12.14
19
AUT
10
7.34
SWE
100
8.40
SWE
570
12.18
20
CAN
10
7.36
DEN
100
8.61
KOR
1000
12.37
21
KOR
10
7.68
KOR
100
9.05
AUT
969
12.66
22
TPE
10
7.99
TPE
100
9.26
CZE
1000
14.81
23
CHN
10
8.85
CHN
100
11.14
CHN
1000
15.31
24
HKG
10
9.47
HKG
100
12.41
HKG
1000
19.22
https://doi.org/10.1371/journal.pone.0199701.t017
The fastest 100-km ultra-marathoners worldwide
PLOS ONE | https://doi.org/10.1371/journal.pone.0199701
July 11, 2018
26 / 29
motivation to succeed athletically for the purpose of economic and social advancement is
known [26]. Elite Kenyan runners become a competitive athlete due to economic reasons. Typi-
cally, Kenyan athletes see athletics as a means of making money to help their families, parents
and siblings [20, 27].
Practical applications for athletes and coaches
The present study advances existing theoretical knowledge as scientists will improve their
understanding of participation and performance trends by nationality in 100-km ultra-mara-
thon running which is relatively less studied compared to shorter distances such as sprint and
marathon distances. Moreover, coaches and runners can use the findings to optimize their
preparation and participation in a 100-km ultra-marathon. Athletes from other countries than
the dominating nationalities (i.e. Russia, Hungary) must be aware that they will most probably
not be able to reach a world-class level in 100-km ultra-marathon running.
Conclusions
In summary, in contrast to existing findings investigating the top 10 by nationality, this analy-
sis of all runners showed that ultra-marathoners from Russia, not Japan, were the fastest
100-km ultra-marathoners worldwide when considering all races held since 1959. Since we
know for the best sprinters and marathoners in the world that specific anthropometric, train-
ing and environmental characteristics are prevalent, future studies need to investigate why
Russian ultra-marathoners dominate the 100-km ultra-marathon distance.
Table 18. Mean time of the top 10, 100 and 1000 of the fastest finishes for each nationality.
Nationality
N of finishers
Mean time of top 10
Nationality
N of finishers
Mean time of top 100
Nationality
N of finishers
Mean time of top 1000
1
RUS
5
6.37
RUS
37
6.57
GER
332
7.35
2
JPN
7
6.37
JPN
50
6.68
FRA
358
7.52
3
BEL
3
6.44
FRA
35
6.74
JPN
557
7.60
4
POL
3
6.47
GER
34
6.80
RUS
351
7.70
5
GBR
7
6.47
GBR
30
6.86
SUI
393
7.74
6
ITA
4
6.49
BEL
23
6.88
ITA
362
8.02
7
FRA
7
6.50
ITA
30
6.94
GBR
402
8.54
8
ESP
4
6.50
ESP
28
6.96
USA
543
8.58
9
GER
5
6.54
SUI
47
6.96
ESP
365
8.60
10
CZE
3
6.55
USA
42
6.99
BEL
359
8.87
11
USA
7
6.60
POL
25
7.00
POL
503
9.25
12
SUI
4
6.68
HUN
29
7.25
NED
415
9.49
13
SWE
5
6.68
NED
29
7.32
FIN
358
9.83
14
HUN
5
6.70
CZE
24
7.33
CAN
463
9.94
15
FIN
7
6.72
AUS
42
7.55
AUT
532
10.04
16
NED
4
6.77
FIN
40
7.64
AUS
536
10.14
17
AUS
4
6.81
SWE
47
7.69
CZE
451
10.82
18
CAN
3
6.89
CAN
41
7.80
TPE
753
11.05
19
DEN
7
7.26
AUT
47
7.86
HUN
419
11.18
20
AUT
6
7.28
DEN
47
8.11
DEN
544
11.59
21
KOR
5
7.42
KOR
72
8.76
KOR
625
11.62
22
TPE
9
7.98
TPE
79
9.15
SWE
570
11.96
23
CHN
9
8.85
CHN
84
10.95
CHN
808
14.82
24
HKG
10
9.47
HKG
90
12.21
HKG
846
18.46
https://doi.org/10.1371/journal.pone.0199701.t018
The fastest 100-km ultra-marathoners worldwide
PLOS ONE | https://doi.org/10.1371/journal.pone.0199701
July 11, 2018
27 / 29
Author Contributions
Conceptualization: Beat Knechtle, Pantelis Theodoros Nikolaidis.
Data curation: Beat Knechtle, Fabio Valeri.
Formal analysis: Fabio Valeri.
Methodology: Beat Knechtle, Fabio Valeri.
Writing – original draft: Beat Knechtle, Pantelis Theodoros Nikolaidis, Fabio Valeri.
References
1.
Knechtle B. Ultramarathon runners: Nature or nurture? Int J Sports Physiol Perform. 2012; 7:310–2.
PMID: 23197583
2.
Cejka N, Rust CA, Lepers R, Onywera V, Rosemann T, Knechtle B. Participation and performance
trends in 100-km ultra-marathons worldwide. J Sports Sci. 2014; 32:354–66. https://doi.org/10.1080/
02640414.2013.825729 PMID: 24015856
3.
Noakes TD, Myburgh KH, Schall R. Peak treadmill running velocity during the VO2 max test predicts
running performance. J Sports Sci. 1990; 8:35–45. https://doi.org/10.1080/02640419008732129 PMID:
2359150
4.
Cona G, Cavazzana A, Paoli A, Marcolin G, Grainer A, Bisiacchi PS. It’s a matter of mind! Cognitive
functioning predicts the athletic performance in ultra-marathon runners. PloS One. 2015; 10:e0132943.
https://doi.org/10.1371/journal.pone.0132943 PMID: 26172546
5.
Lambert MI, Dugas JP, Kirkman MC, Mokone GG, Waldeck MR. Changes in running speeds in a 100
km ultra-marathon race. J Sports Sci Med. 2004; 3:167–73. PMID: 24482594
6.
Tanda G, Knechtle B. Effects of training and anthropometric factors on marathon and 100 km ultramara-
thon race performance. Open Access J Sports Med. 2015; 6:129–36. https://doi.org/10.2147/OAJSM.
S80637 PMID: 25995653
7.
Knechtle B, Knechtle P, Rosemann T, Lepers R. Predictor variables for a 100-km race time in male
ultra-marathoners. Percept Mot skills. 2010; 111:681–93. https://doi.org/10.2466/05.25.PMS.111.6.
681-693 PMID: 21319608
8.
Knechtle B, Wirth A, Knechtle P, Rosemann T. Training volume and personal best time in marathon, not
anthropometric parameters, are associated with performance in male 100-km ultrarunners. J Strength
Cond Res. 2010; 24:604–9. https://doi.org/10.1519/JSC.0b013e3181c7b406 PMID: 20145568
9.
Nikolaidis PT, Onywera VO, Knechtle B. Running Performance, Nationality, Sex and Age in 10km, Half-
marathon, Marathon and 100km Ultra-marathon IAAF 1999–2015. J Strength Cond Res. 2017;
31:2189–207. https://doi.org/10.1519/JSC.0000000000001687 PMID: 28731980
10.
R Development Core Team. R: A language and environment for statistical computing. Vienna, Austria:
The R Foundation for Statistical Computing; 2011.
11.
Nikolaidis PT, Heller J, Knechtle B. The Russians are the fastest in marathon cross-country skiing: The
“Engadin Ski Marathon”. BioMed Res Int. 2017;9821757. https://doi.org/10.1155/2017/9821757 PMID:
28904979
12.
Bairner A. Assessing the sociology of sport: On national identity and nationalism. Int Rev Sociol Sport.
2015; 50:375–9.
13.
Houlihan B. Politics and sport. In: Coakley J, Dunning E, editors. Handbook of sports studies. London,
UK: Sage publications; 2006.
14.
Rupert JL. The search for genotypes that underlie human performance phenotypes. Comp Biochem
Physiol A Mol Integr Physiol. 2003; 136:191–203. PMID: 14527640
15.
Collins M, Xenophontos SL, Cariolou MA, Mokone GG, Hudson DE, Anastasiades L, et al. The ACE
gene and endurance performance during the South African Ironman Triathlons. Med Sci Sports Exerc.
2004; 36:1314–20. PMID: 15292738
16.
Ntoumanis N, Biddle SJH. Affect and achievement goals in physical activity: A meta-analysis. Scand J
Med Sci Sports. 1999; 9:315–32. PMID: 10606097
17.
Hallmann K, Breuer C, Dallmeyer S. Germany: running participation, motivation and images. In:
Scheerder J, Breedveld K, Borgers J, editors. Running across Europe: the rise and size of one of the
largest sport markets. New York, USA: Palgrave Macmillan; 2015.
18.
de Hon O, Kuipers H, van Bottenburg M. Prevalence of Doping Use in Elite Sports: A Review of Num-
bers and Methods. Sports Med. 2014; 45:57–69
The fastest 100-km ultra-marathoners worldwide
PLOS ONE | https://doi.org/10.1371/journal.pone.0199701
July 11, 2018
28 / 29
19.
Marc A, Sedeaud A, Schipman J, Jacquemin JA, Saulière G, Kryger KO, Toussaint JF. Geographic
enrolment of the top 100 in athletics running events from 1996 to 2012. J Sports Med Phys Fitness.
2017; 57(4):418–25. https://doi.org/10.23736/S0022-4707.16.06019-9 PMID: 26632850
20.
Onywera VO, Scott RA, Boit MK, Pitsiladis YP. Demographic characteristics of elite Kenyan endurance
runners. J Sports Sci. 2006; 24(4):415–22. https://doi.org/10.1080/02640410500189033 PMID:
16492605
21.
Scott RA, Georgiades E, Wilson RH, Goodwin WH, Wolde B, Pitsiladis YP. Demographic characteris-
tics of elite Ethiopian endurance runners. Med Sci Sports Exerc. 2003; 35(10):1727–32. https://doi.org/
10.1249/01.MSS.0000089335.85254.89 PMID: 14523311
22.
Marc A, Sedeaud A, Guillaume M, Rizk M, Schipman J, Antero-Jacquemin J, Haida A, Berthelot G,
Toussaint JF. Marathon progress: demography, morphology and environment. J Sports Sci. 2014; 32
(6):524–32. https://doi.org/10.1080/02640414.2013.835436 PMID: 24191965
23.
Irving R, Charlton V, Morrison E, Facey A, Buchanan O. Demographic characteristics of world class
Jamaican sprinters. ScientificWorldJournal. 2013 10; 2013:670217. https://doi.org/10.1155/2013/
670217 PMID: 24396303
24.
Scott RA, Pitsiladis YP. Genotypes and distance running: clues from Africa. Sports Med. 2007; 37(4–
5):424–7. PMID: 17465625
25.
Mooses M, Hackney AC. Anthropometrics and Body Composition in East African Runners: Potential
Impact on Performance. Int J Sports Physiol Perform. 2017; 12(4):422–30. https://doi.org/10.1123/
ijspp.2016-0408 PMID: 27631418
26.
Wilber RL, Pitsiladis YP. Kenyan and Ethiopian distance runners: what makes them so good? Int J
Sports Physiol Perform. 2012; 7(2):92–102. PMID: 22634972
27.
Onywera VO. East African runners: their genetics, lifestyle and athletic prowess. Med Sport Sci. 2009;
54:102–9. https://doi.org/10.1159/000235699 PMID: 19696510
The fastest 100-km ultra-marathoners worldwide
PLOS ONE | https://doi.org/10.1371/journal.pone.0199701
July 11, 2018
29 / 29
| Russians are the fastest 100-km ultra-marathoners in the world. | 07-11-2018 | Knechtle, Beat,Nikolaidis, Pantelis Theodoros,Valeri, Fabio | eng |
PMC9390899 | STUDY PROTOCOL
Effects of nutritional and hydration strategies
during ultramarathon events between
finishers and non-finishers: A systematic
review protocol
James W. NavaltaID1*, Victor D. Y. BeckID2, Taylor M. Diaz1, Vernice E. Ollano1
1 Department of Kinesiology and Nutrition Sciences, University of Nevada, Las Vegas, Las Vegas, NV,
United States of America, 2 Department of Physical Therapy, University of Nevada, Las Vegas, Las Vegas,
NV, United States of America
* [email protected]
Abstract
Ultramarathon running is a sport that is growing in popularity. Competing in an ultramara-
thon event is physiologically taxing on the human body, and it should not be surprising that
not all individuals who enroll for an event ultimately finish. While many factors can contribute
to this phenomenon, it is likely that nutritional and hydration strategies play a large role
between finishing and not finishing an ultramarathon. No published paper has systematically
reviewed the effects of nutritional and hydration strategies during ultramarathon events
between finishers and non-finishers. This paper details our intended protocol with the follow-
ing steps that create the flow of the systematic review: 1) Determine the review question
and Participant, Intervention, Comparator, Outcome, Study Design (PICOS) criteria; 2) Cre-
ate inclusion and exclusion criteria; 3) Create and follow a search strategy; 4) Document
sources that are included and excluded according to the pre-determined eligibility criteria; 5)
Assess final sources for risk of bias; 6) Extract pertinent data from final full-text articles and
synthesize the information; and 7) Disseminate findings of the systematic review.
Introduction
Ultramarathon running is a sport that is growing in popularity. The oldest ultramarathon
event, the Comrades Marathon, began in 1921 covering 89.9 km (55.9 mi), and 17 of the 34
participants did not finish (50% DNF) [1]. Since that time, over 300,000 people have com-
pleted the race, which currently caps yearly enrollment at 20,000 participants [2]. Participation
in ultramarathon events has risen exponentially since the year 2000, with over a million run-
ners participating (1,042,156) [3]. The sport involves running or walking a distance greater
than a traditional marathon (42.2 km, or 26.2 mi). The most popular (over three-quarters of a
million entrants) ultramarathon distance is 50 km (31.1 mi) as it represents a distance just
above the traditional marathon [3]. Participation decreases as the distance gets longer (just
over 100,000 entrants for 100km [62.1 mi], and approximately 40,000 for 24 h events) [3].
Competing in an ultramarathon event is physiologically taxing on the human body. Ultra-
marathon running has been associated with an increase in cardiac troponin T (a measure of
PLOS ONE
PLOS ONE | https://doi.org/10.1371/journal.pone.0272668
August 19, 2022
1 / 8
a1111111111
a1111111111
a1111111111
a1111111111
a1111111111
OPEN ACCESS
Citation: Navalta JW, Beck VDY, Diaz TM, Ollano
VE (2022) Effects of nutritional and hydration
strategies during ultramarathon events between
finishers and non-finishers: A systematic review
protocol. PLoS ONE 17(8): e0272668. https://doi.
org/10.1371/journal.pone.0272668
Editor: Samuel Penna Wanner, Universidade
Federal de Minas Gerais, BRAZIL
Received: April 8, 2022
Accepted: July 24, 2022
Published: August 19, 2022
Copyright: © 2022 Navalta et al. This is an open
access article distributed under the terms of the
Creative Commons Attribution License, which
permits unrestricted use, distribution, and
reproduction in any medium, provided the original
author and source are credited.
Data Availability Statement: No datasets were
generated or analysed during the current study. All
relevant data from this study will be made available
upon study completion.
Funding: The authors received no specific funding
for this work.
Competing interests: The authors have declared
that no competing interests exist.
myocardial damage) [4], malondialdehyde and creatine kinase (lipid peroxidation) [5], and
oxidative stress [6]. Other reported effects include measures associated with cardiac fatigue
such as increased atrial volume [7]. Given these effects it should not be surprising that not all
individuals who enroll for an ultramarathon event ultimately finish. It is estimated that
between 20–50% of individuals who begin an ultramarathon do not finish [8]. While many fac-
tors such as gastrointestinal distress and discomfort can contribute to this phenomenon, it is
likely that nutritional and hydration strategies play a large role.
Several systematic reviews have been conducted on various ultramarathon running topics
including psychology [9], limiting factors [10], long-term health problems [11], and patho-
physiology [12]. There has been a 50-year State of the Science offering [13], as well as a Posi-
tion Statement on Nutrition specific to training for a single-stage ultramarathon [14]. To our
knowledge, no published paper has reviewed the effects of nutritional and hydration strategies
during ultramarathon events between finishers and non-finishers. A “systematic review” was
selected as the methodology after reading published guidance on review types [15, 16]. The
primary aim of the proposed systematic review will be to find and describe the nutritional and
hydration strategies between single-stage ultramarathon finishers and non-finishers, and out-
comes reported in published literature. Based on the Preferred Reporting Items for Systematic
Reviews and Meta-Analyses (PRISMA) guidelines, the best practices for performing a system-
atic review include publishing the protocol independent of the review to ensure procedural
transparency [17]. As our systematic review of ultramarathon nutritional strategies will be, to
our knowledge, the first published on the topic, it is especially important to publish our proto-
col. This methods paper was written to achieve two objectives: 1) To adhere to the best prac-
tices stated in the PRISMA guidelines; and 2) To ensure procedural transparency.
Materials and methods
Developing the systematic review protocol began with members of the team performing scoping
searches in Google Scholar and PubMed. The scoping searches suggested that there were no
published systematic reviews of nutritional strategies between finishers and non-finishers of
ultramarathon events. The first author consulted peer-reviewed guidance [15, 16] about con-
ducting systematic reviews in health fields and has participated in the process previously [18].
The team deliberated and agreed upon the protocol presented in this article. The details of the
protocol are available via PROSPERO, an online international prospective register of systematic
reviews. The protocol was submitted on February 9, 2022 and registered on March 3, 2022
(PROSPERO ID: 42022308733). The protocol has the following steps that create the flow of the
systematic review: 1) Determine the review question and Participant, Intervention, Comparator,
Outcome, Study Design (PICOS) criteria; 2) Create inclusion and exclusion criteria; 3) Create
and follow a search strategy; 4) Document sources that are included and excluded according to
the pre-determined eligibility criteria; 5) Assess final sources for risk of bias; 6) Extract pertinent
data from final full-text articles and synthesize the information; and 7) Disseminate findings of
the systematic review. An optional part of systematic reviews that will be omitted is a meta-anal-
ysis. A meta-analysis will not be performed in this systematic review because the studies are
expected to include different ultramarathon distances, sample different populations, and mea-
sure different outcomes. Because of differences among these characteristics, the studies will not
be homogenous, which is necessary for a meta-analysis [19].
Step 1: Determine the review question and PICOS criteria
The first step is to determine the review question and PICOS criteria (Table 1). Because multi-
stage ultramarathon events stress the body from a nutritional and hydrational standpoint in a
PLOS ONE
A systematic review protocol: Nutrition and hydration strategies in ultramarathon finishers and non-finishers
PLOS ONE | https://doi.org/10.1371/journal.pone.0272668
August 19, 2022
2 / 8
much different manner compared to single-stage events, we decided to focus on single-stage
events not lasting longer than 24 h. The event can be performed in any location (including
indoors on a treadmill). Studies with adults who enrolled in a single-stage ultramarathon
between the ages of 18–60 will be relevant. Controlled, uncontrolled, randomized, nonrando-
mized, and observational studies will be considered. Information about participants’ charac-
teristics, ultrarunning experience, and finisher or non-finisher status will be collected.
Step 2: Create inclusion and exclusion criteria
The second step is to create inclusion and exclusion criteria, or eligibility criteria. The review
question and PICOS criteria form the basis of this systematic review’s eligibility criteria, which
aim to be both explicit and succinct. These characteristics provide clarity to the review team
members, enabling the exclusion of irrelevant sources during the screening process. Expedi-
tious screening is important to the process because teams may be required to screen hundreds
to thousands of sources. In consultation of the review question and PICOS criteria, the eligibil-
ity criteria were created (Table 2). The inclusion criteria will allow for the inclusion of unpub-
lished master’s theses and doctoral dissertations. This decision was made to capture as much
data about nutritional and hydrational strategies during a single-stage ultramarathon event as
possible.
Step 3: Create and follow a search strategy
After determining eligibility criteria, the review team will proceed to the third step by creating
and following a search strategy (Table 3). Consistent with our PICOS and eligibility criteria,
we will search in four databases for all relevant studies, regardless of publication year. The
same search combination string will be used in each database (Table 3). The search combina-
tion was initially created by the full review team. An iterative process of revision ensued until
the search combination returned a manageable number of hits in each database (about 100–
1,000 hits).
Table 1. Review question and PICOS table.
Review
Question
Do finishers of single-stage ultramarathon events employ different nutritional and hydration
strategies than individuals who do not finish?
Population
Ultramarathoners of any sex who enrolled in a single-stage ultramarathon event, between the
ages of 18–60 years old
Intervention
Single-stage ultramarathon event not longer than 24 h
• The ultramarathon must be performed continuously within 24 h and excludes multi-stage
events
Comparator
Completion of the single-stage ultramarathon event
Non completion of the single-stage ultramarathon event (considered as ‘did not finish’ or ‘DNF’)
Outcomes
Nutrition and hydration strategies employed during the event
• Macromolecules (carbohydrates, fats, proteins)
• Hydration (fluid type and pacing)
• Other supplements (vitamins and minerals)
Ultramarathon event characteristics: distance of race, start time, elevation profile, environmental
profile (temperature, humidity, windspeed)
Participants’ age, sex, body mass, height, years of ultrarunning experience, finisher or non-
finisher
Setting
Any physical environment (indoors, outdoors, urban, rural, built-up, or natural)
Study Design
Studies with interventions, as well as observational studies
• Controlled or uncontrolled
• Randomized or nonrandomized
https://doi.org/10.1371/journal.pone.0272668.t001
PLOS ONE
A systematic review protocol: Nutrition and hydration strategies in ultramarathon finishers and non-finishers
PLOS ONE | https://doi.org/10.1371/journal.pone.0272668
August 19, 2022
3 / 8
With the operational search combination established, each member volunteered for roles.
Two members formed Team A, and two members formed Team B. The two members of Team
A will search in Google Scholar and SPORTDiscus. The two members of Team B will search in
PubMed and Web of Science. This assignment of databases will divide the workload of the
search somewhat equally between the teams. The search, screening process, and inclusion pro-
cess represents the “search flow.” The search flow will channel an initially broad collection of
sources into increasingly smaller collections (Fig 1). The search flow is described in Step 4.
Table 2. Eligibility criteria.
Participants
Adults between the age of 18–60 years old, and any sex, gender, or nationality who has enrolled
in a single-stage ultramarathon not lasting longer than 24 h
Inclusion
Criteria
1. The source is a published article in a peer-reviewed journal or is an unpublished or findable
master’s thesis or doctoral dissertation
2. The source is written in English
3. The source reports the findings of an interventional or observational study
a. The intervention is any nutritional or hydrational supplement
b. At least one reported outcome is finishing or not finishing an ultramarathon event
The observation is nutritional, or hydration strategy utilized while performing an
ultramarathon event
Exclusion
Criteria
1. The source is not a published, peer-reviewed journal article or a findable and available
master’s thesis or doctoral dissertation
2. The source is written in any language other than English
3. The source reports the findings of an interventional study with an intervention or outcomes
irrelevant to this systematic review
a. The intervention is an ultramarathon without nutritional or hydration component
b. None of the reported outcomes are finishing or not finishing the ultramarathon event
https://doi.org/10.1371/journal.pone.0272668.t002
Table 3. Search strategy.
Investigators
Team A: TD and JN
Team B: VB and VO
Techniques
Search research databases for sources, including them in four stages:
1. Include sources by title
2. Include sources by abstract
3. Include sources by full text
4. Include sources from the reference lists of sources included by full text (journal articles, master’s
theses, and doctoral dissertations)
Databases
Google Scholar, PubMed, and SPORTDiscus, Web of Science
Included Types of Literature
Published, peer-reviewed journal articles; unpublished and published master’s theses and doctoral
dissertations
Publication Date Range
No limit
Intervention Search Terms
Outcome Search Terms
“Ultramarathon”
“24h ultramarathon”
“Ultra endurance”
“24h race”
“Finish”
“Completion”
“Complete”
“DNF”
“Dropout”
“Nutrition”
“Carbohydrate”
“Fats”
“Protein”
“Vitamins”
“Minerals”
“Hydration”
“Electrolytes”
“Water”
“Fluid”
“Supplements”
“Supplementation”
Search Combination
((ultramarathon) OR (“24h ultramarathon”) OR (“ultra endurance”) OR (“24h race”)) AND ((finish)
OR (completion) OR (complete) OR (DNF) OR (“drop out”)) AND ((nutrition) OR (carbohydrate)
OR (fats) OR (protein) OR (vitamins) OR (minerals) OR (hydration) OR (electrolytes) OR (water)
OR (fluid) OR (supplements) OR (supplementation))
https://doi.org/10.1371/journal.pone.0272668.t003
PLOS ONE
A systematic review protocol: Nutrition and hydration strategies in ultramarathon finishers and non-finishers
PLOS ONE | https://doi.org/10.1371/journal.pone.0272668
August 19, 2022
4 / 8
Step 4: Document sources that are included and excluded
according to the pre-determined eligibility criteria
The fourth step is the application of the systematic review process, the search flow (Fig 2). The
search flow is modeled on the 2020 PRISMA statement [17], containing four steps: 1) Identify
relevant sources by title, 2) Screen sources by abstract, 3) Assess and include sources by full
text, and 4) Include eligible sources from the references of full texts included in the third step.
During the search flow, it will be critical to document sources’ inclusion and exclusion clearly
[16, 19]. Clear documentation allows the systematic review to be transparent and reproducible.
Reproducibility is a hallmark of a systematic review that sets it apart from traditional literature
reviews [19].
Fig 1. The search flow funnels sources into smaller collections until the final articles are included.
https://doi.org/10.1371/journal.pone.0272668.g001
PLOS ONE
A systematic review protocol: Nutrition and hydration strategies in ultramarathon finishers and non-finishers
PLOS ONE | https://doi.org/10.1371/journal.pone.0272668
August 19, 2022
5 / 8
Eligibility criteria will be determined by four individuals (two independent teams of two—
Team A and Team B) who will work on selecting the studies. Individuals on the same team
will be blinded to the other’s decisions with screening completed independently. The alternate
team will resolve any disagreements.
To make the search flow reproducible, a specific tool in Google Sheets will be utilized [20].
The tool is a spreadsheet for Teams A and B to coordinate with each other and has four sepa-
rate sheets (one for each team member). During the first step of the search flow, members of
Team A and B will enter three types of values into the sheet: 1) the number of hits each data-
base returns, 2) the number of sources deemed relevant by title, and 3) the number of duplicate
sources identified across the databases (identical sources found in the other database). The
sheet will automatically sum these values. The sheet represents a precise record of members’
progression through the first step of the search flow. The sheet also helps members record
Steps 2–4.
Step 5: Assess final sources for risk of bias
The fifth step acknowledges that it is important for all systematic reviews to assess the included
sources’ risk of bias [16, 19]. Being transparent about the risk of bias allows readers to draw
conclusions concerning the quality and strength of evidence for interventions affecting the
outcome [16]. This systematic review will assess risk of bias at the study-level using tools spe-
cific to the study design. Randomized, parallel trials will be assessed by using the revised
Cochrane risk of bias tool for randomized trials (RoB 2) [21]. Non-randomized trials will be
assessed by using the risk of bias in non-randomized studies—of interventions (ROBINS-I)
tool [22]. The following characteristics will be assessed: deviations from the intended interven-
tion, missing outcome data, measurement of the outcome, selection of the reported result. Bias
will be evaluated by two independent teams of two individuals. The alternating team will
check the others work and settle any disagreements of individual judgements.
Fig 2. Search flow.
https://doi.org/10.1371/journal.pone.0272668.g002
PLOS ONE
A systematic review protocol: Nutrition and hydration strategies in ultramarathon finishers and non-finishers
PLOS ONE | https://doi.org/10.1371/journal.pone.0272668
August 19, 2022
6 / 8
Step 6: Extract pertinent data from final full-text articles and
synthesize the information
The sixth step is to extract pertinent data from the included full-text articles and write a narra-
tive synthesis. The following data will be extracted from study documents: participant charac-
teristics of age, body mass, height, years of ultrarunning experience, finisher or non-finisher.
Ultramarathon characteristics: distance of race, start time, elevation profile, environmental
profile (temperature, humidity, windspeed, start elevation). Nutritional supplementation dur-
ing the event: macromolecules (carbohydrates, fats, proteins), hydration (fluid type and pac-
ing), and other supplements (vitamins and minerals). We will contact investigators for
information that is not provided in the published document. We will report the average differ-
ences between finishers and non-finishers for macromolecules (carbohydrates, fats, proteins),
hydration (fluid type and pace of ingestion), and other supplements (vitamins and minerals).
Once all data are extracted, a narrative synthesis of the data will be written to report the main
findings and implications of the systematic review.
Step 7: Disseminate the findings of the systematic review
The final step is to disseminate the main findings and implications. The intention is to com-
plete the systematic review by August 2022 and submit the narrative synthesis for publication
in a peer-reviewed academic journal thereafter.
Final remarks
To our knowledge, the proposed systematic review will be the first to describe the effects of
nutrition and hydration between finishers and non-finishers of ultramarathon events as
described by the scientific literature. Because of this, the review will fill an important gap in the
literature. The protocol is presented here as a best practice [17] and to earn readers’ trust in the
review protocol. Additionally, the procedures described here can provide others a framework
to conduct their own systematic reviews if so desired.
Supporting information
S1 Checklist. PRISMA-P (Preferred Reporting Items for Systematic review and Meta-
Analysis Protocols) 2015 checklist: Recommended items to address in a systematic review
protocol.
(DOC)
Author Contributions
Conceptualization: James W. Navalta, Victor D. Y. Beck, Taylor M. Diaz, Vernice E. Ollano.
Writing – original draft: James W. Navalta, Victor D. Y. Beck, Taylor M. Diaz, Vernice E.
Ollano.
Writing – review & editing: James W. Navalta, Victor D. Y. Beck, Taylor M. Diaz, Vernice E.
Ollano.
References
1.
Comrades Marathon History: Comrades Marathon; Available from: https://www.web.comrades.com/
history/.
2.
Comrades Marathon; Available from: https://www.comrades.com/.
PLOS ONE
A systematic review protocol: Nutrition and hydration strategies in ultramarathon finishers and non-finishers
PLOS ONE | https://doi.org/10.1371/journal.pone.0272668
August 19, 2022
7 / 8
3.
Scheer V. Participation Trends of Ultra Endurance Athletes. Sports Medicine and Arthroscopy Review.
2019; 27(1):3–7. https://doi.org/10.1097/JSA.0000000000000198 PMID: 30601393
4.
Laslett L, Eisenbud E, Lind R. Evidence of myocardial injury during prolonged strenuous exercise. Am J
Cardiol. 1996; 78(4):488–90. https://doi.org/10.1016/0002-9149(97)00003-9 PMID: 8801807
5.
Kanter MM, Lesmes GR, Kaminsky LA, La Ham-Saeger J, Nequin ND. Serum creatine kinase and lac-
tate dehydrogenase changes following an eighty kilometer race. Relationship to lipid peroxidation. Eur J
Appl Physiol Occup Physiol. 1988; 57(1):60–3. https://doi.org/10.1007/BF00691239 PMID: 3342795
6.
Knez WL, Coombes JS, Jenkins DG. Ultra-endurance exercise and oxidative damage: implications for
cardiovascular health. Sports Med. 2006; 36(5):429–41. https://doi.org/10.2165/00007256-200636050-
00005 PMID: 16646630
7.
Passaglia DG, Emed LG, Barberato SH, Guerios ST, Moser AI, Silva MM, et al. Acute effects of pro-
longed physical exercise: evaluation after a twenty-four-hour ultramarathon. Arq Bras Cardiol. 2013;
100(1):21–8. https://doi.org/10.1590/s0066-782x2012005000118 PMID: 23250832
8.
Gorichanaz T. Did Not Finish’: A Phenomenology of Failure. Sport, Ethics and Philosophy. 2021; 15
(1):27–42.
9.
Roebuck GS, Fitzgerald PB, Urquhart DM, Ng SK, Cicuttini FM, Fitzgibbon BM. The psychology of
ultra-marathon runners: A systematic review. Psychol Sport Exerc. 2018; 37:43–58.
10.
Garbisu-Hualde A, Santos-Concejero J. What are the Limiting Factors During an Ultra-Marathon? A
Systematic Review of the Scientific Literature. J Hum Kinet. 2020; 72:129–39. https://doi.org/10.2478/
hukin-2019-0102 PMID: 32269654
11.
Scheer V, Tiller NB, Doutreleau S, Khodaee M, Knechtle B, Pasternak A, et al. Potential Long-Term
Health Problems Associated with Ultra-Endurance Running: A Narrative Review. Sports Med. 2022; 52
(4):725–40. https://doi.org/10.1007/s40279-021-01561-3 PMID: 34542868
12.
Knechtle B, Nikolaidis PT. Physiology and Pathophysiology in Ultra-Marathon Running. Front Physiol.
2018; 9:634. https://doi.org/10.3389/fphys.2018.00634 PMID: 29910741
13.
Hoffman MD. State of the Science on Ultramarathon Running After a Half Century: A Systematic Analy-
sis and Commentary. Int J Sports Physiol Perform. 2020:1–5. https://doi.org/10.1123/ijspp.2020-0151
PMID: 32580165
14.
Tiller NB, Roberts JD, Beasley L, Chapman S, Pinto JM, Smith L, et al. International Society of Sports
Nutrition Position Stand: nutritional considerations for single-stage ultra-marathon training and racing. J
Int Soc Sports Nutr. 2019; 16(1):50. https://doi.org/10.1186/s12970-019-0312-9 PMID: 31699159
15.
Sutton A, Clowes M, Preston L, Booth A. Meeting the review family: exploring review types and associ-
ated information retrieval requirements. Health Info Libr J. 2019; 36(3):202–22. https://doi.org/10.1111/
hir.12276 PMID: 31541534
16.
Siddaway AP, Wood AM, Hedges LV. How to Do a Systematic Review: A Best Practice Guide for Con-
ducting and Reporting Narrative Reviews, Meta-Analyses, and Meta-Syntheses. Annu Rev Psychol.
2019; 70:747–70. https://doi.org/10.1146/annurev-psych-010418-102803 PMID: 30089228
17.
Page MJ, McKenzie JE, Bossuyt PM, Boutron I, Hoffmann TC, Mulrow CD, et al. The PRISMA 2020
statement: An updated guideline for reporting systematic reviews. PLoS Med. 2021; 18(3):e1003583.
https://doi.org/10.1371/journal.pmed.1003583 PMID: 33780438
18.
Carrier B, Barrios B, Jolley BD, Navalta JW. Validity and Reliability of Physiological Data in Applied Set-
tings Measured by Wearable Technology: A Rapid Systematic Review. Technologies. 2020; 8(4).
19.
Boland A, Cherry G, Dickson R. Doing a systematic review: A student’s guide. 2nd ed. ed. London,
England: SAGE Publications Ltd; 2017.
20.
Davis DW, Carrier B, Barrios B, Cruz K, Navalta JW. A protocol and novel tool for systematically review-
ing the effects of mindful walking on mental and cardiovascular health. PLoS One. 2021; 16(10):
e0258424. https://doi.org/10.1371/journal.pone.0258424 PMID: 34637455
21.
Sterne JAC, Savovic J, Page MJ, Elbers RG, Blencowe NS, Boutron I, et al. RoB 2: a revised tool for
assessing risk of bias in randomised trials. BMJ. 2019; 366:l4898. https://doi.org/10.1136/bmj.l4898
PMID: 31462531
22.
Sterne JA, Hernan MA, Reeves BC, Savovic J, Berkman ND, Viswanathan M, et al. ROBINS-I: a tool
for assessing risk of bias in non-randomised studies of interventions. BMJ. 2016; 355:i4919. https://doi.
org/10.1136/bmj.i4919 PMID: 27733354
PLOS ONE
A systematic review protocol: Nutrition and hydration strategies in ultramarathon finishers and non-finishers
PLOS ONE | https://doi.org/10.1371/journal.pone.0272668
August 19, 2022
8 / 8
| Effects of nutritional and hydration strategies during ultramarathon events between finishers and non-finishers: A systematic review protocol. | 08-19-2022 | Navalta, James W,Beck, Victor D Y,Diaz, Taylor M,Ollano, Vernice E | eng |
PMC10651037 | PLOS ONE
Dose response of running on blood biomarkers of wellness in the generally healthy
--Manuscript Draft--
Manuscript Number:
PONE-D-23-25168
Article Type:
Research Article
Full Title:
Dose response of running on blood biomarkers of wellness in the generally healthy
Short Title:
Biomarker signature of runners
Corresponding Author:
Bartosz Nogal
InsideTracker
Cambridge, MA UNITED STATES
Keywords:
physical activity, exercise, blood biomarkers, running, generally healthy, mendelian
randomization
Abstract:
Exercise is effective toward delaying or preventing chronic disease, with a large body
of evidence supporting its effectiveness. However, less is known about the specific
healthspan-promoting effects of exercise on blood biomarkers in the disease-free
population. In this work, we examine 23,237 generally healthy individuals who self-
report varying weekly running volumes and compare them to 4,428 generally healthy
sedentary individuals, as well as 82 professional endurance athletes. We estimate the
significance of differences among blood biomarkers for groups of increasing running
levels using analysis of variance (ANOVA), adjusting for age, gender, and BMI. We
attempt and add insight to our observational dataset analysis via two-sample
Mendelian randomization (2S-MR) using large independent datasets. We find that
self-reported running volume associates with biomarker signatures of improved
wellness, with some serum markers apparently being principally modified by BMI,
whereas others show a dose-effect with respect to running volume. We further detect
hints of sexually dimorphic serum responses in oxygen transport and hormonal traits,
and we also observe a tendency toward pronounced modifications in magnesium
status in professional endurance athletes. Thus, our results further characterize blood
biomarkers of exercise and metabolic health, particularly regarding dose-effect
relationships, and better inform personalized advice for training and performance.
Order of Authors:
Bartosz Nogal
Svetlana Vinogradova
Gil Blander
Milena Jorge
Paul Fabian
Ali Torkamani
Additional Information:
Question
Response
Financial Disclosure
Enter a financial disclosure statement that
describes the sources of funding for the
work included in this submission. Review
the submission guidelines for detailed
requirements. View published research
articles from PLOS ONE for specific
examples.
This statement is required for submission
and will appear in the published article if
InsideTracker was the sole funding source.
Powered by Editorial Manager® and ProduXion Manager® from Aries Systems Corporation
the submission is accepted. Please make
sure it is accurate.
Unfunded studies
Enter: The author(s) received no specific
funding for this work.
Funded studies
Enter a statement with the following details:
Initials of the authors who received each
award
•
Grant numbers awarded to each author
•
The full name of each funder
•
URL of each funder website
•
Did the sponsors or funders play any role in
the study design, data collection and
analysis, decision to publish, or preparation
of the manuscript?
•
NO - Include this sentence at the end of
your statement: The funders had no role in
study design, data collection and analysis,
decision to publish, or preparation of the
manuscript.
•
YES - Specify the role(s) played.
•
* typeset
Competing Interests
Use the instructions below to enter a
competing interest statement for this
submission. On behalf of all authors,
disclose any competing interests that
could be perceived to bias this
work—acknowledging all financial support
and any other relevant financial or non-
financial competing interests.
This statement is required for submission
and will appear in the published article if
the submission is accepted. Please make
sure it is accurate and that any funding
sources listed in your Funding Information
later in the submission form are also
declared in your Financial Disclosure
statement.
View published research articles from
PLOS ONE for specific examples.
B.N., S.V., P.F., and G.B. are employees of InsideTracker.
Powered by Editorial Manager® and ProduXion Manager® from Aries Systems Corporation
NO authors have competing interests
Enter: The authors have declared that no
competing interests exist.
Authors with competing interests
Enter competing interest details beginning
with this statement:
I have read the journal's policy and the
authors of this manuscript have the following
competing interests: [insert competing
interests here]
* typeset
Ethics Statement
Enter an ethics statement for this
submission. This statement is required if
the study involved:
Human participants
•
Human specimens or tissue
•
Vertebrate animals or cephalopods
•
Vertebrate embryos or tissues
•
Field research
•
Write "N/A" if the submission does not
require an ethics statement.
General guidance is provided below.
Consult the submission guidelines for
detailed instructions. Make sure that all
information entered here is included in the
Methods section of the manuscript.
BRANY IRB File # 22-12-501-1095
BRANY IRB has determined this research is
exempt from IRB review under category(ies) # (4)(ii), as detailed in 45 CFR 46.104(d)
and BRANY’s
Standard Operating Procedures (category excerpted below).
(4) Secondary research for which consent is not required: Secondary research uses of
identifiable
private information or identifiable biospecimens, with the following criterion met:
(ii) Information, which may include information about biospecimens, is recorded by the
investigator in such a manner that the identity of the human subjects cannot readily be
ascertained directly or through identifiers linked to the subjects, the investigator does
not
contact the subjects, and the investigator will not re-identify subjects
Powered by Editorial Manager® and ProduXion Manager® from Aries Systems Corporation
Format for specific study types
Human Subject Research (involving human
participants and/or tissue)
Give the name of the institutional review
board or ethics committee that approved the
study
•
Include the approval number and/or a
statement indicating approval of this
research
•
Indicate the form of consent obtained
(written/oral) or the reason that consent was
not obtained (e.g. the data were analyzed
anonymously)
•
Animal Research (involving vertebrate
animals, embryos or tissues)
Provide the name of the Institutional Animal
Care and Use Committee (IACUC) or other
relevant ethics board that reviewed the
study protocol, and indicate whether they
approved this research or granted a formal
waiver of ethical approval
•
Include an approval number if one was
obtained
•
If the study involved non-human primates,
add additional details about animal welfare
and steps taken to ameliorate suffering
•
If anesthesia, euthanasia, or any kind of
animal sacrifice is part of the study, include
briefly which substances and/or methods
were applied
•
Field Research
Include the following details if this study
involves the collection of plant, animal, or
other materials from a natural setting:
Field permit number
•
Name of the institution or relevant body that
granted permission
•
Data Availability
Authors are required to make all data
underlying the findings described fully
available, without restriction, and from the
time of publication. PLOS allows rare
exceptions to address legal and ethical
concerns. See the PLOS Data Policy and
FAQ for detailed information.
No - some restrictions will apply
Powered by Editorial Manager® and ProduXion Manager® from Aries Systems Corporation
A Data Availability Statement describing
where the data can be found is required at
submission. Your answers to this question
constitute the Data Availability Statement
and will be published in the article, if
accepted.
Important: Stating ‘data available on request
from the author’ is not sufficient. If your data
are only available upon request, select ‘No’ for
the first question and explain your exceptional
situation in the text box.
Do the authors confirm that all data
underlying the findings described in their
manuscript are fully available without
restriction?
Describe where the data may be found in
full sentences. If you are copying our
sample text, replace any instances of XXX
with the appropriate details.
If the data are held or will be held in a
public repository, include URLs,
accession numbers or DOIs. If this
information will only be available after
acceptance, indicate this by ticking the
box below. For example: All XXX files
are available from the XXX database
(accession number(s) XXX, XXX.).
•
If the data are all contained within the
manuscript and/or Supporting
Information files, enter the following:
All relevant data are within the
manuscript and its Supporting
Information files.
•
If neither of these applies but you are
able to provide details of access
elsewhere, with or without limitations,
please do so. For example:
Data cannot be shared publicly because
of [XXX]. Data are available from the
XXX Institutional Data Access / Ethics
Committee (contact via XXX) for
researchers who meet the criteria for
access to confidential data.
The data underlying the results
presented in the study are available
from (include the name of the third party
•
All relevant data are within the manuscript and its Supporting Information files.
Powered by Editorial Manager® and ProduXion Manager® from Aries Systems Corporation
and contact information or URL).
This text is appropriate if the data are
owned by a third party and authors do
not have permission to share the data.
•
* typeset
Additional data availability information:
Powered by Editorial Manager® and ProduXion Manager® from Aries Systems Corporation
B Nogal
1
1
Dose response of running on blood biomarkers of wellness in the
generally healthy
Bartek Nogal PhD1ナ, Svetlana Vinogradova PhD1ナ, Milena Jorge MD,PhD1, Ali Torkamani
PhD 2,3, Paul Fabian BSc1, and Gil Blander PhD1*
1InsideTracker, Cambridge, Massachusetts, United States of America.
2The Scripps Translational Science Institute, The Scripps Research Institute, La Jolla, CA, USA.
3Department of Integrative Structural and Computational Biology, The Scripps Research
Institute, La Jolla, CA, USA.
ナEqual Contribution: These authors contributed equally
* Correspondence and reprint requests: Gil Blander, [email protected]
Bartek Nogal
[email protected]
Svetlana Vinogradova
[email protected]
Ali Torkamani
[email protected]
Paul Fabian
[email protected]
Running title: Biomarker signature of runners
Main Text
Click here to
access/download;Manuscript;Runners_PONE_v1.docx
B Nogal 2
Abstract
Exercise is effective toward delaying or preventing chronic disease, with a large body of evidence
supporting its effectiveness. However, less is known about the specific healthspan-promoting
effects of exercise on blood biomarkers in the disease-free population. In this work, we examine
23,237 generally healthy individuals who self-report varying weekly running volumes and
compare them to 4,428 generally healthy sedentary individuals, as well as 82 professional
endurance athletes. We estimate the significance of differences among blood biomarkers for
groups of increasing running levels using analysis of variance (ANOVA), adjusting for age,
gender, and BMI. We attempt and add insight to our observational dataset analysis via two-sample
Mendelian randomization (2S-MR) using large independent datasets. We find that self-reported
running volume associates with biomarker signatures of improved wellness, with some serum
markers apparently being principally modified by BMI, whereas others show a dose-effect with
respect to running volume. We further detect hints of sexually dimorphic serum responses in
oxygen transport and hormonal traits, and we also observe a tendency toward pronounced
modifications in magnesium status in professional endurance athletes. Thus, our results further
characterize blood biomarkers of exercise and metabolic health, particularly regarding dose-effect
relationships, and better inform personalized advice for training and performance.
B Nogal
3
3
Keywords: physical activity, exercise, blood biomarkers, running, generally healthy, mendelian
randomization
B Nogal 4
1
Introduction
Physical inactivity is one of the leading modifiable behavioral causes of death in the US 1.
Worldwide, physical inactivity is estimated to account for about 8.3% of premature mortality, an
effect size that is on the same order as smoking and obesity 2. At the same time, the potent health
benefits of exercise have been proven time and time again, with results so consistent across a wide
variety of chronic diseases that some posit it can be considered a medical intervention 3-5.
However, since most investigators report the effects of exercise in either diseased populations or
athletes 6,7, there exists a significant gap in knowledge as to the measurable effects of exercise in
the generally healthy population who exercise for the purpose of improving their healthspan, which
can be projected via established measures such as blood biomarkers 8-11.
It is well established that routine laboratory biomarkers are validated proxies of the state of an
individual’s overall metabolic health and other healthpan-related parameters 12. A large body of
evidence supports the effectiveness of exercise in modifying blood biomarkers toward disease
mitigation in clinical cohorts as well as athletes, where the effect sizes may be larger 6,13. Indeed,
it’s been shown that more favorable changes in response to exercise training occur usually in those
with more pronounced dyslipidemia 13. In professional athletes, the sheer volume and/or intensity
of physical activity may drive large effects in various hematological, lipid, immune, and endocrine
variables 6. Our aim is to help fill the gap in understanding of the effects of exercise on blood
biomarkers in the generally healthy, free-living population. Toward this end, we endeavored to
explore the effects of vigorous exercise such as running in apparently healthy, mostly non-athletic
cohort to better understand the landscape of blood biomarker modifications expected in the
individual who partakes in recreational physical activity for the purpose of maintaining good
health.
B Nogal
5
5
For this purpose, we leveraged the InsideTracker dataset that includes information on self-reported
exercise habits combined with blood biomarker and genomics data. We have previously reported
on the results of a longitudinal analysis on blood biomarker data from 1032 generally healthy
individuals who used our automated, web-based personalized nutrition and lifestyle platform 14.
For the purpose of this investigation, we focused on running as the exercise of choice as it is one
of the most common (purposeful) physical activity modalities practiced globally by generally
healthy individuals and would thus be relevant. Moreover, since this was a cross-sectional study
based on self-reported exercise habits, we attempted to increase our capacity to infer intervention
effects, as well as tease out potential confounders, by performing 2S-MR in large independent
cohorts.
2
Methods
2.1
Dataset
We conducted an observational analysis of data from InsideTracker users. InsideTracker is a
direct-to-consumer (DTC) company established in 2009 that markets and sells InsideTracker
(insidetracker.com), a personalized lifestyle recommendation platform. The platform provides
serum biomarker and genomics testing, and performs integrative analysis of these datasets,
combined with activity/sleep tracker data toward biomarker and healthspan optimization (of note,
at the time of this analysis, we did not have sufficient users with activity/sleep tracker data to
include this data stream in the current study). New users were continuously added to the
InsideTracker database from January 2011 to March 2022.
B Nogal 6
2.2
Recruitment of participants
Recruitment of participants aged between 18 and 65 and residing in North America was conducted
through company marketing and outreach. Participants were subscribing members to the
InsideTracker platform and provided informed consent to have their blood test data and self-
reported information used in an anonymized fashion for research purposes. Research was
conducted according to guidelines for observational research in tissue samples from human
subjects. Eligible participants completed a questionnaire that included age, ethnicity, sex, dietary
preferences, physical activity, and exposure to sunlight. This study employed data from 23,237
participants that met our analysis inclusion requirements, namely absence of any chronic disease
as determined by questionnaire and metabolic blood biomarkers within normal clinical reference
ranges. The platform is not a medical service and does not diagnose or treat medical conditions,
so medical history and medication use were not collected. The Institutional Review Board (IRB)
determine this work was not subject to a review based on category 4 exemption (“secondary
research” with de-identified subjects).
2.3
Biomarker collection and analysis
Blood samples were collected and analyzed by Clinical Laboratory Improvement Amendments
(CLIA)–approved, third-party clinical labs (primarily Quest Diagnostics and LabCorp).
Participants were instructed to fast for 12 hours prior to the phlebotomy, with the exception of
water consumption. Results from the blood analysis were then uploaded to the platform via
electronic integration with the CLIA-approved lab. Participants chose a specific blood panel from
7 possible offerings, each comprising some subset of the biomarkers available. Due to the variation
in blood panels offered, the participant sample size per biomarker is not uniform.
B Nogal
7
7
2.3
Biomarker dataset preparation
In our raw dataset, occasional outlier values were observed that were deemed implausible (e.g.
fasting glucose < 65 mg/dL). To remove anomalous outliers in a systematic way, we used the
Interquartile Range (IQR) method of identifying outliers, removing data points which fell below
Q1 – 1.5 IQR or above Q3 + 1.5 IQR.
2.4
Calculation of polygenic scores
The variants (SNPs) comprising the polygenic risk scores were derived from publicly available
GWAS summary statistics (https://www.ebi.ac.uk/gwas/). Scores were calculated across users by
summing the product of effect allele doses weighted by the beta coefficient for each SNP, as
reported in the GWAS summary statistics. Variant p-value thresholds were generally chosen based
on optimization of respective PGS-blood biomarker correlation in the entire InsideTracker cohort
with both blood and genomics datasets (~1000-1500 depending on the blood biomarker at the time
of analysis). Genotyping data was derived from a combination of a custom InsideTracker array
and third party arrays such as 23andMe and Ancestry. Not all variants for any particular PGS were
genotyped on every array; proxies for missing SNPs were extracted via the “LDlinkR” package
using the Utah Residents (CEPH) with Northern and Western European ancestry (CEU) population
(R2 > 0.8 cut-off). Only results PGSs for which there was sufficient biomarker-genotyping dataset
overlap were reported (note that none of the blood biomarker PGSs met this requirement).
2.5
Blood biomarker analysis with respect to running volume and polygenic scores
To estimate significance of differences for blood biomarkers levels among exercise groups, we
performed 3-way analysis of variance (ANOVA) analysis adjusting for age, gender, and BMI
B Nogal 8
(type-II analysis-of-variance tables function ANOVA from ‘car’ R package, version 3.0-12).
When estimating the effort of reported training volume on biomarkers, we assigned numerical
values corresponding to 4 levels of running and performed ANOVA analysis with those levels
treating it as an independent variable. P-values were adjusted using the Benjamini & Hochberg
method 15. P-values for interaction plots were calculated with ANOVA including interaction
between exercise group and polygenic scores category. When comparing runners (PRO and
HVAM combined) versus sedentary individuals, we used propensity score matching method to
account for existing covariates (age and gender): we identified 745 sedentary individuals with
similar to runners’ age distributions among both males and females. We used ‘MatchIt’ R package
(version 4.3.3) implementing nearest neighbor method for matching 16.
2.6
Mendelian randomization
We attempted to add insight around the causality of exercise vs. BMI differences with respect to
serum marker improvement by performing MR analyses on a subset of biomarker observations
where BMI featured as a strong covariate and was thus used as the IV in the 2S-MR. Thus, our
hypothesis here was that BMI differences were the primary (causal) driver behind the improvement
behind some biomarkers. MR uses genetic variants as modifiable exposure (risk factor) proxies
to evaluate causal relationships in observational data while reducing the effects of confounders
and reverse causation (Figure 1S). These SNPs are used as instrumental variables and must meet
3 basic assumptions: (1) they must be robustly associated with the exposure; (2) they must exert
their effect on outcome via the exposure, and (3) there must be no unmeasured confounders of the
associations between the genetic variants and outcome (e.g. horizontal pleiotropy) 17. Importantly,
SNPs are proper randomization instruments because they are determined at birth and thus serve as
proxies of long-term exposures and cannot, in general, be modified by the environment. If the 3
B Nogal
9
9
above mentioned assumptions hold, MR-estimate effects of exposure on outcomes are not likely
to be significantly affected by reverse causation or confounding. In the 2S-MR performed here,
where GWAS summary statistics are used for both exposure and outcome from independent
cohorts, reverse causation and horizontal pleiotropy can readily be assessed, and weak instrument
bias and the likelihood of false positive findings are minimized as a result of the much larger
samples sizes 17. Indeed, the bias in the 2S-MR using non-overlapping datasets as performed here
is towards the null 17. Furthermore, to maintain the SNP-exposure associations and linkage
disequilibrium (LD) patterns in the non-overlapping populations we used GWAS datasets from
the MR-Base platform that were derived from ancestrally similar populations (“ukb”: analysis of
UK Biobank phenotypes, and “ieu”: GWAS summary datasets generated by many different
European consortia). To perform the analysis we used the R package “TwoSampleMR” that
combines the effects sizes of instruments on exposures with those on outcomes via a meta-analysis.
We used “TwoSampleMR” package functions for allele harmonization between exposure and
outcome datasets, proxy variant substitution when SNPs from exposure were not genotyped in the
outcome data (Rsq>0.8 using the 1000G EUR reference data integrated into MR-Base), and
clumping to prune instrument SNPs for LD (the R script used for MR analyses is available upon
request). We used 5 different MR methods that were included as part of the “TwoSampleMR”
package to control for bias inherent to any one technique 18. For example, the multiplicative
random effects inverse variance-weighted (IVW) method is a weighted regression of instrument-
outcome effects on instrument-exposure effects with the intercept is set to zero. This method
generates a causal estimate of the exposure trait on outcome traits by regressing the, for example,
SNP-BMI trait association on the SNP-biomarker measure association, weighted by the inverse of
the SNP-biomarker measure association, and constraining the intercept of this regression to zero.
B Nogal 10
This constraint can result in unbalanced horizontal pleiotropy whereby the instruments influence
the outcome through causal pathways distinct from that through the exposure (thus violating the
second above-mentioned assumption). Such unbalanced horizontal pleiotropy distorts the
association between the exposure and the outcome, and the effect estimate from the IVW method
can be exaggerated or attenuated. However, unbalanced horizontal pleiotropy can be readily
assessed by the MR Egger method (via the MR Egger intercept), which provides a valid MR causal
estimate that is adjusted for the presence of such directional pleiotropy, albeit at the cost of
statistical efficiency. Finally, to ascertain the directionality of the various causal relationships
examined, we also performed each MR analysis in reverse where possible.
3
Results
Study population characteristics
Table 1 shows the demographic characteristics of the study population. We observed a
significant trend toward younger individuals reporting higher running volume, with more than
75% of the professional (PRO) group falling between the ages of 18 and 35 (Table 1S). Significant
differences were also observed in the distribution of males and females within study groups (Table
1). Moreover, higher running volume associated with significantly lower body mass index (BMI).
Thus, moving forward, combined comparisons of blood biomarkers as they relate to running
volume were adjusted for age, gender, and BMI.
Endurance exercise exhibits a modest association with clusters of blood biomarker features
In order to begin to understand the most important variables that may associate with endurance
exercise in the form of running, we performed a principal component analysis (PCA), dividing the
B Nogal
11
11
cohort into two most divergent groups in terms of exercise volume: PRO/high volume amateur
(HVAM) and sedentary (SED) groups. Using propensity matching, PRO and amateur athletes
who reported running >10h per week were combined into the PRO-HVAM group to balance out
the sample size between the exercising and non-exercising groups. Using this approach, we did
not observe a significant separation between these groups (data not shown). However, dividing
this dataset further into males and females yielded a modest degree of separation, with
hematological, inflammation, and lipid features, as well as BMI explaining some of the variance
(Figure 1 A through D). We hypothesized that there may more subtle relationships between
running volume and the blood biomarker features that contributed to distinguishing the endurance
exercise and sedentary groups, thus we next performed ANOVA analyses stratified by running
volume as categorized in Table 1.
Significant trends in glycemic, hematological, blood lipid, and inflammatory serum traits with
increasing running volumes
Weighted ANOVA analyses adjusted for age, gender, and BMI showed significant differences
among groups for multiple blood biomarkers (Table 2 and 2S, Figures 2 and 3). We observed a
trend toward lower HbA1c, hsCRP, RDW, WBC, ferritin, gamma-glutamyl transferase (GGT),
and LDL. HDL, hemoglobin (Hb), transferrin saturation (TS), alanine aminotransferase (ALT),
aspartate aminotransferase (AST), vitamin B12, folate, 25-hydroxy vitamin D, and creatine kinase
(CK) tended to be higher with increasing reported training volume, particularly in PRO runners
(Tables 2 and 2S, Figures 2 and 2S, Figure 3). Hct and Hb were higher only in PRO males,
whereas increased running volume associated with upward trend in these biomarkers in females
(Figure 3 A and B). Increased running volume was associated with markedly lower Fer in males,
whereas female runners did not exhibit varying levels, and SED females showed increased levels
B Nogal 12
(Figure 3 C). The low ferritin observed in male and female runners was not clinically significant.
ALT positively associated with running volume in females only (Figure 2S). Serum and RBC
magnesium (Mg) were both significantly lower in PRO runners relative to all other groups (Table
2 and Figure 3 D and E). Increasing levels of endurance exercise also appeared to be associated
with higher sex-hormone binding globulin (SHBG), particularly in PRO male runners (Figure 3
F).
Endurance exercise correlates with lower BMI across categories of genetic risk
Using publicly available GWAS summary statistics, we constructed blood biomarker polygenic
risk scores (PGSs) to explore potential genetic risk-mitigating effects of endurance exercise. Since
only a subset of the individuals in our cohort were genotyped, we aggregated the groups into 2
categories—PRO-HVAM and sedentary—to increase statistical power. This across-group sample
size increase generally did not sufficiently power the ANOVA analysis to detect statistically
significant trends (data not shown), though the BMI polygenic risk was suggestively mitigated for
both males and female PRO-HVAM runners across categories of genetic risk (Figure 4 B).
Increased running volume is associated with lower BMI which may drive biomarker changes
We observed a significant downward trend in the BMI with increased running volume for both
males and females, and, although some of the biomarker differences between sedentary and
exercising individuals remained significant after adjustment for BMI, their significance was
attenuated (Figure 4 A, p-value attenuation data not shown). Thus, we hypothesized that BMI may
be driving a significant portion of the observed variance in some of the biomarkers across the
groups. Thus, to explore causal relationships between weight and biomarker changes, we
performed 2S-MR with BMI-associated single-nucleotide polymorphisms (SNPs) as the
B Nogal
13
13
instrumental variables (IVs) for a subset of the healthspan-related biomarkers where BMI
explained a relatively large portion of the variance in our analysis. In general, these blood
biomarkers associated with inflammation (hsCRP and RDW), lipid metabolism (Tg and HDL),
glycemic control (HbA1c and Glu), as well as Alb and SHBG. We used GWAS summary statistics
and found that most of these BMI-blood biomarker relationships examined directionally aligned
with our study (except for LDL), and some were indicative of causal relationships in the BMI-
biomarker direction even after considering directional pleiotropy (Table 3S). We entertained the
possibility of reverse causality and thus repeated the 2S-MR using each of the biomarker levels as
the exposure and BMI as the outcome, and the results were generally not significant (except for
WBC – see Table 4S). Of note, to estimate the direct causal effects of running on blood
parameters, we attempted to find an instrumental variable for to approximate running as the
exposure from publicly available GWAS summary statistics. Toward this end, we found that
increasing levels of vigorous physical activity did associate with lower hsCRP, HbA1C, higher
HDL, and possibly higher SHBG (although the explained variance (R2) in this exposure was just
0.001009, the F statistic was 37.7, thus meeting the criteria of F > 10 for minimizing weak
instrument bias) (Figures 5 and 3S; Table 5S).
Vigorous physical activity associates with healthier behaviors
We hypothesized that those who exercise regularly may also partake in other healthful lifestyle
habits that may be contributing to more optimal blood biomarker signatures of wellness. However,
our dataset did not allow for systematic accounting of other lifestyle habits across all running
groups. Thus, we again leveraged the potential of the 2S-MR approach to inform potential
confounding associations between modifiable exposures and found that vigorous physical activity
such as running is at least suggestively associated with several behaviors associated with improved
B Nogal 14
health (Figure 4S). Our analysis showed that those who participate in increasing levels of vigorous
physical activity may be less likely to eat processed meat (IVW p = 0.0000013), sweets (IVW p =
0.32), and nap during the day (IVW p = 0.13), while increasing their intake of oily fish (IVW p =
0.029), salad/raw vegetable intake (IVW p = 0.00016), and fresh fruit (IVW p = 0.0027) (Table
6S). Furthermore, following our assessment of reverse causality, we found evidence for the
bidirectionality in the causal relationship between vigorous activity and napping during the day
and salad/raw vegetable intake, perhaps suggesting some degree of confounding due to population
stratification (Table 7S). The suggestive positive effect of fresh fruit and processed meat intake on
vigorous physical activity appeared to violate MR assumption (3) (Figure 1S) (horizontal
pleiotropy p-values 0.051 and 0.17, respectively – Figure 5S).
4
Discussion
In this report, we describe the variance in wellness-related blood biomarkers among self-reported
recreational runners, PRO runners, and individuals who do not report any exercise. Overall, we
find that 1) recreational running as an exercise appears to be an effective intervention toward
modifying several biomarkers indicative of improved metabolic health, 2) an apparent dose-
response relationship between running volume and BMI may itself be responsible for a proportion
of the apparent metabolic benefits, and 3) both PRO-level status and gender appear to associate
with heterogeneous physiological responses, particularly in iron and magnesium metabolism, as
well as some hormonal traits.
4.1
Self-reported running improves glycemia and lipidemia
We did not observe distinct clusters corresponding to self-reported high-volume/PRO runners and
the sedentary upon dimension reduction. This is, perhaps, not unexpected due, in part, to the self-
B Nogal
15
15
selected healthspan-oriented nature of our cohort, where even the sedentary subset of individuals
tends to exhibit blood biomarker levels in the normal clinical reference ranges. Furthermore, the
measurement of running volume via self-report may be vulnerable to overestimation, which may
have contributed to the blending of sedentary and exercise groups with respect to the serum
markers measured, resulting in only marginal separation between the groups 19,20. However, we
did observe significant individual blood biomarker variance with respect to reported running
volumes when the dataset was subjected to ANOVA, even after adjustment for age, sex, and BMI.
From among glycemic control blood biomarkers, we were able to detect a relatively small exercise
effect in both fasting glucose and HbA1c in this generally healthy cohort, where the average
measures of glycemia were below the prediabetic thresholds in even the sedentary subset of the
cohort. Larger exercise intervention effects on metabolic biomarkers may be expected in cohorts
that include individuals with more clinically significant baseline values 21.
Similarly, blood lipids improved with higher self-reported running volume, and this result has been
reported before in multiple controlled endurance exercise trials 22. The literature indicates that
HDL and Tg are two exercise-modifiable blood lipid biomarkers, with HDL being the most widely
reported to be modified by aerobic exercise 23,24. Although the mechanism behind this is not
entirely clear, it likely involves the modification of lecithin acyltransferase and lipoprotein lipase
activities following exercise training 25. We observed a similar trend in our blood biomarker
analysis, with HDL exhibiting an upward trend with increasing reported running volume. While
we also found Tg and LDL to decrease with increasing exercise volume, these trends were less
pronounced. Reports generally suggest that, in order to reduce LDL more consistently, the
intensity of aerobic exercise must be high enough 23. In the case of Tg, baseline levels may have
B Nogal 16
a significant impact on the exercise intervention effect, with individuals exhibiting higher baselines
showing greater improvements 13.
Importantly, these results suggests that exercise has a significant effect on glycemic control and
blood lipids even in the self-selected, already healthy individuals who are proactive about
preventing cardiometabolic disease.
4.2
Self-reported running and serum proxies of systemic inflammation
Chronic low-grade inflammation is one of the major risk factors for compromised cardiovascular
health and metabolic syndrome (MetS). While there is no shortage of inflammation-reducing
intervention studies on CVD patients with clinically high levels of metabolic inflammation, there
is less emphasis on modifiable lifestyle factors that can help stave off CVD and extend healthspan
in the generally healthy individual. Indeed, considering the pathological cardiovascular processes
begin shortly after birth, prevention in asymptomatic individuals may be a more appropriate
strategy toward decreasing the burden of CVD on the healthcare system 26.
Toward this end, increasing self-reported running volume appeared to associate with improved
markers of inflammation, as shown by the lower levels of hsCRP, WBC, as well as ferritin. Of
note, while the acute-phase protein, ferritin, is often used in the differential diagnosis of iron
deficiency anemia, the biomarker’s specificity appears to depend on the inflammatory state of the
individual, as it associates with hsCRP and inflammation more than iron stores, particularly in
those with higher BMI 27. Although serum ferritin and iron is reported to be lower in male and
female elite athletes 28, the observed overall negative association of ferritin with increased running
volume in our cohort may be an indication of lower levels of inflammation rather than
compromised iron stores, particularly since the average ferritin level across all groups was above
B Nogal
17
17
the clinical iron deficiency thresholds. Moreover, increased levels of ferritin have been associated
with insulin resistance and lower levels of adiponectin in the general population, both indicators
of increased systemic inflammation 29. Here, exercising groups with lower levels of ferritin also
exhibited glycemic and blood lipid traits indicative of improved metabolic states, further
supporting ferritin’s role as an inflammation proxy. Finally, Hb, TS and iron tended to be higher
in those who run for exercise compared to the SED group (with the TIBC lower), again suggesting
that runners, including the PRO group, were iron-sufficient in this cohort.
4.3
PRO athletes exhibit distinct biomarker signatures
PRO athletes exhibited lower serum and RBC Mg, which may be indication of the often-reported
endurance athlete hypomagnesaemia 30. While the serum Mg was still within normal clinical
reference range for both PRO female and male athletes, RBC Mg, a more sensitive biomarker of
Mg status 31, was borderline low in female PRO athletes and might suggest suboptimal dietary
intakes and/or much higher volume of running training compared to the other running groups (i.e.
>>10h /week). Indeed, this group also had elevated baseline CK and AST, which suggests a much
higher training intensity and/or volume. Moreover, PRO level athletes had adequate iron status
and serum B12 and folate in the upper quartile of the normal reference range, suggesting that these
athletes’ general nutrition status may have been adequate. These observations suggest that elite
endurance runners may need to pay particular attention to their magnesium status.
Further, we observed higher levels of SHBG in PRO male runners, a biomarker whose levels
positively correlate with various indexes of insulin sensitivity 32. However, since the average
SHBG levels in the SED group were not clinically low in both sexes, the observed increase in
SHBG levels induced by running in males may be a catabolic response, as cortisol levels in this
B Nogal 18
group were also higher. Indeed, Popovic et al have shown that endurance exercise may increase
SHBG, cortisol, and total testosterone levels at the expense of free testosterone levels 33. This
could perhaps in part be explained by higher exercise-induced adiponectin levels, which have been
shown to increase SHBG via cAMP kinase (AMPK) activation 34. However, since our data is
observational, we cannot rule out overall energy balance as a significant contributor to SHBG
levels. For example, caloric restriction (CR) has been shown to result in higher SHBG and cortisol
levels 32.
Finally, regarding the abovementioned PRO group elevated AST and CK biomarkers, evidence
suggests that normal reference ranges in both CK and AST in well-recovered athletes should be
adjusted up, as training and competition have a profound, non-pathological, impact on the activity
of these enzymes 35,36. Indeed, the recommendation appears to be not to use reference intervals
derived from the general population with hard-training (particularly competitive) athletes 36.
4.4
Effect of BMI on blood biomarkers
Since the current study is a cross-sectional analysis of self-reported running, we could not rule out
the possibility that factors other than exercise were the driving force behind the observed
biomarker variance among the groups examined. While factors such as diet, sleep, and/or
medication use were not readily ascertained in this free-living cohort at the time of this study, BMI
was readily available to evaluate this biomarker’s potential relative contribution to the observed
mean biomarker differences among self-reported runner groups.
Multiple studies have attempted to uncouple the effects of exercise and BMI reduction on blood
biomarker outcomes, with mixed results 37. For example, it is relatively well-known that acute
bouts of exercise improve glucose metabolism, but long-term effects are less well described 38.
B Nogal
19
19
Indeed, whether exercise without significant weight-loss is effective toward preventing metabolic
disease (and the associated blood biomarker changes) is inconclusive. From the literature, it
appears that, for endurance exercise to have significant effect on most blood biomarkers, the
volume of exercise needs to be very high, and this typically results in significant reduction in
weight. Thus, in practice, it is difficult to demonstrably uncouple the effects of significant exercise
and the associated weight-loss, and the results may depend on the blood biomarker in question.
Indeed, there is evidence that exercise without weight-loss does improve markers of insulin
sensitivity but not chronic inflammation, with the latter apparently requiring a reduction in
adiposity in the general population 39-41.
In our study of apparently healthy individuals, we observed a downward trend in BMI with
increasing self-reported running volume, and, although this study was not longitudinal and we are
thus unable to claim weight-loss, our 2S-MR analysis using BMI as the exposure nonetheless
suggests this biomarker to be responsible for a significant proportion of the modification of some
blood biomarkers.
4.5.1 Serum markers of systemic inflammation
Through our 2S-MR analyses, we show that BMI is causally associated with markers of systemic
inflammation, including RDW, folate, and hsCRP 27,42,43. Similar analyses have reported that
genetic variants that associate with higher BMI were associated with higher CRP levels, but not
the other way around 44. The prevailing mechanism proposed to explain this relationship appears
to be the pathological nature of overweight/obesity-driven adipose tissue that results in secretion
of proinflammatory cytokines such as IL-6 and TNFa, which then stimulate an acute hepatic
response, resulting in increased hsCRP levels (among other effects) 45. Thus, our 2S-MR analyses
B Nogal 20
and those of others 44 would indicate that the primary factor behind the lower systemic
inflammation in our cohort may be the exercise-associated lower BMI and not running exercise
per se, though the lower hsCRP in runners remained significant after adjustment for BMI in our
analysis.
Indeed, although a major driver behind reduced systemic inflammation may be a reduction in BMI
in the general population, additive effects of other lifestyle factors such as exercise cannot be
excluded. For example, a large body of cross-sectional investigations does indicate that physically
active individuals exhibit CRP levels that are 19-35% lower than less active individuals, even
when adjusted for BMI as was the case in the current analysis 41. Further, it’s been reported that
physical activity at a frequency of as little as 1 day per week is associated with lower CRP in
individuals who are otherwise sedentary, while more frequent exercise further reduces
inflammation 41.
Significantly, our entire cohort of self-selected apparently healthy individuals did not exhibit
clinically high hsCRP, with average BMI also below the overweight thresholds. Because all
subjects were voluntarily participating in a personalized wellness platform intended to optimize
blood biomarkers that included hsCRP, it is possible that some individuals from across the study
groups (both running and sedentary) in our cohort partook in some form of inflammation-reducing
dietary and/or lifestyle-based intervention. Thus, that we detected a significant difference in
hsCRP between exercising and non-exercising individuals in this self-selected already generally
healthy cohort may be suggestive of the potential for additional preventative effect of scheduled
physical activity on low-grade systemic inflammation in the generally healthy individual.
4.5.2 Blood lipids
B Nogal
21
21
Controlled studies that tightly track exercise and the associated adiposity reduction have reported
that body fat reduction (and not improvement in fitness as measured via VO2max) is a predictor of
HDL, LDL, and Tg 46. Similarly, though BMI is an imperfect measure of adiposity, our 2S-MR
analysis suggests that this biomarker is causally associated with improved levels of HDL and Tg,
though not LDL. This latter finding replicates a report by Hu et al. who, using the Global Lipids
Genetics Consortium GWAS summary statistics, applied a network MR approach that revealed
causal associations between BMI and blood lipids, where Tg and HDL, but not LDL, were found
to trend toward unhealthy levels with increasing adiposity 47. On the other hand, others
implemented a robust BMI genetic risk score and demonstrated a causal association of adiposity
with peripheral artery disease and a multiple linear regression showed a strong association with
HDL, TC, and LDL, among other metabolic parameters 48. In our cohort, given the lack of
evidence for a causal BMI-LDL association and the overall healthy levels of BMI, the observed a
significant improvement in LDL may be a result of marked running intensity and/or volume,
possibly combined with the aforementioned additional wellness program intervention variables.
4.5.3 Hormonal traits
As described above, we observed a trend toward increased plasma cortisol and SHBG in runners,
particularly PRO level athletes. The effects on cortisol are consistent with a report by Houmanrd
et al, who found male distance runners to exhibit higher levels of baseline cortisol 49. With respect
to the effects of BMI on baseline cortisol levels, this observation is generally supported by our 2S-
MR analyses with evidence for a consistent effect of increased cortisol with decreasing BMI.
However, this association was suggestive at best, indicating that the higher levels of cortisol
exhibited in the PRO runners with significant lower adiposity are not likely to be solely explained
by their lower BMI. Indeed, the relationship between BMI and cortisol appears to be complex,
B Nogal 22
with some reports suggesting a U-shaped relationship, where the glucocorticoid’s levels associate
negatively up to about a BMI of 30 kg/m2, then exhibiting a positive correlation into obesity
phenotypes 50. MR statistical models generally do not account for such non-linearity and would
require a more granular demographical treatment, which is not possible using only GWAS
summary statistics data in the context of 2S-MR 17,51.
4.6
Behavioral traits associated with increase physical activity
The combination of the body of the literature that describes the effects of endurance training on
blood biomarkers, and our own analysis that included markers such as CK and AST, makes us
cautiously assured that most of the abovementioned blood biomarker signatures are indeed a result
of the interplay between self-reported running and the associated lower BMI. However, as this is
a self-report-based analysis and we were unable to track other subject behaviors in this free-living
cohort, we acknowledge that multiple behaviors that associate with exercise may be influencing
our results.
Toward this end, our exploratory 2S-MR analyses revealed potentially causal relationships
between vigorous exercise and multiple dietary habits that have been shown to improve the
biomarkers we examined. Indeed, diets that avoid processed meat and sweets while providing
ample amounts of fresh fruits, as well as oily fish have been shown to be anti-inflammatory, and
improve glycemic control and dyslipidemia 52,53. That physically active individuals are also more
likely to make healthier dietary choices adds insight to the potential confounders in ours and
others’ observational analyses, and this similar associations have previously been reported 54-56.
For example, using a calculated healthy eating motivation score, Naughton et al. showed that those
who partake in more than 2 hours of vigorous physical activity are almost twice as likely to be
B Nogal
23
23
motivated to eat healthy 56. Indeed, upon closer examination, the genetic instruments used to
approximate vigorous physical activity as the exposure in this work included variants in the genes
DPY19L1, CADM2, CTBP2, EXOC4, and FOXO3 57. Of these, CADM2 encodes proteins that are
involved in neurotransmission in brain regions well known for their involvement in executive
function, including motivation, impulse regulation and self-control 58. Moreover, variants within
this locus have been associated with obesity-related traits 59. Thus, it is likely that the improved
metabolic outcomes seen here with our self-reported runners are a composite result of both these
individuals exercise and dietary habits. Importantly, the above suggests that a holistic wellness
lifestyle approach is in practice the most likely to be most effective toward preventing
cardiometabolic disease. Nonetheless, the focus of this work – exercise in the form of running –
is known to significantly improve cardiorespiratory fitness (CRF), which has been shown to be
an independent predictor of CVD risk and total mortality, outcomes that indeed correlate with
dysregulated levels in many of the blood biomarkers examined in this work 7.
4.7
Study limitations
This study is based on self-reported running and thus has several limitations. First, it is generally
known that subjects tend to overestimate their commitment to exercise when self-reporting,
although in our cohort is a self-selected health-oriented population that is possibly less likely to
over-report their running volume. Furthermore, although the robust increasing trend in baselines
for muscle damage biomarkers (CK, AST) that have been shown to be associated with participation
in sports and exercise provides indirect evidence that the running groups were indeed participating
in increasing volumes of strenuous physical activity, we cannot confirm whether the reported
running was performed overground or on a treadmill, which may result in some heterogeneity in
physiological responses , nor can we ascertain the actual training volume of PRO-level runners.
B Nogal 24
We also cannot exclude the possibility that the running groups also participated in other forms of
exercise (such as strength training) or partook in other wellness program interventions that may
have influenced their blood biomarkers and/or BMI via lean muscle accretion. Toward this end,
we have attempted to shed light on potential behavioral covariates related to vigorous physical
activity via 2S-MR. Finally, while this cohort is generally healthy, we cannot exclude the potential
for unmeasured confounders such as medications, nutritional supplements, and unreported health
conditions.
2S- MR enables the assessment of causal relationships between modifiable traits and is less prone
to the so-called “winner’s curse” that more readily affects one-sample MR analyses 17,51. Because
2S-MR uses GWAS summary statistics for both exposure and outcome, it is possible to increase
statistical power because of the increased sample sizes. However, horizontal pleiotropy is still a
concern that can skew the results. Currently, there is no gold standard MR analysis method, thus
we used different techniques (IVW, MR-Egger, and median-based estimations – all of which are
based on different assumptions and thus biases) to evaluate the consistency among these estimators
and only reported associations as ‘causal’ if there was cross-model consistency. Nonetheless, an
exposure such as BMI is a complex trait that is composed of multiple sub-phenotypes (such as
years of education) that could be driving the causal associations.
5
Conclusion
Running is one of the most common forms of vigorous exercise practiced globally, thus making it
a compelling target of research studies toward understanding its applicability in chronic disease
prevention. Our cross-sectional study offers insight into the biomarker signatures of self-reported
B Nogal
25
25
running in generally healthy individuals that suggest improved insulin sensitivity, blood lipid
metabolism, and systemic inflammation. Furthermore, using 2S-MR in independent datasets we
provide additional evidence that some biomarkers are readily modified BMI alone, while others
appear
to
respond
to
the
combination
of
varying
exercise
and
BM
I. Our additional bi-directional 2S-MR analyses toward understanding the causal relationships
between partaking in vigorous physical activity and other healthy behaviors highlight the inherent
challenge in disambiguating exercise intervention effects in cross sectional studies of free-living
populations, where healthy behaviors such as exercising and healthy dietary habits co-occur.
Overall, our analysis shows that the differences between those who run and the sedentary in our
cohort are likely a combination of the specific physiological effects of exercise, the associated
changes in BMI, and lifestyle habits associated with those who exercise, such as food choices and
baseline activity level. Looking ahead, the InsideTracker database is continuously augmented
with blood chemistry, genotyping, and activity tracker data, facilitating further investigation of the
effects of various exercise modalities on phenotypes related to healthspan, including longitudinal
analyses and more granular dose-response dynamics.
Data Availability Statement
The full set of biomarker change correlations has been made available in the Supplementary
Information files. Specific components of the raw dataset are available upon reasonable request
from the corresponding author. 2S-MR analysis was performed using publicly available datasets
via the TwoSampleMR R package.
Ethics statement
B Nogal 26
This study was submitted to The Institutional Review Board (IRB), which determined this work
was not subject to a review based on category 4 exemption (“secondary research” with de-
identified subjects).
Author contributions
BN performed the 2S-MR analyses, calculated PGSs, and wrote the manuscript; SV performed
blood biomarker and blood biomarker X PGS interaction analysis; PF calculated PGSs; MJ, AT,
and GB provided guidance. All authors have read and agreed to the published version of the
manuscript.
Funding
InsideTracker was the sole funding source.
Conflict of interest
InsideTracker is a direct-to-consumer blood biomarker and genomics company providing its
users with nutritional and exercise recommendations toward improving wellness. B.N., S.V.,
P.F., and G.B. are employees of InsideTracker.
B Nogal
27
27
Acknowledgments
InsideTracker is the sole funding source. We thank Michelle Cawley and Renee Deehan for their
assistance with background subject matter research and insightful conversations.
B Nogal 28
References
1.
Lavie CJ, Ozemek C, Carbone S, Katzmarzyk PT, Blair SN. Sedentary Behavior, Exercise,
and Cardiovascular Health. Circ Res. Mar 2019;124(5):799-815.
doi:10.1161/CIRCRESAHA.118.312669
2.
Carlson SA, Adams EK, Yang Z, Fulton JE. Percentage of Deaths Associated With
Inadequate Physical Activity in the United States. Prev Chronic Dis. Mar 29 2018;15:E38.
doi:10.5888/pcd18.170354
3.
Antonicelli R, Spazzafumo L, Scalvini S, et al. Exercise: a "new drug" for elderly patients
with chronic heart failure. Aging (Albany NY). May 2016;8(5):860-72.
doi:10.18632/aging.100901
4.
Sgro P, Emerenziani GP, Antinozzi C, Sacchetti M, Di Luigi L. Exercise as a drug for
glucose management and prevention in type 2 diabetes mellitus. Curr Opin Pharmacol. Aug
2021;59:95-102. doi:10.1016/j.coph.2021.05.006
5.
Vina J, Sanchis-Gomar F, Martinez-Bello V, Gomez-Cabrera MC. Exercise acts as a drug;
the pharmacological benefits of exercise. Br J Pharmacol. Sep 2012;167(1):1-12.
doi:10.1111/j.1476-5381.2012.01970.x
6.
Lee EC, Fragala MS, Kavouras SA, Queen RM, Pryor JL, Casa DJ. Biomarkers in Sports and
Exercise: Tracking Health, Performance, and Recovery in Athletes. J Strength Cond Res. Oct
2017;31(10):2920-2937. doi:10.1519/JSC.0000000000002122
7.
Lin X, Zhang X, Guo J, et al. Effects of Exercise Training on Cardiorespiratory Fitness and
Biomarkers of Cardiometabolic Health: A Systematic Review and Meta-Analysis of Randomized
Controlled Trials. J Am Heart Assoc. Jun 26 2015;4(7)doi:10.1161/JAHA.115.002014
B Nogal
29
29
8.
Li X, Ploner A, Wang Y, et al. Clinical biomarkers and associations with healthspan and
lifespan: Evidence from observational and genetic data. EBioMedicine. Apr 2021;66:103318.
doi:10.1016/j.ebiom.2021.103318
9.
Mailliez A, Guilbaud A, Puisieux F, Dauchet L, Boulanger E. Circulating biomarkers
characterizing physical frailty: CRP, hemoglobin, albumin, 25OHD and free testosterone as best
biomarkers. Results of a meta-analysis. Exp Gerontol. Oct 1 2020;139:111014.
doi:10.1016/j.exger.2020.111014
10.
Hirata T, Arai Y, Yuasa S, et al. Associations of cardiovascular biomarkers and plasma
albumin with exceptional survival to the highest ages. Nat Commun. Jul 30 2020;11(1):3820.
doi:10.1038/s41467-020-17636-0
11.
Erema VV, Yakovchik AY, Kashtanova DA, et al. Biological Age Predictors: The Status Quo
and Future Trends. Int J Mol Sci. Dec 1 2022;23(23)doi:10.3390/ijms232315103
12.
Hartmann A, Hartmann C, Secci R, Hermann A, Fuellen G, Walter M. Ranking Biomarkers
of Aging by Citation Profiling and Effort Scoring. Front Genet. 2021;12:686320.
doi:10.3389/fgene.2021.686320
13.
Trejo-Gutierrez JF, Fletcher G. Impact of exercise on blood lipids and lipoproteins. J Clin
Lipidol. Jul 2007;1(3):175-81. doi:10.1016/j.jacl.2007.05.006
14.
Westerman K, Reaver A, Roy C, et al. Longitudinal analysis of biomarker data from a
personalized nutrition platform in healthy subjects. Sci Rep. Oct 2 2018;8(1):14685.
doi:10.1038/s41598-018-33008-7
15.
Fox J WS. An R Companion to Applied Regression. Third ed. Sage, Thousand Oaks CA;
2019.
B Nogal 30
16.
Ho D, Imai K, King G, Stuart EA. MatchIt: Nonparametric Preprocessing for Parametric
Causal Inference. Journal of Statistical Software. 06/14 2011;42(8):1 - 28.
doi:10.18637/jss.v042.i08
17.
Burgess S, Davey Smith G, Davies NM, et al. Guidelines for performing Mendelian
randomization investigations. Wellcome Open Res. 2019;4:186.
doi:10.12688/wellcomeopenres.15555.2
18.
Hemani G, Zheng J, Elsworth B, et al. The MR-Base platform supports systematic causal
inference across the human phenome. Elife. May 30 2018;7doi:10.7554/eLife.34408
19.
Bogl LH, Pietilainen KH, Rissanen A, Kaprio J. Improving the accuracy of self-reports on
diet and physical exercise: the co-twin control method. Twin Res Hum Genet. Dec
2009;12(6):531-40. doi:10.1375/twin.12.6.531
20.
Yuen HK, Wang E, Holthaus K, et al. Self-reported versus objectively assessed exercise
adherence. Am J Occup Ther. Jul-Aug 2013;67(4):484-9. doi:10.5014/ajot.2013.007575
21.
Ostman C, Smart NA, Morcos D, Duller A, Ridley W, Jewiss D. The effect of exercise
training on clinical outcomes in patients with the metabolic syndrome: a systematic review and
meta-analysis. Cardiovasc Diabetol. Aug 30 2017;16(1):110. doi:10.1186/s12933-017-0590-y
22.
Fikenzer K, Fikenzer S, Laufs U, Werner C. Effects of endurance training on serum lipids.
Vascul Pharmacol. Feb 2018;101:9-20. doi:10.1016/j.vph.2017.11.005
23.
Mann S, Beedie C, Jimenez A. Differential effects of aerobic exercise, resistance training
and combined exercise modalities on cholesterol and the lipid profile: review, synthesis and
recommendations. Sports Med. Feb 2014;44(2):211-21. doi:10.1007/s40279-013-0110-5
B Nogal
31
31
24.
Tambalis K, Panagiotakos DB, Kavouras SA, Sidossis LS. Responses of blood lipids to
aerobic, resistance, and combined aerobic with resistance exercise training: a systematic review
of current evidence. Angiology. Oct-Nov 2009;60(5):614-32. doi:10.1177/0003319708324927
25.
Calabresi L, Franceschini G. Lecithin:cholesterol acyltransferase, high-density
lipoproteins, and atheroprotection in humans. Trends Cardiovasc Med. Feb 2010;20(2):50-3.
doi:10.1016/j.tcm.2010.03.007
26.
Kemper HC, Snel J, Verschuur R, Storm-van Essen L. Tracking of health and risk indicators
of cardiovascular diseases from teenager to adult: Amsterdam Growth and Health Study. Prev
Med. Nov 1990;19(6):642-55. doi:10.1016/0091-7435(90)90061-n
27.
Khan A, Khan WM, Ayub M, Humayun M, Haroon M. Ferritin Is a Marker of
Inflammation rather than Iron Deficiency in Overweight and Obese People. J Obes.
2016;2016:1937320. doi:10.1155/2016/1937320
28.
Nabhan D, Bielko S, Sinex JA, et al. Serum ferritin distribution in elite athletes. J Sci Med
Sport. Jun 2020;23(6):554-558. doi:10.1016/j.jsams.2019.12.027
29.
Ku BJ, Kim SY, Lee TY, Park KS. Serum ferritin is inversely correlated with serum
adiponectin level: population-based cross-sectional study. Dis Markers. 2009;27(6):303-10.
doi:10.3233/DMA-2009-0676
30.
Pollock N, Chakraverty R, Taylor I, Killer SC. An 8-year Analysis of Magnesium Status in
Elite International Track & Field Athletes. J Am Coll Nutr. Jul 2020;39(5):443-449.
doi:10.1080/07315724.2019.1691953
31.
Arnaud MJ. Update on the assessment of magnesium status. Br J Nutr. Jun 2008;99
Suppl 3:S24-36. doi:10.1017/S000711450800682X
B Nogal 32
32.
Simo R, Saez-Lopez C, Barbosa-Desongles A, Hernandez C, Selva DM. Novel insights in
SHBG regulation and clinical implications. Trends Endocrinol Metab. Jul 2015;26(7):376-83.
doi:10.1016/j.tem.2015.05.001
33.
Popovic B, Popovic D, Macut D, et al. Acute Response to Endurance Exercise Stress:
Focus on Catabolic/anabolic Interplay Between Cortisol, Testosterone, and Sex Hormone
Binding Globulin in Professional Athletes. J Med Biochem. Mar 2019;38(1):6-12.
doi:10.2478/jomb-2018-0016
34.
Simo R, Saez-Lopez C, Lecube A, Hernandez C, Fort JM, Selva DM. Adiponectin
upregulates SHBG production: molecular mechanisms and potential implications.
Endocrinology. Aug 2014;155(8):2820-30. doi:10.1210/en.2014-1072
35.
Mougios V. Reference intervals for serum creatine kinase in athletes. Br J Sports Med.
Oct 2007;41(10):674-8. doi:10.1136/bjsm.2006.034041
36.
Banfi G, Morelli P. Relation between body mass index and serum aminotransferases
concentrations in professional athletes. J Sports Med Phys Fitness. Jun 2008;48(2):197-200.
37.
Ross R, Janiszewski PM. Is weight loss the optimal target for obesity-related
cardiovascular disease risk reduction? Can J Cardiol. Sep 2008;24 Suppl D:25D-31D.
doi:10.1016/s0828-282x(08)71046-8
38.
Ross R. Does exercise without weight loss improve insulin sensitivity? Diabetes Care.
Mar 2003;26(3):944-5. doi:10.2337/diacare.26.3.944
39.
Cerqueira E, Marinho DA, Neiva HP, Lourenco O. Inflammatory Effects of High and
Moderate Intensity Exercise-A Systematic Review. Front Physiol. 2019;10:1550.
doi:10.3389/fphys.2019.01550
B Nogal
33
33
40.
Church TS, Earnest CP, Thompson AM, et al. Exercise without weight loss does not
reduce C-reactive protein: the INFLAME study. Med Sci Sports Exerc. Apr 2010;42(4):708-16.
doi:10.1249/MSS.0b013e3181c03a43
41.
Plaisance EP, Grandjean PW. Physical activity and high-sensitivity C-reactive protein.
Sports Med. 2006;36(5):443-58. doi:10.2165/00007256-200636050-00006
42.
Salvagno GL, Sanchis-Gomar F, Picanza A, Lippi G. Red blood cell distribution width: A
simple parameter with multiple clinical applications. Crit Rev Clin Lab Sci. 2015;52(2):86-105.
doi:10.3109/10408363.2014.992064
43.
Carmel R, Green R, Rosenblatt DS, Watkins D. Update on cobalamin, folate, and
homocysteine. Hematology Am Soc Hematol Educ Program. 2003:62-81.
doi:10.1182/asheducation-2003.1.62
44.
Welsh P, Polisecki E, Robertson M, et al. Unraveling the directional link between
adiposity and inflammation: a bidirectional Mendelian randomization approach. J Clin
Endocrinol Metab. Jan 2010;95(1):93-9. doi:10.1210/jc.2009-1064
45.
Maachi M, Pieroni L, Bruckert E, et al. Systemic low-grade inflammation is related to
both circulating and adipose tissue TNFalpha, leptin and IL-6 levels in obese women. Int J Obes
Relat Metab Disord. Aug 2004;28(8):993-7. doi:10.1038/sj.ijo.0802718
46.
Katzmarzyk PT, Leon AS, Rankinen T, et al. Changes in blood lipids consequent to aerobic
exercise training related to changes in body fatness and aerobic fitness. Metabolism. Jul
2001;50(7):841-8. doi:10.1053/meta.2001.24190
B Nogal 34
47.
Hu X, Zhuang XD, Mei WY, et al. Exploring the causal pathway from body mass index to
coronary heart disease: a network Mendelian randomization study. Ther Adv Chronic Dis.
2020;11:2040622320909040. doi:10.1177/2040622320909040
48.
Huang Y, Xu M, Xie L, et al. Obesity and peripheral arterial disease: A Mendelian
Randomization analysis. Atherosclerosis. Apr 2016;247:218-24.
doi:10.1016/j.atherosclerosis.2015.12.034
49.
Houmard JA, Costill DL, Mitchell JB, Park SH, Fink WJ, Burns JM. Testosterone, cortisol,
and creatine kinase levels in male distance runners during reduced training. Int J Sports Med.
Feb 1990;11(1):41-5. doi:10.1055/s-2007-1024760
50.
Schorr M, Lawson EA, Dichtel LE, Klibanski A, Miller KK. Cortisol Measures Across the
Weight Spectrum. J Clin Endocrinol Metab. Sep 2015;100(9):3313-21. doi:10.1210/JC.2015-2078
51.
O'Donnell CJ, Sabatine MS. Opportunities and Challenges in Mendelian Randomization
Studies to Guide Trial Design. JAMA Cardiol. Oct 1 2018;3(10):967.
doi:10.1001/jamacardio.2018.2863
52.
Hosseini B, Berthon BS, Saedisomeolia A, et al. Effects of fruit and vegetable
consumption on inflammatory biomarkers and immune cell populations: a systematic literature
review and meta-analysis. Am J Clin Nutr. Jul 1 2018;108(1):136-155. doi:10.1093/ajcn/nqy082
53.
Djousse L, Arnett DK, Coon H, Province MA, Moore LL, Ellison RC. Fruit and vegetable
consumption and LDL cholesterol: the National Heart, Lung, and Blood Institute Family Heart
Study. Am J Clin Nutr. Feb 2004;79(2):213-7. doi:10.1093/ajcn/79.2.213
54.
L D. Physical Activity and Dietary Habits of College Students. The Journal of Nurse
Practitioners. February 2015 2015;11(2):192-198.e2.
B Nogal
35
35
55.
Shi X, Tubb L, Fingers ST, Chen S, Caffrey JL. Associations of physical activity and dietary
behaviors with children's health and academic problems. J Sch Health. Jan 2013;83(1):1-7.
doi:10.1111/j.1746-1561.2012.00740.x
56.
Naughton P, McCarthy SN, McCarthy MB. The creation of a healthy eating motivation
score and its association with food choice and physical activity in a cross sectional sample of
Irish adults. Int J Behav Nutr Phys Act. Jun 6 2015;12:74. doi:10.1186/s12966-015-0234-0
57.
Klimentidis YC, Raichlen DA, Bea J, et al. Genome-wide association study of habitual
physical activity in over 377,000 UK Biobank participants identifies multiple variants including
CADM2 and APOE. Int J Obes (Lond). Jun 2018;42(6):1161-1176. doi:10.1038/s41366-018-0120-
3
58.
Arends RM, Pasman JA, Verweij KJH, et al. Associations between the CADM2 gene,
substance use, risky sexual behavior, and self-control: A phenome-wide association study.
Addict Biol. Nov 2021;26(6):e13015. doi:10.1111/adb.13015
59.
Morris J, Bailey MES, Baldassarre D, et al. Genetic variation in CADM2 as a link between
psychological traits and obesity. Scientific Reports. 2019/05/14 2019;9(1):7339.
doi:10.1038/s41598-019-43861-9
B Nogal 36
Table 1 Study Population Demographics
Group
N
Female, %
Age, yrs
Body mass index, kg/m2
PRO
82
53.7%
33.68
20.15
HVAM
1103
52.9%
39.48
22.57
MVAM
6747
54.2%
41.49
23.35
LVAM
10877
34.2%
41.16
24.72
SED
4428
48.9%
44.25
27.83
PRO = Professional, HVAM = high volume amateur (>10 hr), MVAM = medium
volume amateur (3-10hr), LVAM = low volume amateur (<3 hr), SED = sedentary
B Nogal
37
37
Table 2 Blood Biomarkers Significantly Different Among Sedentary
Individuals and Those Who Partake in Running for Exercise to
Various Degrees
Biomarker
ANOVA p-value
Trend p-value
lowest mean
highest mean
Alb
<1e-16
<0.001
MVAM
PRO
ALT
<1e-16
<1e-16
SED
PRO
AST
<1e-16
<0.001
SED
PRO
B12
<0.001
<0.001
SED
PRO
Chol
<0.001
0.005
PRO
SED
CK
<1e-16
<1e-16
SED
PRO
Cor
<0.001
0.675
SED
PRO
FE
<0.001
0.119
SED
PRO
Fer
<1e-16
<1e-16
MVAM
SED
Fol
<1e-16
<0.001
SED
PRO
FT
<0.001
0.013
SED
PRO
GGT
<1e-16
<0.001
PRO
SED
Glu
0.087
0.184
PRO
SED
Hb
0.002
<0.001
MVAM
PRO
HCT
0.053
0.055
MVAM
PRO
HDL
<1e-16
<0.001
SED
PRO
HbA1c
<0.001
0.010
PRO
SED
B Nogal 38
hsCRP
<0.001
0.176
PRO
SED
LDL
<0.001
0.006
PRO
SED
Mg
<0.001
0.276
PRO
SED
MPV
0.058
0.089
SED
HVAM
Na
<1e-16
0.622
HVAM
SED
RBC_Mg
<0.001
0.773
PRO
SED
RDW
<1e-16
0.002
PRO
SED
SHBG
<1e-16
0.004
SED
PRO
Tg
<1e-16
<1e-16
PRO
SED
WBC
<1e-16
<1e-16
PRO
SED
Figure 1
Click here to access/download;Figure;Figure_1.jpg
Figure 2
Click here to access/download;Figure;Figure_2.jpg
Figure 3
Click here to access/download;Figure;Figure_3.jpg
Figure 4
Click here to access/download;Figure;Figure_4.jpg
Figure 4
Click here to access/download;Figure;Figure_5 .jpg
Supporting Information
Click here to access/download
Supporting Information
Supplementary_Materials_PONE.pdf
| Dose response of running on blood biomarkers of wellness in generally healthy individuals. | 11-15-2023 | Nogal, Bartek,Vinogradova, Svetlana,Jorge, Milena,Torkamani, Ali,Fabian, Paul,Blander, Gil | eng |
PMC8771017 | 3182
http://journals.tubitak.gov.tr/medical/
Turkish Journal of Medical Sciences
Turk J Med Sci
(2021) 51: 3182-3193
© TÜBİTAK
doi:10.3906/sag-2106-271
Short-term impact of the Covid-19 pandemic on the global and Turkish economy
Ömer AÇIKGÖZ*, Aslı GÜNAY
Department of Economics, Faculty of Political Science, Social Sciences University of Ankara, Ankara, Turkey
* Correspondence: [email protected]
1. Introduction
The Covid-19 pandemic, a new strain discovered in China
in December 2019, has killed millions of people and
transformed the world forever. It is a historic event since it is
not just a health issue; it also has global economic, political,
and social dimensions. As of August 2021, the virus had
infected more than 205 million people worldwide, resulting
in around 4.3 million deaths, and more than 2.4 billion
vaccine doses have been administered globally [1]. Besides,
the International Monetary Fund (IMF) indicates that the
global economy was experiencing its worst crisis with a 3.5%
global gross domestic product (GDP) fall in 2020 since the
Great Depression of the 1930s [2, 3] compared to an estimated
15% GDP decline in between 1929 and 1932 worldwide [4].
While output in the United States (US) declined by 3.5%,
the economy contracted by 6.7% and 4.7% in the euro area
and Japan, respectively, in 2020. The pandemic is estimated
to have pushed 119–124 million people into poverty in 2020
due to the global economic recession [5].
Many scientific comparisons have been drawn between
the Covid-19 pandemic and preceding pandemics
(Spanish flu, Asian flu, Hong Kong flu, and swine flu)
[6] to demonstrate the magnitude of mortality rate and
economic collapse caused by the Covid-19 pandemic.
However, considering their worldwide dissemination,
nature, intensity, and socioeconomic characteristics,
drawing comparisons between them can be difficult. The
mortality impact of the Covid-19 pandemic, for example,
will be less than the Spanish flu that is estimated to have
killed roughly 40 million people globally in 1918 [7,
8]. Furthermore, despite the lack of economic data for
analyzing the economic impact of the Spanish flu, it is
estimated that GDP and consumption fell by 6% and 8% in
the typical country, respectively, and that many businesses,
particularly those in the service and entertainment
industries, suffered significant amount losses in revenue
[8, 9]. On the other hand, existing studies show that the
global death number of the Asian flu in 1957 and the Hong
Background/aim: The Covid-19 pandemic is one of those rare events that affects everyone on earth and changes our lives. The pandemic,
which has killed over four million people worldwide, is putting unprecedented pressure on governments to maintain essential health
and social services, as well as keep their economies running, even as the virus threatens people’s daily life on every level. Thus, the
purpose of this study is to discuss the short-term economic impact of the pandemic by assessing its costs using official economic data for
both the world and Turkey. Furthermore, this research highlights possible economic, social, and political pathways for a postpandemic
new world.
Materials and methods: This study is a review article that overviews and tracks the economic development of the Covid-19 pandemic
from the start, synthesizes and compares current data of reliable institutions, and provides an overall assessment.
Results: The pandemic has certainly caused short-term and long-term damage to economies and living standards for many people.
Although there are estimates on what this damage is, the exact degree of the damage is still unknown. However, it seems that the
recovery will be gradual, long-lasting, and unpredictable due to the unprecedented uncertainty characteristic of the pandemic.
Conclusion: Early economic growth projections show that there will be no ordinary recovery for the world economy since short-term
countries’ recovery paths are different. It is likely to remain uneven and depend on the effectiveness of the vaccination process, fiscal
policy support, public health management, and hard-hit sectors’ growth size in economies. Due to the uncertainty and lack of confidence,
governments should ensure an equal and sustainable economic recovery from the Covid-19 pandemic by conducting flexible monetary
and fiscal policies. However, without structural reforms, economies can not boost either in the short-term and long-term.
Key words: Covid-19, economy, pandemic, Turkey, world
Received: 22.06.2021 Accepted/Published Online: 07.08.2021 Final Version: 17.12.2021
Review Article
This work is licensed under a Creative Commons Attribution 4.0 International License.
AÇIKGÖZ and GÜNAY / Turk J Med Sci
3183
Kong flu in 1968 was around one million people. Moreover,
both pandemics had no significant worldwide economic
impact [10, 11]. As a result, early indications suggest that
the Covid-19 pandemic will have a similar global impact
with the Spanish flu rather than others in the end.
Covid-19 pandemic is a global crisis, and all countries
have been affected by this crisis. Developed as well as
emerging and developing countries are experiencing a
recession. Countries whose economies substantially rely
on tourism and hospitality, travel, and entertainment
sector have been particularly hard hit. For example, the
global GDP loss from the pandemic crisis could be around
9 trillion dollars over 2020 and 2021, greater than the
economies of Japan and Germany combined [2]. Moreover,
with inferior health systems, smaller fiscal support, and
high debt levels, both emerging and developing countries
and low-income countries have faced extra challenges for
recovery from that noticeable recession.
At the beginning of the Covid-19 pandemic, global
uncertainty was at an all-time high, and it continues to be
so. Although global economic and policy uncertainty has
decreased by about 60% since the onset of the Covid-19
pandemic in the first quarter of 2020, the World Uncertainty
Index (WUI) shows that it is still about 50% higher than
its historical average from 1996 to 2010 [12]. Hence, the
Covid-19 pandemic’s uncertainty is unprecedented since
there is a great deal of uncertainty about almost every
aspect of the Covid-19 crisis [11]. These are labeled as the
virus’ infectiousness and lethality [13], the time required
to develop and deploy effective vaccines [14], the duration
of social isolation [15, 16], macroeconomic consequences,
and government policy responses in both the short-term
and long-term [17], the shifts in consumer spending
patterns, travel, logistics, new business and working
formation [18, 19]. This situation suggests that the global
economy will not recover regularly; in other words, the
time it takes for each country to recover will most likely
vary due to the effectiveness of the vaccination process
and public health care measurements implemented by
each country. For example, according to the Organisation
for Economic Co-operation and Development (OECD),
much of Europe will take nearly three years, whereas
Korea and the US are likely to recover to prepandemic per
capita income levels in roughly 18 months [3]. As a result,
the most confusing impact of the Covid-19 pandemic
on society, economics, and policies is unprecedented
uncertainty since how the pandemic will evolve and end
is still ambiguous.
The Covid-19 crisis demonstrates that governments,
not markets, are the ones that provide much-needed help
during the global economic recession. In other words,
the Covid-19 pandemic has led to a collapse of the free
market phenomena, indicating markets are the only
solution mechanism for practically every problem that
societies encounter since 1981. Nevertheless, the Covid-19
pandemic has promoted the governments’ intervention
rather than the market intervention. Almost every major
industry has sought financial aid from the government
during the pandemic. Moreover, small enterprises have
been pleading for zero-interest loans, tax cuts, and outright
cash. As a result, the Covid-19 pandemic has demonstrated
that markets alone cannot recover economies in this global
crisis, and more market and government collaboration will
most likely be new economic policy in the postpandemic
world [20].
Already existing disparities and gaps in health and
social protection systems between countries have been
severely revealed, and in many cases worsened, by the
Covid-19 pandemic. Countries with strong health and
social protection systems responded better to the crisis by
guaranteeing access to health care services, also providing
jobs and income security for the neediest, such as informal
workers, daily wage earners, self-employed workers,
migrants, and the homeless. Countries that do not have
robust health and social protection systems, on the other
hand, have required international assistance to enable an
adequate initial reaction to the pandemic. In this respect,
the Covid-19 pandemic presents a chance for countries to
prioritize investments in their health and social protection
systems and develop them to better deal with future crises
[21].
Population mental health has deteriorated significantly
since the start of the Covid-19 pandemic. The OECD
shows that rates of anxiety and depression increased
in 2020 compared to previous years [22]. Economic
insecurity, unemployment, lower-income, death fear,
domestic violence, mobility restrictions, media exposure
about the pandemic, and social isolation have been the
main factors that has led to an unprecedented worsening
of population mental health during the Covid-19
pandemic [22, 23]. Due to these risk factors, loneliness
and individualism are likely to have associated with the
pandemic [24]. As individuals are isolated from social life,
they have begun to behave more individually than before.
People have started to form their own living spaces for
conducting their socioeconomic lives based on personal
freedoms with the accelerated digitalization in all areas of
life. Hence, populations’ social well-being, and their social
life and relationships have worsened noticeably during
the pandemic. Despite the new positive developments like
vaccines, many are still wondering how the postpandemic
world will be like.
Governments face formidable difficulties in their
efforts to safeguard their citizens from the threat of the
Covid-19 pandemic. It is recognized that society’s regular
functioning cannot be maintained, particularly in light
AÇIKGÖZ and GÜNAY / Turk J Med Sci
3184
of the virus’s primary protective measurement, namely
confinement. Furthermore, it is acknowledged that the
imposed measurements will invariably intrude on rights
and freedoms that are an essential feature of a democratic
society ruled by law [25]. Countries have no choice but
to take extraordinary steps to overcome the pandemic’s
unusual situation and save lives, like extensive lockdowns
enacted to slow virus transmission and restrict freedom
of movement. Such policies may have an unintended
impact on people’s lives and security, and access to health
care, food, water, sanitation, work, education, and leisure
[26]. Furthermore, more security measures may damage
democratic principles and fundamental human rights in
communities and countries [6]. As a result, the Covid-19
pandemic might lead to a rise in authoritarianism at the
global level.
Our daily lives have forever changed by the Covid-19
pandemic and the resulting economic crisis. One of the
most notable developments has been the acceleration of the
movement to digital payments, as customers avoided using
cash for fear of spreading the virus, and retailers responded
by moving their operations online. For example, in 2019,
the overall number of noncash payments in the euro area
climbed by 81% to €98 billion from €90.7 billion in 2018,
and card payments accounted for 48% of total noncash
payments in the euro area in 2019 [27]. In total, the global
digital payments industry hit more than $4.7 trillion value
in 2019 and increased to $5.4 trillion value in 2020, almost
a 16% rise compared with the previous year. The entire
sector is expected to continue its impressive growth in
2021, with over $6.6 trillion transaction value. According
to the growth rates in 2020, while Europe was the leading
digital payments market with the 28.3% growth rate to
$1.17 trillion compared to 2019, the US market follows
with $1.26 trillion worth of digital payments, 22.6% more
than the previous year1.
Moreover, a cryptocurrency based on blockchain
technology, such as Bitcoin and Ethereum, has gained
popularity and traction worldwide as a faster and
cheaper way to transmit money across borders during the
pandemic. Notably, demand for Bitcoin has been surging
globally since the beginning of the Covid-19 pandemic.
Despite a significant drop in its value in recent months,
the value of Bitcoin increased by more than 300% in 20202.
Today, especially emerging economies, are increasingly
turning to cryptocurrencies to help them recover from the
1 CPA Practice Advisor (2021). Digital Payments to Hit $6.6 Trillion in 2021, a 40% Jump in Two Years [online]. Website https://www.cpapracticeadvisor.
com/accounting-audit/news/21208440/digital-payments-to-hit-66-trillion-in-2021-a-40-jump-in-two-years [accessed 19 June 2021].
2 Bitcoin.com (2021). Bitcoin Price [online]. Website https://markets.bitcoin.com/crypto/BTC [accessed 10 June 2021].
3 Oxford Business Group (2021). Economic News [online]. Website https://oxfordbusinessgroup.com/news/can-cryptocurrencies-drive-covid-19-
recovery-emerging-markets [accessed 10 June 2021].
4 CNBC (2021). The Fed this summer will take another step in developing a digital currency [online]. Website https://www.cnbc.com/2021/05/20/the-
fed-this-summer-will-take-another-step-ahead-in-developing-a-digital-currency.html [accessed 11 June 2021].
pandemic’s adverse economic impacts3. On the other hand,
central bank digital currency (CBDC) and stablecoins have
received more attention recently. According to the Bank
for International Settlements (BIS) survey, more than
85% of central banks are studying or investigating CBDC;
however, many of issuances have yet to be completed [28].
For example, China, the European Central Bank (ECB)
and the Federal Reserve are working to build CBDC4 [29].
In contrast to the CBDC, stablecoins are private entities
designed to maintain a steady value concerning another
asset like a unit of currency and commodity or a basket
of assets, unlike cryptocurrencies [30]. Even though
the pandemic has highlighted the importance of digital
financial services, digital currencies have raised concerns
about consumer protection, data privacy, potential
cybersecurity risks, disrupting bank lending, and erasing
local liquidity from bank deposits [31].
Countries have used big data to combat the
Covid-19 pandemic, which enhanced the effectiveness
of their efforts in pandemic monitoring, virus tracking,
prevention, control, and treatment, as well as resource
allocation [32]. However, while using big data to fight the
Covid-19 pandemic may improve health care services and
their performance, it might also raise other issues related
to personal data protection. Before the pandemic, the use
of personal data by governments without the permission
of individuals was a point of debate; however, it has now
become a focal point for human rights violations with the
pandemic’s severe measurements.
In line with the situations and discussions mentioned
above, it is crucial to study the impact of the Covid-19
pandemic on the world and Turkish economies, mainly
because of its massive destruction in all areas. While many
studies have examined the global economic consequences
of the Covid-19 pandemic [33, 34, and 35], some of them
investigated the only Turkish economy [36, 37, and 38].
Furthermore, this research might be viewed as an updated
version of our previous work, which examined only the
early stages of the pandemic [6]. Both studies might be
considered complementary, this one investigates the
impact of the pandemic on the global and the Turkish
economies from the beginning to the present in contrast to
the previous one. Hence, the main objective of this study
is to evaluate the potential short-term macroeconomic
impacts of the Covid-19 pandemic on the world and
Turkish economy based on reliable data released by the
AÇIKGÖZ and GÜNAY / Turk J Med Sci
3185
IMF, the OECD, the World Bank (WB), Turkish public
institutions, and other international institutions and
current debate. Analyzing the Covid-19 pandemic impact
on the economies in the medium and long term is not our
priority since it is not known where the pandemic will
evolve.
Nevertheless, every data related to the world economy
could not be reached due to the dynamic structure of the
pandemic since the national data generally has not been
published in the middle of the year. Hence, the international
economic organizations could not access some countries’
data. In this context, our study focuses on average OECD
area data if the world value of the related economic
variable does not exist. Since the OECD economies share
in world GDP is around 50% with 38 member countries5, it
is thought that average OECD data can be used as a proxy
for the world economic data in this study.
As a result, some basic macroeconomic variables such
as growth rate, inflation, interest and unemployment rate,
trade volume, fiscal balance, travel and tourism, health
spending, and fiscal support are presented to analyze the
global and Turkish real sector, financial sector, public
sector, labor market, foreign trade, and travel and tourism
sector developments in the short-term. Furthermore,
based on current data and discussions, some assumptions
are made concerning possible changes in the global and
Turkish economies. Finally, this study is concluded.
2. Global economic costs of the pandemic until mid-2021
The OECD reveals that global output fell by 3.5% in 2020
due to a sharp decline in global economic activity [3].
Despite new virus outbreaks in several economies in the
fourth quarter of 2020, the global economy recovered
faster than projected, and global output remained roughly
better than an estimated 4.2% contraction [39]. It should
be emphasized that sector specialization in different
economies has led to variations in countries’ output
growth rates. Those most have relied on international
travel and tourism sector faced a more considerable GDP
loss in 2020. Global GDP growth is forecast to be 5.8% in
2021 and 4.4% in 2022, with global output exceeding the
prepandemic level (% 2.7) by mid-2021. However, global
income will still be around $3 trillion, roughly equal to the
size of the entire French economy, lower by the end of 2022
than forecast before the crisis [3].
Before the Covid-19 pandemic, the travel and tourism
sector employed 10.6% of the worldwide workforce (334
million), but 62 million jobs were lost, representing a drop
of 18.5% in 2020. In addition, the global GDP contribution
of this sector fell from 10.4% ($9.2 trillion) in 2019 to 5.5%
5 Organisation for Economic Co-operation and Development (2021). OECD share in world GDP stable at around 50% in PPP terms in 2017 [online].
Website https://www.oecd.org/sdd/prices-ppp/oecd-share-in-world-gdp-stable-at-around-50-per-cent-in-ppp-terms-in-2017.htm [accessed 21 June
2021].
($4.7 trillion) in 2020 due to ongoing mobility restrictions.
Hence, the loss of travel and tourism sector was almost $4.5
trillion in 2020 [40]. Furthermore, destinations worldwide
welcomed one billion fewer international arrivals in 2020
than in the previous year. International arrivals dropped
by 74% due to an unprecedented fall in service demand
and widespread travel restrictions. Also, the collapse in
international travel represents an estimated loss of $1.3
trillion in export revenues, which is more than 11 times
the loss recorded during the 2009 global economic crisis
[41]. As of 2019, international visitor expenditure totaled
$1.7 trillion, accounting for 6.8% of total exports. While
domestic visitor expenditure fell by 45%, international
visitor expenditure fell by a staggering 69.4% in 2020 [40].
Government retention plans and reduced hours support
many jobs, but the potential of job losses and contraction
persists without full recovery of the travel and tourism
sector.
The International Civil Aviation Organization (ICAO)
reveals that global passenger traffic declined by 60%, and
the revenue loss was nearly $371 billion in 2020 because
of the widespread lockdowns, border closures, and travel
restrictions worldwide. Moreover, the estimated decline
in total world passengers will be between 44% and 49%,
and approximately $289 to 323 billion loss of revenues of
airlines is projected for the year 2021 compared to 2019
levels [42]. Nevertheless, cargo flights increased 40% in
April 2020 in contrast to the fall in passenger traffic. Also,
air cargo demand continued to outperform pre-Covid-19
pandemic levels, with demand up 9% in February 2021
compared to February 2019 level [42, 43].
The
International
Labour
Organization
(ILO)
announces that 8.8% of global working hours were lost
relative to the fourth quarter of 2019, equivalent to 255
million full-time jobs in 2020, approximately four times
greater than during the global financial crisis in 2009. In
total, there were unprecedented global employment losses
in 2020 of 114 million jobs relative to 2019. In contrast
to previous crises, employment losses in 2020 translated
mainly into rising inactivity rather than unemployment,
leading to an additional 81 million people shifting to
inactivity alongside 33 million additional unemployed.
Hence, the unemployment rate rose by 1.1% points
to 6.5% around the world in 2020 [44]. Similarly, the
unemployment rate increased from 5.4% in 2019 to 7.1%
in 2020 in the OECD countries [3]. Before taking into
account income support measurements, global labour
income in 2020 is estimated to have declined by 8.3%,
which amounts to $3.7 trillion, or 4.4% of global GDP [44].
Overall, world trade volumes are expected to increase
to 8.2% in 2021, after falling by 8.5% in 2020 despite
AÇIKGÖZ and GÜNAY / Turk J Med Sci
3186
ongoing weak services trade due to travel restrictions
and lack of travellers’ confidence [3]. Similarly, global
merchandise trade volume will likely rise by 8% in 2021
after declining by 5.3% in 2020 [45]. In the third quarter of
2020, shipments of computers and electronic components
increased by 11%, while textile shipments increased
by 24%, boosted by demand for face masks and other
protective equipment compared to the second quarter
of 2020. Surprisingly, after growing at a 10% annual rate
in the first half of 2020, global pharmaceutical exports
fell by 1% in the third quarter of 2020, mainly due to
summer stockpiling [5]. Hence, the increased demand
for digital technology and medical supplies has boosted
global trade above prepandemic levels [3]. The better-
than-expected performance at the end of the year can be
attributed in part to the November release of additional
Covid-19 vaccines, which helped to increase business and
consumer confidence. On the other hand, commercial
services exports fell by 20% in 2020 due to international
travel limitations that impeded the delivery of services that
required physical presence or face-to-face interaction [45].
Since oil is a crucial intermediate good, particularly for
manufactured products and the energy sector, fluctuations
in the oil market have a spillover effect. Because of the
general recession of the world economy and the decline
in demand for fuels and gasoline due to travel restrictions,
the Covid-19 pandemic has had a significant impact on
global oil demand. Oil prices fell from $67.3 per barrel in
December 2019 to $18.4 per barrel in April 2020, but with
a steady downward trend in Covid-19 cases in the second
half of 2020, oil prices improved as well, pushing the price
of Brent oil to around $50 by December 2020 [5].
Inflation rate in OECD countries was 1.5% in 2020
and is expected to reach 2.7% for 2021. The inflation rate
is projected to increase significantly due to the past rise
in commodity prices, particularly oil, and some one-off
effects of the crisis. A combination of possible negative
supply-side effects such as higher operating costs due
to containment measurements, disruptions in global
supply chains, and a desire to make up for past losses in
revenues could push up inflation by more than projected.
Upside risks to inflation include further exchange rate
depreciation, and food and energy price increases,
especially in countries where central bank credibility has
already been weakened [3].
Global stock markets have fallen sharply as investors
continue to worry about the broader uncertain economic
impact of the pandemic. The FTSE, Dow Jones Industrial
Average, and the Nikkei all saw massive declines in the
first months of the Covid-19 pandemic. Although the
major Asian and US stock markets recovered following the
announcement of the first vaccine in November, the FTSE
6 BBC (2021). FTSE 100 suffers worst year since financial crisis [online]. Website https://www.bbc.com/news/business-55500103 [accessed 11 June 2021].
is still in negative territory. The FTSE dropped by 14.3% in
2020, its worst performance since 20086.
The rise in fiscal deficits has stemmed primarily from
the collapse in revenues caused by lower economic activity.
In response, central banks in many countries reduced the
interest rates to make borrowing cheaper and encourage
spending to boost the economy. The longer the pandemic
lasts, the greater the challenge is to public finances, and
government deficits and debt have risen to unprecedented
levels. To calm down markets and encourage spending,
central banks have lowered policy rates and purchased
government bonds, thereby, facilitating the fiscal
responses to the pandemic. The size, composition, and
duration of fiscal support have varied across countries,
which has influenced its effectiveness. Among economies,
the majority of supports was devoted to employment
protection and household income support [46]. As a
result, the fiscal balance increased from a deficit of 3.1%
of GDP in 2019 to 10.8% in 2020 in OECD countries.
The estimated deficit value for 2021 is 10.1% due to the
continuing fiscal supports [3]. Global additional spending
and foregone revenues of governments was 9.2% of 2020
GDP [46].
Along with governments, some international financial
institutions provide financial supports to economies to
rapid recovery from the Covid- 19 pandemic. For instance,
the ECB has introduced a Pandemic Emergency Purchase
Programme (PEPP) to support the euro area banks, firms,
and households through the Covid-19 crisis. In this
context, PEPP increased from €500 billion to €1850 billion
in December 2020 [5]. In addition, the IMF is providing
financial assistance and debt service relief to member
countries, facing the economic impact of the Covid-19
pandemic. Overall, the IMF is currently making about
$250 billion, a quarter of its $1 trillion lending capacity,
available to member countries [47]. Furthermore, the WB
has expected to deploy up to $160 billion over 15 months
through June 2021 to support countries’ responses to the
Covid-19 crisis [48].
Around the world, an estimated 400 million people do
not have access to basic health care services. Nearly 100
million individuals are pushed into extreme poverty each
year due to having to pay for their own health care. These
numbers have increased with the Covid-19 pandemic, as
they will continue to rise as people lose jobs and health
insurance costs increase [49]. Before the onset of the
Covid-19 pandemic, average health spending as a share
of GDP across the OECD countries was around 8.8%
[50]. Governments have increased their health spending
with additional spending or foregone revenues during
the Covid-19 pandemic, which is accounted for 1.2% of
2020 GDP. The IMF projected that while average advanced
AÇIKGÖZ and GÜNAY / Turk J Med Sci
3187
economies’ share of health spending in GDP will increase
by 2.6% in between 2020 and 2030, this share will be
estimated to be 0.5% and 0.1% for emerging market and
middle-income countries, and low-income countries,
respectively [46]. Hence, it seems that current inequalities
in health care services worldwide will continue in the
postpandemic period.
Briefly, as shown in Figure, global economic output is
projected to rise by nearly 6% in this year, an impressive
surge after the 3.5% contraction in 2020. While the
unemployment rate in OECD countries is estimated to
fall from 7.1% in 2020 to 6.5% in 2021, there will be no
significant change in the fiscal balance deficit of the OECD
area, and it is expected to be around 10% of GDP level
2021. Nevertheless, world trade volumes are projected to
increase by close to 8.2% in 2021, falling by 8.5% in 2020.
Although inflation in the OECD area declined by 0.4%
point to 1.5% in 2020, it is expected to rise 2.7% in 2021
due to the delayed higher commodity prices [3].
3. Costs of the pandemic on Turkish economy until
mid-2021
Despite the OECD’s prediction of a 1.3% decrease in
Turkish GDP in December 2020, Turkey demonstrated
a more remarkable recovery from the Pandemic in 2020,
with a growth rate of 1.8%. Turkey’s GDP is expected to
rise at a rate of 5.7% in 2021 before slowing to 3.4% in 2022
in the absence of possible future shocks [3]. Turkey’s GDP
increased by 7% in the first quarter of 2021 [51], but the
surge in infections which appears to have peaked in May
2021. Therefore, new confinement measures have affected
employment, incomes, and private consumption from the
second quarter of 2021. The vaccine rollout started fast in
February, but the authorities ran into serious procurement
issues, causing them to scale down their goals and seek
more diverse measurements until June 2021 [3].
Turkey’s overall export and import values were nearly
$169.482 billion and $219.397 billion, respectively,
resulting in a negative trade balance of $49.915 billion in
2020, which is higher than the $29.512 billion negative
trade balance in 2019 [52]. Turkey’s net exports declined
by 7.3% in 2020 but it is expected to increase 2.7% in
2021 [3]. In May 2021, exports increased by 65.50%, and
imports increased by 54% compared to the previous year’s
same month [53]. As of 2020, the top export destinations
of Turkey were Germany (9.4%), the US (6%), and Italy
(4.7%) and the top import origins were China (10.4%),
Germany (9.8%), and Russia (8.1%) [52]. These data show
that the supply chain of Turkey does not depend on one
country primarily, so the Covid-19 pandemic’s detrimental
influence on the manufacturing sector did not persist
long in Turkey. For instance, the manufacturing sector
in Turkey shrank by 8% in the second quarter of 2020,
but then this sector showed a strong rebound in the next
quarter with a 28.6% growth [51]. Therefore, the pandemic
has not hurt the Turkish manufacturing sector as expected
in the end. Turkish current account deficit recorded $1.127
million, indicating a decrease of $1.947 million compared
to June of the 2020, bringing the 12-month rolling deficit
to $29.674 million [54]. Although Turkey had a current
account surplus in 2019 with 0.9% of GDP value, the
Covid-19 pandemic widened the current account deficit to
5.2% in 2020 [3]. However, this deterioration was mainly
driven by a decrease in the goods trade deficit and an
increase in services surplus, which results from the global
economic disruptions caused by the Pandemic.
The tourism sector has been hardly damaged in Turkey
since tourism income fell by 65.1% and declined to $12
billion in 2020 compared to the previous year. In parallel,
departing visitors decreased by 69.5% in 2020 compared to
Source: [3], [58], [74] and [75]
-15.0
-10.0
-5.0
0.0
5.0
10.0
15.0
20.0
2019
2020
2021 projection
2019
2020
2021 projection
OECD Countries
Turkey
Real GDP growth
Unemployment rate
Inflation rate
Fiscal balance
Trade growth
Figure. Economic outlook on OECD Countries and Turkey.
AÇIKGÖZ and GÜNAY / Turk J Med Sci
3188
2019 and declined to nearly 16 million people [55]. In April
2020, global international passenger capacity experienced
an unprecedented 94% reduction for Turkey [42]. The
navigational charge losses of Turkey were approximately
$210 million in 2020, declined from $363 million in 2019 to
$153 million at the end of 2020 [56]. Additional measures
to facilitate a strong tourism season in 2021 summer have
been implemented in Turkey, including a vaccination
process for employees of the travel and tourism sector.
Turkey has struggled with the high unemployment
rate (13.5%) in the first quarter of 2021 since 2018. The
number of unemployed persons increased by 89 thousand
to 4 million 277 thousand persons compared to the
same quarter of the previous year [57]. It seems that the
unemployment rate increased, especially among the
blue-collar workers and service sector employees, due to
the bankruptcies and closures of factories and small and
medium workplaces, who have been probably suffered the
major economic losses of the Covid-19 pandemic. This
situation has led to income losses for workers because of
the weak labour market in Turkey.
Moreover, the Consumer Price Index (CPI) of Turkey
increased by 16.59% annually in May 2021. On the
other hand, transportation with 28.39%, furnishings and
household equipment with 21.79%, and health with 19.30%
were the main groups where high annual increases were
realized [58]. However, the central bank’s management
has reiterated its commitment to the 5% inflation target
against the 16,59% current inflation rate and persisting
inflationary pressures in Turkey [59].
Additionally, unprecedented uncertainty will bring
more risks for investors in Turkey. The value of Turkey’s
Economic Confidence Index, 99.3 in January 2020,
decreased by 3.1 points to 96.2 in January 2021. This
fall stemmed from the decline in service and retail trade
confidence indices [60]. On the other hand, the Turkish
Real Sector Confidence Index increased from 62.30
to 107.40 in April 2021 compared to the same month
in the previous year [61]. Therefore, the data indicate
that increased confidence will bring better business
performance in Turkey soon.
The Central Bank of the Republic of Turkey (CBRT)
decided to keep the policy interest rate at 19% in May 2021
to improve the financial conditions, which is higher than
compared to the same month in the previous year (8.25%)
[62]. Furthermore, on March 31, the CBRT introduced
a program of outright purchases of sovereign bonds and
has substantially increased its liquidity facilities to banks
[63]. Today, Turkey’s 5 Years Credit Default Swaps (CDS)
premium value is high with 382.62 points on 21 June 2021
but lower than 2020’s maximum value of 643.15. This
7 World Government Bonds (2021). Turkey 5 Years CDS [online]. Website http://www.worldgovernmentbonds.com/cds-historical-data/ [accessed 13
June 2021].
can be interpreted as the Turkish economy is under the
pressure of financial risk7 since the higher CDS premium
might cause pressure on the Turkish foreign borrowing
interest rate to rise.
The Turkish authorities predict that the total
discretionary fiscal support package will cost 638 billion
Turkish liras, 12.7% of GDP, in March 2021 to fight the
Covid-19 pandemic. Debt guarantees to businesses and
people, loan service deferrals by public banks, tax deferrals
for businesses, equity injections into public banks, and
a short-term labor program are examples of crucial
fiscal measurements in Turkey [63]. In addition, Turkey
implemented substantial Value-Added Tax (VAT) cuts for
services and a withholding tax reduction for tradespeople.
For instance, the VAT on passenger transportation,
wedding organizations, residential maintenance and repair,
dry cleaning, and tradespeople services like tailoring was
reduced from 18% to 8%. Moreover, until mid-May 2021, a
nationwide prohibition on layoffs was in effect. Besides, for
families with monthly salaries of less than 5.000 Turkish
liras ($740), state lenders have proposed a low-interest
credit package of up to 10.000 Turkish liras ($1.477). The
government also announced that it would pay 60% of the
staff salaries of firms forced out of business under the
short-term employment allowance program. In addition,
the minimum pension was raised to 1.500 Turkish liras
($221) to protect retirees from the pandemic’s harmful
consequences, and bonus payments were pushed to earlier
dates. The government has recently begun paying 1.000
Turkish liras ($148) to 4.4 million needy families [64]. In
the first quarter of 2021, compared to the same quarter the
previous year, employee compensation climbed by 16%,
while net operational surplus/mixed-income increased by
39.1% [57].
Among G20 emerging economies, Turkey provided the
most liquidity support compared to its GDP in response
to the Pandemic. Turkey left behind countries, including
China, Brazil, India, and South Africa, with a liquidity
support to GDP ratio of 9.5%. Brazil followed it with 6.2%
and India with 5.2%. This ratio was 1.5% in Russia, and
1.3% in China. Besides, Turkey’s additional spending and
foregone revenues were 2.7% of 2020 GDP and lag behind
many countries [65].
Public deposit banks began new retail loan campaigns
for home buying and consumer spending in June 2021. Also,
loans to farmers have been postponed for six months. As
part of the government’s Coronavirus Economic Stability
Shield program, the Treasury-Backed Credit Guarantee
Fund increased from 25 billion Turkish liras ($3.67
million) to 50 billion liras ($7.34 million). The enterprises’
principal and interest payments were postponed for at
AÇIKGÖZ and GÜNAY / Turk J Med Sci
3189
least three months, and public lenders refinanced them.
Turkey extended repayment periods for certain credit card
loans, introduced low-interest credit packages for low-
income households, allowed tradespeople to postpone
payments without penalty, provided new low-interest
loans and credit cards with longer repayment periods for
tradespeople, and offered new credit packages for their
jobs [66].
Turkey’s health spending was roughly 4.4% of its GDP
in 2019, lower than the OECD average of 8.8%. However,
health spending in OECD countries increased sharply
in 2020 due to the pandemic. According to preliminary
estimates, health spending in a group of 16 OECD
countries jumped to roughly 9.9% of GDP on average in
2020 [50]. Turkey has also boosted its health spending
through additional spending or revenue foregone during
pandemic, accounting for 0.3% of GDP in 2020. As a
result, IMF estimated that Turkey’s health spending as a
share of GDP would rise 0.5% on average until 2030 [46].
Overall, Turkey is the 19th largest economy globally,
with a GDP of $761 billion [67]. Turkey was among the
few countries to experience positive economic growth in
2020. As shown in Figure, GDP growth is expected to be
strong in 2021 with 5.7% if there will be no unprecedented
shocks. Turkey’s unemployment rate decreased by 0.6%
point to 13.1% in 2020 due to the decline in labor force
participation. It is estimated to increase 14% in 2021 due
to the weak labour market during the Covid-19 pandemic
[3]. Besides, inflation rate increased by 2.8% point to
14.6% in 2020, probably due to inflation expectations and
risk premia. Inflation is projected to rise to 16.5% in 2021
in Turkey [68]. Turkey’s net export volume fell from 3.2%
in 2019 to -7.3% in 2020, and it will be expected to rise to
2.7% in 2021 as in other countries [3]. On the other hand,
Turkey’s fiscal balance deficit rose from 2.9% of GDP in
2019 to 3.4% in 2020. However, the Turkish fiscal balance
will be worsened by an estimated 5% deficit in 2021 [68].
4. Discussion
Estimating the real economic consequences of the
Covid-19 pandemic is currently difficult due to the
Pandemic’s spiral effects on both the national and global
economies as a result of increased trade and financial
linkages brought on by globalization [69]. In general, the
crisis response is organized around four thematic pillars
that are aligned with economies’ comparative advantages:
saving lives threatened by the pandemic, protecting the
poor and vulnerable, assisting in the retention of jobs and
businesses, and working to build a more strong recovery
8 Our World in Data (2021). Statistics and Research: Coronavirus (COVID-19) Vaccinations [online]. Website https://ourworldindata.org/covid-
vaccinations [accessed 16 August 2021].
9 Financial Times (2021). Covid-19 looks like a hinge in history [online]. Website https://www.ft.com/content/de643ae8-9527-11ea-899a-f62a20d54625
[accessed 14 June 2021].
[48]. Countries who were fast to vaccinate their populations
against the Covid-19 pandemic and manage infections
through efficient public health measurements will likely
rebound more quickly. However, while vaccination rates in
many advanced economies increase, poorer and emerging
market countries lag behind [3]. In the short term, global
economic recovery will be impossible if equal vaccination
distribution between and within countries is not achieved.
For example, only 1.2% of people in low-income countries
have gotten at least one dose of the Covid-19 vaccination,
even though 31.2% of the world’s population has received
at least one dose8. This divergence will probably tend to
increase economic inequality within countries in the
short-term. However, this condition might hurt the world’s
social peace in the medium and long-term.
It is crucial to highlight that in the early stages of the
Covid-19 pandemic, the global leader of the US failed to
safeguard its citizens, leaving them unwell and ruined.
Millions of Americans have become impoverished and
unable to access health care services. The Covid-19
pandemic killed nearly as many Americans than all of the
military conflicts of the last 70 years combined9. Despite
warnings and considerable advantages, including vast
resources, biomedical infrastructures, and scientific skills,
the US missed every opportunity to contain the virus. In
contrast to many other countries, refused to take effective
measurements to reverse the virus’s upward trend. This
delay has affected the American economy badly, and the
US economy contracted by 3.5% in 2020 [3]. According
to the US experience, acting quickly is likely the most
critical lesson in determining national outcomes during
pandemics.
Although
increasing
vaccine
production
and
distribution is the most substantial current economic
policy for boosting economic development, the future of
the global economy remains uncertain, and the recovery
will be uneven. According to the IMF’s economic
predictions, between 90 million and 130 million full-time
equivalent jobs will be lost, with global output growth of
about 6% in 2021. However, this estimation is subject to
vary based on virus evaluations around the world. Thus,
the picture is highly unpredictable, with both upside and
downside risks making forecasting difficult. Therefore,
rebuilding confidence in macroeconomic policy and
structural reforms are vital for rebounding the national
and world economy [70].
The Covid-19 pandemic demonstrates the importance
of public health management. However, many nations’
health systems are overburdened, and expenditures will
AÇIKGÖZ and GÜNAY / Turk J Med Sci
3190
be required to improve staff and health care capacity
to manage the possibility for Covid-19 renewal and
subsequent pandemics. Governments all over the world, on
the other hand, have promised billions of dollars towards
a Covid-19 vaccine and treatment options; as a result,
the market shares of some pharmaceutical companies
involved in vaccine development have increased [71]. As
a result, it is not incorrect to suggest that pharmaceutical
corporations have benefited from the Covid-19 pandemic
at most. In addition, several countries supply vials,
syringes, needles, and even cool boxes and freezers needed
to manufacture, distribute, and administrate vaccines.
Hence, mass production, distribution and administering
the vaccine supply chain are essential as vaccination [3].
ILO reveals the contrast between massive job losses
in hard-hit sectors such as accommodation and food
services, arts and culture, retail, and construction and
the positive job growth evident in many high-skilled
services sectors such as information and communications
technology (ICT), and financial and insurance activities
[44]. The World Economic Forum’s Future of Jobs Report
2020 projected that technological change is set to displace
a range of skills in the labour market while driving greater
demand for a new set of core skills such as analytical
thinking, creativity, critical thinking, and digital skills
[72]. Because of a particular shortfall in digital skills,
there is a significant additional disruption in the labour
market due to the Covid-19 crisis. Also, only 53.6% of
the global population uses the internet, which refers to
the digital divide around the world. Hence, the impact of
the pandemic should serve as a wake-up call for countries
that need to embrace the digitalization process, incentivize
companies to move towards digital business models, and
invest more in ICT development and digital skills [73].
The Covid-19 pandemic measurements also disrupted
global education worldwide, and over 1.6 billion students
in more than 190 countries were out of school in months.
Two-thirds of an academic year has been missed on
average worldwide due to full or partial closures in
months, while half of the world’s student population
is still affected by closures. The long-term closure of
schools poses a great risk for students’ future since they
will probably fall below the minimum proficiency levels
in both theory and practice. Besides, 24 million children
and youth are at risk of dropping out due to the digital
divide. Since digital transformation is closely associated
with human capital development, more education funding
is needed for education recovery in most countries to
improve online education services and upgrade curricula
for changing world [74]. Also, more alignment is needed
between employers and educators to enable students to get
the new digital economy skills. However, these reforms
will probably bring additional economic costs to the fiscal
budget of countries.
For many economies, currently, debt affordability is
not a risk, but it will be inevitable due to the high amount
of income support and government support packages.
Vaccine supply brings an extra economic burden to the
budget of countries. These additional spendings will bring
more taxation on consumers and producers in the future.
This situation is likely to deepen the economic crisis in
the medium and long-term. Today, international financial
institutions like IMF and WB provide financial assistance
and debt service relief to member countries facing the
economic impact of the Covid-19 pandemic. Nevertheless,
whether these aids are given equal or transparent between
countries is a separate debate.
There will inevitably be paradigm shifts in the economy
in the postpandemic period. In contrast to the free market
economy, public and private sector cooperations will
be essential for economic activities since the Covid-19
pandemic shows that neither governments nor businesses
can achieve this complex postpandemic economic
transformation alone. In addition, it seems the well-
balanced fiscal support and flexible approach need to the
economic recovery with the cooperation of the public and
private sectors [75].
Today, global viruses will be seen as a threat like a
nuclear attack, biological weapon, or global terrorism
since their impacts on humans and the world are similarly
very destructive, as seen in the Covid-19 pandemic [6].
Furthermore, the pandemic has brought to light another
potential global threat, named bioterrorism, because the
virus’s terrible impact, which these terrorist groups have
acknowledged, has renewed their interest in acquiring,
manufacturing, and employing biological weapons [76].
For this reason, countries that could develop a biological
defense system or infrastructure for the virus attack will
have more power in the postpandemic world.
5. Conclusion
Although there were variations in recovery across
economies, Turkey was among the few countries with
China that showed a solid and positive rebound with 1.8%
economic growth during the Covid-19 pandemic in 2020
though the global output declined by 3.5%. Hence, Turkey
is apart from its peers due to the rapid recovery resulting
from initial better policy responses such as monetary and
credit expansion and large liquidity support. From this
perspective, it is expected that the world and Turkey’s GDP
growth will be 5.8% and 5.7%, respectively, in 2021 without
further major shocks. Nevertheless, new confinement
measures taken in May 2021 due to the pandemic’s third
wave might adversely affect employment, incomes,
private consumption, and the travel and tourism sector in
Turkey. This unprecedented uncertainty characteristic of
the Covid-19 pandemic has likely changed all economic
AÇIKGÖZ and GÜNAY / Turk J Med Sci
3191
predictions, and sustainable growth for Turkey might not
be possible in the short-term.
Global economic output is projected to rise by nearly
6% in 2021, an impressive surge after the 3.5% contraction
in 2020, but there will be no ordinary recovery in the
world due to the unprecedented economic uncertainty.
Also, economic rebound depends on the effectiveness of
vaccination programs, public health policies, fiscal polices,
and the country’s dependence on a hard-hit sector such
as travel and tourism in the short-term. Although the
unemployment rate and fiscal balance deficit are projected
to fall in 2021 in OECD countries, world trade volumes are
estimated to rise. Besides, inflation is expected to increase
in the OECD area due to the past rise in commodity and
oil prices, and some one-off effects of the crisis.
In Turkey, as in other countries, the economic impact
of the Covid-19 pandemic has been severe. Turkey’s
unemployment rate decreased due to the decline in labor
force participation but it is estimated to increase due to
the weak labour market during the Covid-19 pandemic.
Besides, poverty in Turkey is estimated to have risen by
about 1.5 million people as a result of the pandemic [68].
Moreover, the inflation rate has increased, most probably
due to the inflation expectations and risk premia. It is
obvious that Turkey’s unemployment and inflation rate
displayed a different trend in 2020 compared to the OECD
economies. Although the tourism sector has been hardly
damaged in Turkey, a sharp increase in tourism revenues
is expected in 2021 once mass vaccination occurs. As of
August 2021, the share of people fully vaccinated against
Covid-19 virus is 39%, and the share of people only partly
vaccinated against Covid-19 virus is 13% in Turkey, and
Turkey ranked 17th in the world10.
Although Turkey’s fiscal balance deficit rose, it showed
a better performance than other OECD countries. Turkey
provided the most liquidity support among G20 emerging
economies during the pandemic. However, the Turkish
fiscal balance will be worsened because of the extension
of fiscal supports. Currently, the actual costs of the fiscal
policy supports and vaccination to the Turkish economy
have not been yet precise.
10 Our World in Data (2021). Statistics and Research: Coronavirus (COVID-19) Vaccinations [online]. Website https://ourworldindata.org/covid-
vaccinations [accessed 16 August 2021].
Today, vaccines are seen as the key to a safe and
permanent transition to more typical economic and
social conditions. Countries that have been quick to
vaccinate their population against the Covid-19 virus
and manage to control infections through effective
public health strategies have likely displayed more quick
recovery. Hence, it seems that the pandemic will provide
some economic and political opportunities to some
countries if they can manage to control the virus earlier
by effective vaccination policy.
As a result of the digital transformation in
every industry, the world will not be the same in
the postpandemic period. Remote working, video-
conferencing,
online
entertainment,
e-commerce,
and digital currencies, for example, have all grown in
popularity since the beginning of the pandemic. Hence,
countries should focus on the digitalization process,
dissemination of ICT, and digital skills. Supporting
students and young people to remain in education and
enabling them to acquire digital skills by upgrading their
curriculum based on new labour market demand should
be a priority for governments. Countries should take
urgent measurements to prevent a youth unemployment
crisis and promote better mental health for young people.
Moreover, cooperation between public and private
sectors will probably be a new economic approach rather
than a free market.
In sum, Turkey must concentrate on new structural
reforms to boost the economy, invest in health care, adapt
to the new postpandemic world, and achieve a sustainable
economy along with the other countries. Overall, trends
in other economies suggest that the cost of the Covid-19
pandemic will not be recovered just by attaining a
vaccination threshold in the short-term. In other words,
vaccination seems to be a beginning rather than an end,
and the policy choices governments make today will
determine their place in the new world. Nevertheless,
global economic recovery will be impossible in the near
future if equal vaccination distribution between and
within countries is not achieved.
References
1.
World Health Organization. WHO Coronavirus (Covid-19)
Dashboard. Geneva; 13 August 2021.
2.
Gopinath G. The great lockdown: Worst economic downturn
since the Great Depression. International Monetary Fund.
Washington; 14 April 2020.
3.
Organisation for Economic Co-operation and Development.
OECD Economic Outlook. Paris; May 2021.
4.
Lowenstein R. History repeating. Wall Street Journal. New
York; 14 January 2015.
5.
Committee for the Coordination of Statistical Activities. How
Covid-19 is changing the world: a statistical perspective. New
York; 30 March 2021.
AÇIKGÖZ and GÜNAY / Turk J Med Sci
3192
6.
Açıkgöz Ö, Günay A. The early impact of the Covid-19 Pandemic
on the global and Turkish economy. Turkish Journal of Medical
Sciences 2020; 50: 520-526. doi:10.3906/sag-2004-6
7.
Barro RJ, Ursua JF, Weng J. The coronavirus and the great
influenza pandemic: Lessons from the ‘Spanish Flu’ for the
coronavirus’s potential effects on mortality and economic
activity. NBER Working Paper 2020; 26866. doi: 10.3386/w26866
8.
Zoppi L. How does the Covid-19 Pandemic compare to other
pandemics?. News Medical Life Sciences. Manchester; 16 March
2021.
9.
Garrett TA. Economic effects of the 1918 influenza pandemic
implications for a modern-day pandemic. Federal Reserve Bank
of St. Louis. Missouri; November 2007.
10.
Niall F. 1918, 1957, 2020: Big pandemics and their economic,
social and political consequences. Hoover Institution. California;
20 May 2020.
11.
Altig D, Baker SR, Barrero JM, Bloom N, Bunn P et al. Economic
uncertainty before and during the Covid-19 Pandemic. Journal
of Public Economics 2020; 191 (104274). doi: 10.1016/j.
jpubeco.2020.104274
12.
World Economic Forum. From Covid-19 to Brexit, this is how
uncertainty affects the global economy. Geneva; 25 January 2021.
13.
Fauci AS, Clifford LH, Redfield RR. Covid-19 navigating the
uncharted. New England Journal of Medicine 2020; 382:1268-
1269.
14.
Koirala A, Joo YJ, Khatami A, Chiu C, Britton PN. Vaccines for
Covid-19: The current state of play. Pediatric Respiratory Review
2020; 35: 43-49.
15.
Anderson RM, Heersterbeek H, Klinkenberg D, Hollingsworth
TD. How will country-based mitigation measures influence the
course of the covid-19 epidemic?. The Lancet 2020; 395 (10228):
931-934. doi: 10.1016/S0140-6736(20)30567-5
16.
Atkeson A. What will be the economic impact of Covid-19 in the
US? Rough estimates of disease scenarios. NBER Working Paper
2020; 26867. doi: 10.3386/w26867
17.
Baqaee D, Farhi E, Mina MJ, Stock JH. Policies for a second wave.
Brooking Papers on Economic Activity Conference Draft; 25
June 2020. https://brook.gs/3f2Al4z
18.
Barrero J, Bloom N, Davis SJ. Covid-19 and labour reallocation:
Evidence from the US. VoxEU. London; 14 July 2020.
19.
Jorda O, Singh SR, Taylor AM. Longer-run economic
consequences of pandemics. Federal Reserve Bank of San
Francisco Working Paper 2020; 09. doi: 10.24148/wp2020-09
20.
Duina F. Op-Ed: Free market ideology won’t help us during
the Covid-19 crisis. Here’s what would. Los Angles Times. Los
Angles; 18 May 2020.
21.
Internation Labor Organization. ILO Brief: Social Protection
Spotlight. Geneva; 23 April 2020.
22.
Organisation for Economic Co-operation and Development.
Tackling the mental health impact of the Covid-19 crisis: An
integrated, whole-of-society response. OECD Policy Responses
to Coronavirus (Covid-19). Paris; May 2021.
23.
World Economic Forum. The loneliness pandemic? How
Covid-19 has created a mental health crisis. Geneva; 9 February
2021.
24.
Santini Z, Koyanagi A. Loneliness and its association with
depressed mood, anxiety symptoms, and sleep problems in
Europe during the Covid-19 Pandemic. Acta Neuropsychiatrica
2021; 33(3):160-163. doi:10.1017/neu.2020.48
25.
Council of Europe. Respecting democracy, rule of law and
human rights in the framework of the Covid-19 sanitary crisis.
Information Documents. Strazburg; 7 April 2020.
26.
Guterres A. We are all in this Together: Human rights and
Covid-19 response and recovery. United Nations. New York;
23 April 2020.
27.
European Central Bank. Press Release: 2019 Payments
statistics. Frankfurt; 11 September 2020.
28.
Codruta B, Andreas W. Ready, steady, go? – Results of the
third BIS survey on central bank digital currency. Bank for
International Settlements Papers 2021; 114.
29.
European Central Bank. Report on a Digital Euro. Frankfurt;
October 2020.
30.
World Economic Forum. Digital currency governance
gonsortium: Vision for 2021 deliverables. Geneva; January 202.
31.
World Economic Forum. International cooperation and the
era of digital currency growth. Geneva; 5 May 2021.
32.
Alsunaidi SJ, Almuhaideb AM, Ibrahim NM, Shaikh FS,
Alqudaihi KS et al. Applications of big data analytics to control
Covid-19 Pandemic. Sensors 2021; 21, 2282. doi:10.3390/
s21072282
33.
Shlomo M, Barzani E. Global economic impact of Covid-19: a
summary of research. Samuel Neaman Institute. Haifa; March
2020.
34.
Warwick M, Fernando R. The economic impact of Covid-19.
In: Baldwin R, Mauro BW, editors. Economics in the time of
Covid-19. London, UK: Centre for Economic Policy Research
(CEPR) Press; 2020. pp. 45-51.
35.
Jackson JK, Weiss MA, Schwarzenberg AB, Nelson RM.
Congressional Research Service Report: Global economic
effects
of
Covid-19.
Congressional
Research
Service.
Washington; 9 July 2021.
36.
Çakmaklı C, Demiralp S, Yeşiltaş S, Yıldırım MA. An evaluation
of the Turkish economy during Covid-19. Centre for Applied
Turkey Studies’ Working Paper 2020; 01.
37.
Öztürk Ö, Şişman MY, Uslu H, Çıtak F. Effect of Covid-19
outbreak on Turkish stock market: a sectoral-level analysis.
Hitit University Journal of Social Sciences Institute 2020; 13(1):
56- 68: doi: 10.17218.hititsosbil.728146
38.
Voyvoda E, Yeldan E. A general equilibrium analysis of the
impact of the Covid-19 outbreak on Turkey’s economy and a
policy alternative to protect labor incomes. Political Economy
Research Institute’s Working Paper 2020; 518.
39.
The
Organisation
for
Economic
Co-operation
and
Development. OECD Economic Outlook. Paris; December
2020.
AÇIKGÖZ and GÜNAY / Turk J Med Sci
3193
40.
World Travel&Tourism Council. Press Release: Economic
Impact Reports. London; 25 March 2021.
41.
World Tourism Organization. Press Release 2020: Worst Year
in Tourism History with 1 Billion Fewer International Arrivals.
Madrid; 28 January 2021.
42.
International Civil Aviation Organization. Report on
Economic Impacts Covid-19: ICAO Economic Impact Analysis
of Covid-19 on Civil Aviation. Montreal; 9 June 2021.
43.
International Air Transport Assosiation. Economic Reports:
Air Cargo Market Analysis. Montreal; February 2021.
44.
International Labour Organization. Covid-19 and the World
of Work: ILO Monitor 7th ed. Geneva; January 2021.
45.
World Trade Organization. Press Release: Trade Statistics and
Outlook. Geneva; 31 March 2021.
46.
International Monetary Fund. Fiscal Monitor: A Fair Shot.
Washington; April 2021.
47.
International Monetary Fund. Covid-19 Financial Assistance
and Debt Service Relief. Washington; 27 May 2021.
48.
World Bank. World Bank Group Covid-19 Crisis Response
Approach Paper. Washington; June 2020.
49.
World Economic Forum. Davos agenda: What you need to
know about the future of global health. Geneva; 24 January
2021.
50.
Organisation for Economic Co-operation and Development.
OECD Health Statistics 2021. Paris; July 2021.
51.
The Turkish Statistical Institute. Press Release: Quarterly Gross
Domestic Product, Quarter I: January-March, 2021. Ankara;
31 May 2021.
52.
Republic of Turkey Ministry of Trade. Data Bulletin: January
2021. Ankara; 2 February 2021.
53.
Republic of Turkey Ministry of Trade. Data Bulletin: May 2021.
Ankara; 2 June 2021.
54.
Central Bank of the Republic of Turkey. Balance of Payments
Statistics. Ankara; June 2021.
55.
The Turkish Statistical Institute. Press Release: Household
Domestic Tourism, Quarter IV: October-December and
Annual, 2020. Ankara; 22 April 2021.
56.
International Civil Aviation Organization. ICAO Covid-19 Air
Traffic Dashboard. Montreal; 9 June 2021.
57.
The Turkish Statistical Institute. Press Release: Labour Force
Statistics, April 2021. Ankara; 10 June 2021.
58.
The Turkish Statistical Institute. Press Release: Consumer Price
Index, May 2021. Ankara; 3 June 2021.
59.
Central Bank of the Republic of Turkey. Inflation Targets.
Ankara; 10 June 2021.
60. The Turkish Statistical Institute. Press Release: Economic
Confidence Index, January 2021. Ankara; 28 January 2021.
61.
Central Bank of the Republic of Turkey. Business Tendency
Statistics and Real Sector Confidence Index. Ankara; June
2021.
62.
Central Bank of the Republic of Turkey. Summary of the
Monetary Policy Committe Decision. 6 May 2021.
63.
International Monetary Fund. Policy Responses to Covid 19:
Turkey. Washington; 3 June 2021.
64.
Çicek D. Turkey: Government keeps economy on track amid
Covid-19. Anadolu Agency. Ankara; 11 March 2021.
65.
International Monetary Fund. Fiscal Monitor Database of
Country Fiscal Measures in Response to the COVID-19
Pandemic. Washington; July 2021.
66.
Gönültaş B, Kurtaran G, Gökkoyun SC. Turkey outpaces G20
in liquidity support amid Covid-19. Anadolu Agency. Ankara;
1 May 2021.
67.
World Bank. World Bank Open Data: Gross Domestic
Production (current US$). Washington; June 2021.
68.
International Monetary Fund. IMF Country Report on Turkey
No. 21/110. Washington; June 2021.
69.
Lee JW, McKibben WJ. Globalization and Disease: The Case of
SARS. Cambridge MA, MIT Press; 2004.
70.
International Monetary Fund. World Economic Outlook
Reports. Washington; April 2021.
71.
Jones L, Palumbo D, Brown D. Coronavirus: How the
pandemic has changed the world economy. BBC News.
London; 24 January 2021.
72.
World Economic Forum. The Future of Jobs Report 2020.
Geneva; 20 October 2020.
73.
World Economic Forum. The Global Competitiveness Report
2020. Geneva; 16 December 2020.
74.
United
Nations
Educational,
Scientific
and
Cultural
Organization. Update education curricula and expand
investment in the skills needed for jobs in markets of tomorrow.
Paris; 19 March 2021.
75.
World Bank. 6 ways to ensure a fair and inclusive economic
recovery from Covid-19. Washington; 2 June 2021.
76.
Dass RAS. Bioterrorism: Lessons from the Covid-19 Pandemic.
Counter Terrorist Trends and Analyses 2021; 13 (2):16-23.
| Short-term impact of the Covid-19 pandemic on the global and Turkish economy. | 12-17-2021 | Açikgöz, Ömer,Günay, Asli | eng |
PMC6655540 | RESEARCH ARTICLE
Sprint mechanical variables in elite athletes:
Are force-velocity profiles sport specific or
individual?
Thomas A. Haugen1*, Felix Breitscha¨del1,2, Stephen Seiler3
1 Norwegian Olympic Federation, Oslo, Norway, 2 Norwegian University of Science and Technology,
Trondheim, Norway, 3 Faculty of Health and Sport Sciences, University of Agder, Kristiansand, Norway
* [email protected]
Abstract
Purpose
The main aim of this investigation was to quantify differences in sprint mechanical variables
across sports and within each sport. Secondary aims were to quantify sex differences and
relationships among the variables.
Methods
In this cross-sectional study of elite athletes, 235 women (23 ± 5 y and 65 ± 7 kg) and 431
men (23 ± 4 y and 80 ± 12 kg) from 23 different sports (including 128 medalists from World
Championships and/or Olympic Games) were tested in a 40-m sprint at the Norwegian
Olympic Training Center between 1995 and 2018. These were pre-existing data from quar-
terly or semi-annual testing that the athletes performed for training purposes. Anthropomet-
ric and speed-time sprint data were used to calculate the theoretical maximal velocity,
horizontal force, horizontal power, slope of the force-velocity relationship, maximal ratio of
force, and index of force application technique.
Results
Substantial differences in mechanical profiles were observed across sports. Athletes in
sports in which sprinting ability is an important predictor of success (e.g., athletics sprinting,
jumping and bobsleigh) produced the highest values for most variables, whereas athletes in
sports in which sprinting ability is not as important tended to produce substantially lower val-
ues. The sex differences ranged from small to large, depending on variable of interest.
Although most of the variables were strongly associated with 10- and 40-m sprint time, con-
siderable individual differences in sprint mechanical variables were observed among equally
performing athletes.
Conclusions
Our data from a large sample of elite athletes tested under identical conditions provides a
holistic picture of the force-velocity-power profile continuum in athletes. The data indicate
PLOS ONE | https://doi.org/10.1371/journal.pone.0215551
July 24, 2019
1 / 14
a1111111111
a1111111111
a1111111111
a1111111111
a1111111111
OPEN ACCESS
Citation: Haugen TA, Breitscha¨del F, Seiler S
(2019) Sprint mechanical variables in elite athletes:
Are force-velocity profiles sport specific or
individual? PLoS ONE 14(7): e0215551. https://doi.
org/10.1371/journal.pone.0215551
Editor: Leonardo A. Peyre´-Tartaruga, Universidade
Federal do Rio Grande do Sul, BRAZIL
Received: January 3, 2019
Accepted: April 3, 2019
Published: July 24, 2019
Copyright: © 2019 Haugen et al. This is an open
access article distributed under the terms of the
Creative Commons Attribution License, which
permits unrestricted use, distribution, and
reproduction in any medium, provided the original
author and source are credited.
Data Availability Statement: The data underlying
this study are deposited at the UiA Open Research
Data repository with the following persistent ID:
https://dataverse.no/dataset.xhtml?persistentId=
doi:10.18710/PJONBM. Researchers interested in
verification/replication of the present study can use
this data source and are also kindly requested to
contact the senior author, Stephen Seiler at
[email protected], or Monica Klungland
Torstveit at [email protected]. The data
underlying this study are owned by the Norwegian
Olympic Federation (NOF). Because there are some
unique athletes in the dataset, and their results may
that sprint mechanical variables are more individual than sport specific. The values pre-
sented in this study could be used by coaches to develop interventions that optimize the
training stimulus to the individual athlete.
Introduction
Running a short distance as fast as possible is a core capacity in many sports. For a sprinter
competing in athletics, 100 m and 200 m, this capability alone defines them as performers. In
bobsleigh, athletes are required to sprint while moving an external mass. Sprinting capacity is
also crucial in most team sports, as the ability to either create or close small gaps can be deci-
sive in goal scoring situations. Even in typical endurance sports, explosive acceleration ability
(in the context of their slow-twitch dominant peers) can be a medal-winning advantage at the
finish of a close race. Accordingly, numerous sprint training studies across a wide range of
sports have been performed over the years. Sprinting under assisted, resisted and normal con-
ditions, maximal and explosive strength training, plyometric training and high-intensity run-
ning have all been investigated in different combinations [1–3]. Although the principle of
specificity is clearly present, no training methods have so far emerged as superior. Individual
predispositions must therefore be considered when prescribing training programs.
Considerable effort has been made over the years to quantify the underlying variables for
sprint performance. Seminal works by Fenn & Marsh [4] described the force-velocity (Fv) rela-
tionship in isolated frog and cat muscles, a relationship that later was confirmed in humans by
Wilkie [5]. Advances in technology have allowed scientists to explore fundamental aspects of
sprinting skills more closely, and presently, the physiology and mechanics of sprint running
are typically interrogated through macroscopic mechanical variables [6–8]. Samozino et al. [9]
have recently developed a simple and practical method for profiling the mechanical capabilities
of the neuromuscular system using an inverse dynamic approach applied to the centre-of-
mass movement, calculating the step-averaged ground reaction forces in runners’ sagittal
plane of motion during accelerated sprinting from only anthropometric and spatiotemporal
data. Theoretical maximal velocity (v0), horizontal force (F0), horizontal power (Pmax) and
force-velocity profile (i.e., the slope of the force-velocity relationship; SFV) can be calculated
from the speed-time curve. Other indices, such as ratio of force (RF) and index of force appli-
cation technique (DRF) can also be computed from the same method. RF is a ratio of the step-
averaged horizontal component of the ground reaction force to the corresponding resultant
force, while DRF expresses the athlete’s ability to maintain a net horizontal force production
despite increasing velocity throughout accelerated sprinting [6]. These variables are determi-
nant factors for sprint performance, in line with the laws of motion, and provide insights into
individual biomechanical limitations [6–8, 10].
A promising application of force-velocity profiling is in the design of individualized sprint
training programs. An effective sprint training program should target the main factors that
limit the athlete’s performance [11–12]. To help tailor the training program to the individual,
the coach could compare test values from the individual to test values that are typical for the
sport. An athlete with a velocity value that is low for the sport could then be prescribed more
maximal velocity sprinting, whereas an athlete with a horizontal force value that is low for the
sport could be prescribed more horizontal strength training [11]. Currently, sprint force-veloc-
ity profile data are only available from athletes in a few selected sports; previous studies only
analyzed specialist sprinters or athletes from selected team sports [8, 10, 13–19]. It is unclear
Sprint mechanical variables in athletes
PLOS ONE | https://doi.org/10.1371/journal.pone.0215551
July 24, 2019
2 / 14
identify them, NOF has determined that the
underlying data cannot be shared. Data are
therefore available only on request to Stephen.
[email protected] for researchers who meet the criteria
for access to confidential data. The authors confirm
that data would be made available to researchers
interested in replication/verification of the present
study.
Funding: The authors received no specific funding
for this work.
Competing interests: The authors have declared
that no competing interests exist.
whether previously measured sprint force-velocity profiles are specific to the sport or specific
to the athlete. To individualize a training program for an athlete in a given sport, the coach
requires a holistic picture of the force-velocity profile continuum in athletes. Therefore, the
aim of the present study was to quantify differences in sprint mechanical variables across
many sports and within each sport. Secondary aims were to quantify sex differences and rela-
tionships among the variables.
Materials and methods
Participants
The Norwegian Olympic training centre is a standardized testing facility used by a large num-
ber of elite athletes from many sports. A database of results from 40-m sprint tests was col-
lected from 1995 to 2018, and this database provides a foundation for exploring sprint
performance and mechanical properties in athletes. In this cross-sectional study we analyzed
sprint tests by 666 athletes from 23 sports. All participants were Norwegian national team ath-
letes, i.e., represented Norway in international senior competitions, and 128 of the athletes
were medalists from the World Championships and/or Olympic Games.
Ethics statement
This study was based on pre-existing data from quarterly or semi-annual testing that the ath-
letes performed for training purposes, and thus no informed consent was obtained [20]. All
data were anonymized to comply with the General Data Protection Regulations of the Euro-
pean Union. The study was reviewed by the Norwegian Data Protection Authority and
approved by the ethics committee at the Faculty of Health and Sport Sciences, University of
Agder (reference number 19/00068).
Procedures and data handling
All included players were tested in the time period 1995–2018. Tests were performed on a ded-
icated indoor 40-m track with 8 mm Mondotrack FTS surface (Mondo, Conshohocken, USA)
at the Norwegian Olympic training center in Oslo. A standard warm-up program was com-
pleted prior to sprint testing, beginning with a 10–15 min easy jog, followed by 5–6 minutes of
sprint specific drill exercises, 2–4 strides with increasing speed and 1–2 trial starts. During test-
ing, athletes assumed the starting position and started running on their own initiative after
being cleared to start by the test leader. New trials were performed every 3–5 min until a per-
formance plateau was observed. In practice, 80% of all players achieved their best performance
within two trials. Body mass was assessed immediately prior to or after the sprint test on a sta-
tionary force platform (AMTI, model OR6-5-1). Data from a single athlete was only included
in one category for each analysis. That category was the player’s affiliation on the day of his/
her best sprint test result. A purpose-built excel spreadsheet developed by Morin & Samozino
[21] formed basis for calculations of F0, v0, Pmax, SFV, RFmax and DRF. These calculations were
based on best individual sprint test, associated split times and body mass. Temperature and
atmospheric pressure were set to 760 mm Hg and 20 ˚C.
Two different timing system setups were used. In the time epoch 1995–2011, a 60x60 cm
start pad was placed under the track at the start line. The clock was initiated when the front
foot stepped off the pad. Split times were recorded with split-beamed timing gates for each
10th m of the sprints. The transmitters where placed 140 cm above ground level, while the
reflectors were placed 130 and 150 cm above the floor. Both beams had to be interrupted to
trigger each timing gate. The timing setup has been assessed for accuracy and reliability [22].
Sprint mechanical variables in athletes
PLOS ONE | https://doi.org/10.1371/journal.pone.0215551
July 24, 2019
3 / 14
In January 2011, the timing system was upgraded. The split-beamed timing gates were
replaced by dual-beamed timing gates, while the start pad was replaced with a single-beamed
timing gate located 60 cm in front of the start line and 50 cm above ground level. Rigorous
pilot testing was performed before deciding the exact location of the timing gate at start to pro-
vide comparable times with the previous setup. Simultaneous comparisons (n = 50) of the old
and new timing setup revealed no differences in 40-m sprint times (mean ± SD: 0.00 ± 0.02 s).
The dual-beamed timing system has also been assessed for accuracy and reliability [23]. Over-
all, typical error (coefficient of variation; CV) was 0.6–2.4% for sprint times, ~1.5% for v0 and
RFmax, and 3.5–5.1% for F0, Pmax, SFV and DRF for both timing setups.
To ensure valid sprint mechanical values when using split times as input in the method pro-
posed by Samozino et al. [9], it is crucial that i) the entire acceleration phase is captured, and
ii) time initiation (the time 0) is very close to the first rise of the force production onto the
ground [24]. For the current procedures, the body’s center-of-mass was ~ 0.5 m in front of the
start line, and possessed a considerable forward momentum, at time triggering. Hence, based
on available correction factors [22, 25], 0.5 s was added to all sprint times for converting to
“first movement” triggering. All sprint times presented are comparable to starts from blocks
and audio signal with reaction time subtracted from the total time.
Statistical analysis
Data are reported as mean ± SD. Pearson’s R (±90%CL) was used to examine the relationship
across variables (after log transformation of physiological data). Correlation values were inter-
preted categorically according to the scale outlined by Hopkins et al. [26], meaning that 0.10,
0.30, 0.50, 0.70, 0.90 and 1.0 were thresholds for small, moderate, large, very large, extremely
large and perfect, respectively. Magnitudes of differences across category means were assessed
by standardization (mean difference divided by the harmonic mean of the SD of the compared
groups). The thresholds for assessing the observed difference in means were 0.2, 0.6, 1.2, 2.0
and 4.0 for small, moderate, large, very large and extremely large, respectively [26]. To make
inferences about true values of effects, non-clinical magnitude-based inference rather than
null-hypothesis significance testing was used [26, 27]. Magnitudes were evaluated mechanisti-
cally: if the confidence interval overlapped substantial positive and negative values, the effect
was deemed unclear; otherwise effects were deemed clear and shown with the probability that
the true effect was substantial using the following scale: 25–75%, possibly; 75–95%, likely; 95–
99.5%, very likely; > 99.5%, most likely [26, 27]. A purpose-built excel spreadsheet for combin-
ing outcomes from several subject groups was used to calculate effect magnitudes, confidence
limits (CL) and inferences [28]. Women’s performance was defined as 100% for comparisons
between male and female athletes (all sport disciplines pooled together).
Results
To keep the results within reasonable limits, only a summary of the results is presented in this
section. However, additional comparisons across category means can be performed by insert-
ing data from the supplementary file into Hopkins’ spreadsheet [28].
Table 1 shows age, body mass and sprint performance across the analyzed sports. Overall,
athletics sprinting (hereafter referred to as “sprinting” or “sprinters”) produced the fastest
sprint times over both the shortest (10 m) and longest (40 m) distances among males, clearly
ahead of bobsleigh (mean difference, ±90%CL: 0.02, ±0.03 and 0.09, ±0.09 s; possibly and
likely; moderate), athletics jumping (hereafter referred to as “jumping” or “jumpers”) (0.07,
±0.03 and 0.15, ±0.09 s; most likely and very likely; large), soccer (0.11, ±0.02 and 0.38, ±0.07 s;
most likely; very large) and all other sports (0.13, ±0.03 to 0.24, ±0.03 and 0.45, ±0.09 to 0.88,
Sprint mechanical variables in athletes
PLOS ONE | https://doi.org/10.1371/journal.pone.0215551
July 24, 2019
4 / 14
±0.09 s; most likely; very large to extremely large). Sprinters also displayed the fastest sprint
times over the same distances among women, clearly ahead of jumpers (0.10, ±0.04 and 0.28,
±0.15 s; most likely and very likely; very large), handball (0.15, ±0.03 and 0.52, ±0.11 s; most
likely; very large), athletics throwing (0.16, ±0.04 and 0.57, ±0.13 s; most likely; very large) and
Table 1. Mean values (± SD) of age, body mass and sprint performance in Norwegian national team athletes (n = 666).
Discipline (sex)
N
Age
BM
10 m
20 m
30 m
40 m
(y)
(kg)
(s)
(s)
(s)
(s)
Alpine skiing (W)
10
22.6 ± 3.3
66.0 ± 4.2
2.19 ± 0.05
3.62 ± 0.10
4.95 ± 0.15
6.29 ± 0.19
Alpine skiing (M)
13
24.7 ± 3.8
84.4 ± 3.7
2.04 ± 0.05
3.30 ± 0.08
4.48 ± 0.10
5.64 ± 0.13
Athletics jumping (W)
8
20.4 ± 4.9
60.4 ± 2.6
2.10 ± 0.06
3.40 ± 0.10
4.60 ± 0.15
5.79 ± 0.19
Athletics jumping (M)
9
21.8 ± 3.6
77.4 ± 6.8
1.97 ± 0.05
3.16 ± 0.06
4.23 ± 0.09
5.28 ± 0.11
Athletics sprinting (W)
5
19.2 ± 3.0
58.4 ± 2.1
2.00 ± 0.02
3.24 ± 0.04
4.39 ± 0.05
5.51 ± 0.09
Athletics sprinting (M)
8
20.7 ± 3.2
71.8 ± 3.8
1.90 ± 0.03
3.05 ± 0.05
4.09 ± 0.08
5.13 ± 0.11
Athletics throwing (W)
10
20.3 ± 3.6
75.0 ± 6.6
2.16 ± 0.06
3.53 ± 0.09
4.81 ± 0.14
6.08 ± 0.20
Athletics throwing (M)
14
22.4 ± 4.7
100.1 ± 8.8
2.03 ± 0.07
3.30 ± 0.11
4.45 ± 0.15
5.58 ± 0.20
Bandy (W)
13
23.0 ± 5.4
64.1 ± 8.6
2.28 ± 0.07
3.76 ± 0.12
5.18 ± 0.16
6.63 ± 0.21
Bandy (M)
23
19.3 ± 2.7
77.4 ± 8.3
2.09 ± 0.06
3.39 ± 0.09
4.60 ± 0.12
5.80 ± 0.16
Basketball (M)
10
22.6 ± 3.3
86.2 ± 9.5
2.06 ± 0.07
3.36 ± 0.12
4.55 ± 0.17
5.74 ± 0.22
Beach-/volleyball (M)
23
24.9 ± 4.7
87.7 ± 7.7
2.04 ± 0.05
3.34 ± 0.09
4.56 ± 0.13
5.78 ± 0.16
Bobsleigh (M)
9
26.7 ± 1.9
92.8 ± 4.4
1.92 ± 0.03
3.10 ± 0.06
4.17 ± 0.07
5.22 ± 0.09
Combat sports (W)
17
23.7 ± 6.0
60.6 ± 5.7
2.30 ± 0.07
3.78 ± 0.12
5.20 ± 0.19
6.62 ± 0.26
Combat sports (M)
32
22.5 ± 4.2
74.8 ± 11.1
2.08 ± 0.07
3.38 ± 0.11
4.59 ± 0.16
5.80 ± 0.23
Cross-country skiing (W)
8
20.0 ± 3.4
59.9 ± 5.1
2.27 ± 0.10
3.73 ± 0.19
5.12 ± 0.28
6.51 ± 0.37
Cross-country skiing (M)
15
21.9 ± 3.4
74.2 ± 4.4
2.11 ± 0.10
3.44 ± 0.16
4.68 ± 0.23
5.88 ± 0.31
Fencing (W)
5
18.9 ± 1.0
64.6 ± 4.0
2.34 ± 0.06
3.86 ± 0.11
5.30 ± 0.18
6.74 ± 0.25
Fencing (M)
10
21.9 ± 2.0
77.1 ± 6.6
2.14 ± 0.04
3.50 ± 0.07
4.76 ± 0.10
6.01 ± 0.13
Handball (W)
32
25.8 ± 4.6
72.8 ± 6.1
2.15 ± 0.07
3.50 ± 0.13
4.77 ± 0.18
6.03 ± 0.24
Handball (M)
18
23.9 ± 3.6
92.7 ± 9.0
2.03 ± 0.04
3.27 ± 0.07
4.43 ± 0.10
5.58 ± 0.14
Ice hockey (M)
34
24.8 ± 4.6
88.7 ± 7.4
2.03 ± 0.06
3.30 ± 0.10
4.46 ± 0.14
5.62 ± 0.19
Mogul skiing (W)
5
19.4 ± 1.9
64.0 ± 9.1
2.18 ± 0.04
3.57 ± 0.10
4.88 ± 0.16
6.19 ± 0.22
Mogul skiing (M)
14
21.2 ± 3.1
72.9 ± 6.3
2.05 ± 0.04
3.32 ± 0.05
4.52 ± 0.09
5.67 ± 0.12
Nordic combined (M)
22
23.5 ± 4.2
69.6 ± 4.0
2.04 ± 0.05
3.33 ± 0.08
4.51 ± 0.11
5.69 ± 0.16
Ski jumping (W)
11
18.4 ± 4.1
56.3 ± 3.1
2.23 ± 0.04
3.68 ± 0.06
5.05 ± 0.11
6.45 ± 0.15
Ski jumping (M)
28
21.2 ± 3.5
64.3 ± 5.0
2.05 ± 0.07
3.34 ± 0.12
4.55 ± 0.18
5.75 ± 0.25
Snowboard (W)
5
21.7 ± 4.6
59.4 ± 3.4
2.24 ± 0.06
3.73 ± 0.11
5.13 ± 0.17
6.61 ± 0.24
Snowboard (M)
9
21.3 ± 3.1
78.5 ± 7.7
2.05 ± 0.05
3.34 ± 0.09
4.55 ± 0.11
5.76 ± 0.16
Soccer (W)
93
23.8 ± 3.9
64.0 ± 4.9
2.17 ± 0.06
3.55 ± 0.11
4.84 ± 0.16
6.12 ± 0.22
Soccer (M)
57
25.4 ± 4.3
78.7 ± 5.8
2.01 ± 0.05
3.24 ± 0.08
4.39 ± 0.12
5.51 ± 0.16
Speed skating (W)
12
21.4 ± 4.0
68.5 ± 5.8
2.20 ± 0.05
3.60 ± 0.08
4.92 ± 0.12
6.25 ± 0.17
Speed skating (M)
22
22.3 ± 3.7
78.6 ± 8.0
2.10 ± 0.06
3.39 ± 0.12
4.59 ± 0.17
5.78 ± 0.23
Table tennis (M)
13
21.1 ± 4.1
69.5 ± 8.8
2.12 ± 0.06
3.46 ± 0.13
4.71 ± 0.19
5.95 ± 0.27
Telemark skiing (W)
5
18.6 ± 1.7
62.0 ± 1.9
2.26 ± 0.13
3.72 ± 0.23
5.10 ± 0.35
6.49 ± 0.49
Telemark skiing (M)
13
23.3 ± 2.6
82.7 ± 6.3
2.08 ± 0.06
3.39 ± 0.08
4.60 ± 0.11
5.80 ± 0.14
Tennis (W)
7
17.5 ± 1.7
65.6 ± 3.3
2.25 ± 0.07
3.70 ± 0.12
5.08 ± 0.18
6.48 ± 0.24
Tennis (M)
11
20.8 ± 3.3
75.3 ± 4.8
2.07 ± 0.05
3.37 ± 0.07
4.57 ± 0.10
5.78 ± 0.13
Weight-/powerlifting (M)
13
20.6 ± 4.2
87.9 ± 22.2
2.12 ± 0.07
3.45 ± 0.10
4.71 ± 0.17
5.96 ± 0.22
W = women, M = men, BM = body mass.
https://doi.org/10.1371/journal.pone.0215551.t001
Sprint mechanical variables in athletes
PLOS ONE | https://doi.org/10.1371/journal.pone.0215551
July 24, 2019
5 / 14
all other sports (0.17, ±0.02 to 0.34, ±0.05 and 0.61, ±0.13 to 1.23, ±0.25 s; most likely; very
large to extremely large). The mean sex difference increased from 6.4% for 10-m sprints to
9.3% for 40-m sprints.
Fig 1A shows F0 across sports. Bobsleigh and sprinting displayed the greatest F0 values
among men (unclear difference between them), clearly ahead of volleyball/beach volleyball
(0.4, ±0.3 to 0.5, ±0.3 Nkg-1; likely to very likely; moderate), snowboard (0.4, ±0.4 to 0.5, ±0.4
Nkg-1; likely to very likely; moderate), soccer (0.5, ±0.3 to 0.6, ±0.3 Nkg-1; very likely to most
likely; moderate to large) and all other sports (0.5, ±0.3 to 1.3, ±0.3; very likely to most likely;
moderate to very large). In women, sprinters exhibited the highest F0 values, clearly ahead of
jumping (0.7, ±0.4 Nkg-1; most likely; very large), handball (0.8, ±0.3 Nkg-1; most likely; very
large), snowboard (0.9, ±0.4 Nkg-1; very likely; very large), alpine skiing (0.9, ±0.3 Nkg-1;
most likely; very large) and all other sports (0.9, ±0.3 Nkg-1 to 1.7, ±0.3 Nkg-1; most likely;
very large to extremely large). The sex difference for F0 was 9.3, ±1.2% (most likely; moderate).
Fig 1B shows v0 across sports. Sprinters showed the highest values among men, clearly
ahead of jumpers (0.4, ±0.4 ms-1; likely; moderate), bobsleigh (0.5, ±0.3 ms-1; very likely;
large), soccer (1.1, ±0.3 ms-1; most likely; very large) and all other male sports (1.2, ±0.3 to 2.1,
±0.3 ms-1; most likely; very large to extremely large). Sprinters also displayed superior values
among women, clearly better than jumpers (0.4, ±0.3 ms-1; very likely; moderate), handball
(1.0, ±0.2 ms-1; most likely; very large), athletics throwing (1.1, ±0.2 ms-1; most likely; very
large) and all other female sports (1.1, ±0.2 to 2.0, ±0.4 ms-1; most likely; very large to
extremely large). The sex difference for v0 was 11.9, ±1.1% (most likely; large).
Fig 2A shows Pmax across sports. In men, sprinters obtained the highest values, clearly
ahead of bobsleigh (0.9, ±1.0 Wkg-1; likely; moderate), jumping (2.2, ±1.1 Wkg-1; most likely;
large), soccer (3.7, ±0.8 Wkg-1; most likely; very large) and all other male sports (4.2, ±0.9 to
7.2, ±0.9 Wkg-1; most likely; very large to extremely large). Sprinting athletes also displayed
the highest Pmax values among women, clearly ahead of jumping (2.4, ±1.1 Wkg-1; most likely;
very large), handball (3.6, ±0.8 Wkg-1; most likely; very large), throwers (4.0, ±0.9 Wkg-1;
most likely; extremely large) and all other female sports (4.2, ±0.7 to 7.3, ±0.8 Wkg-1; most
likely; extremely large). The sex difference for Pmax was 21.9, ±1.1% (most likely; large).
Fig 2B shows SFV across sports. Jumpers produced the highest values (i.e., most velocity-ori-
ented) among the males, ahead of sprinting specialists (0.02, ±0.04 Ns-1m-1kg-1; unclear;
small), speed skating (0.03, ±0.04 Ns-1m-1kg-1; possibly; small) and all other male sports
(0.06, ±0.04 to 0.16, ±0.03 Ns-1m-1kg-1; very likely to most likely; moderate to very large. At
the other end, volleyball/beach volleyball and snowboard were the most force-oriented disci-
plines (unclear difference between them), showing clearly lower SFV values than weight-/
powerlifting (0.03, ±0.04 to 0.04, ±0.04 Ns-1m-1kg-1; likely: small to moderate) and all other
male sports (0.03, ±0.04 to 0.16, ±0.03 Ns-1m-1kg-1; likely to most likely; small to very large).
Among female athletes, jumpers displayed the most velocity-based SFV values, clearly higher
than athletic sprinting (0.04, ±0.05 Ns-1m-1kg-1; likely; moderate) and all other female sports
(0.05, ±0.04 to 0.16, ±0.08 Ns-1m-1kg-1; likely to most likely; moderate to very large). Snow-
board was the most force-oriented group, ahead of bandy (0.05, ±0.07 Ns-1m-1kg-1; unclear;
moderate), ski jumping (0.05, ±0.08 Ns-1m-1kg-1; unclear; moderate) and all other female
sports (0.06, ±0.07 to 0.16, ±0.08 Ns-1m-1kg-1; likely to most likely; moderate to very large).
The sex difference for SFV was 2.4, ±0.7% (most likely; small).
Fig 3A shows RFmax across sports. Sprinters produced the highest percentage values among
men, ahead of bobsleigh (0.4, ±0.8%; unclear; small), jumping (1.4, ±0.8%; very likely; large)
and all other male sports (2.1, ±0.7 to 4.6, ±0.7%; most likely; large to extremely large). Sprint-
ers also displayed the highest values among women, clearly ahead of jumpers (1.7, ±1.1%; very
Sprint mechanical variables in athletes
PLOS ONE | https://doi.org/10.1371/journal.pone.0215551
July 24, 2019
6 / 14
Fig 1. Maximal horizontal force (F0) (Panel A) and theoretical maximal velocity (v0) (Panel B) across sports. The sports are
ranked according to mean values for men.
https://doi.org/10.1371/journal.pone.0215551.g001
Fig 2. Maximal horizontal power (Pmax) (Panel A) and force-velocity slope (SFV) (Panel B) across sports. The sports are ranked
according to mean values for men.
https://doi.org/10.1371/journal.pone.0215551.g002
Sprint mechanical variables in athletes
PLOS ONE | https://doi.org/10.1371/journal.pone.0215551
July 24, 2019
7 / 14
likely; large) and all other sports (2.9, ±0.9 to 5.9, ±1.2%; most likely; very large to extremely
large).
Fig 3B shows DRF across sports. Jumpers displayed the highest values, ahead of sprinters
(0.2, ±0.4%; unclear; small), speed skating (0.4, ±0.4%; likely; moderate) and all other male
sports (0.7, ±0.4 to 1.5, ±0.3%; very likely to most likely; moderate to very large). Among
females, jumpers also obtained the highest values, clearly ahead of sprinters (0.2, ±0.4%;
unclear; small), handball (0.5, ±0.3%; very likely; moderate) and all other female sports (0.5,
±0.3 to 1.5, ±0.7%; very likely to most likely; moderate to very large).
Table 2 shows correlations (±90%CL) across the analyzed variables. The correlations
between sprint mechanical variables and sprint times ranged from trivial to perfect. The corre-
lations between sprint performance and SFV/DRF increased with increasing sprint distance.
Fig 3. Maximal ratio of force (RFmax) (Panel A) and index of force application technique (DRF) (Panel B) across sports. The
sports are ranked according to mean values for men.
https://doi.org/10.1371/journal.pone.0215551.g003
Table 2. Correlations (±90%CL) across analyzed variables.
10-m time
40-m time
F0
v0
Pmax
SFV
RFmax
40-m time
0.96, ±0.01
F0
-0.89, ±0.01
-0.73, ±0.03
v0
-0.87, ±0.02
-0.97, ±0.01
0.57, ±0.04
Pmax
-1.00, ±0.01
-0.97, ±0.01
0.88, ±0.02
0.89, ±0.01
SFV
-0.02, ±0.06
-0.30, ±0.06
-0.43, ±0.05
0.50, ±0.05
0.05, ±0.06
RFmax
-1.00, ±0.01
-0.96, ±0.01
0.89, ±0.01
0.88, ±0.02
1.00, ±0.01
0.03, ±0.06
DRF
-0.15, ±0.06
-0.42, ±0.05
-0.30, ±0.06
0.62, ±0.04
0.19, ±0.06
0.99, ±0.01
0.17, ±0.06
F0 = maximal horizontal force (relative to body mass), v0 = theoretical maximal velocity, Pmax = maximal horizontal power (relative to body mass), SFV = force-velocity
slope (relative to body mass), RFmax = ratio of force, DRF = index of force application technique.
https://doi.org/10.1371/journal.pone.0215551.t002
Sprint mechanical variables in athletes
PLOS ONE | https://doi.org/10.1371/journal.pone.0215551
July 24, 2019
8 / 14
Discussion
To our knowledge, this is the first study to explore and compare underlying physiological and
mechanical variables of sprint performance across a wide range of sports. Up to very large and
even extremely large differences in sprint mechanical variables were observed across sports.
Overall, sports in which sprinting ability is an important predictor of success scored the high-
est values for most variables, while sports involving other locomotion modalities than running
tended to produce substantially lower values. The current data from a large sample of elite ath-
letes tested under identical conditions provides a holistic picture of the Fv profile continuum
in sprinting athletes. In the following paragraphs, we will discuss each of the analysed variables
more in detail.
F0 in the present sample was in the range 7–10 Nkg-1 for men and 6–9 Nkg-1 for women,
corresponding to a mean sex difference of 9.3%. Athletic sprinters and bobsleigh contestants
were at the upper end of the scale. Previously published studies have shown that world-class
male and female sprinters can reach 11 and 10 Nkg-1, respectively [10], representing the cur-
rent upper limits for horizontal force production relative to body mass during accelerated
sprinting. However, F0, calculated with the simple method outlined by Samozino et al. [9], is
larger during resisted sprinting compared to unloaded sprints [14, 15, 29]. Hence, F0 derived
from normal sprinting appears not to be a true F0, as the resistance in overcoming body mass
inertia appears insufficient for maximal horizontal force-capacity generation. For practition-
ers, the importance of F0 is perfectly illustrated by the fact that bobsleigh athletes displayed
slightly higher F0-values (and clearly higher body mass) than sprinters despite slightly poorer
sprint times across all time splits. Thus, F0 is a particularly crucial measure for athletes who
perform brief sprints while moving an external mass. The shorter the distance considered, the
higher the correlation between sprint performance and F0 (see Table 2).
Cross et al. [17] reported 8.5 ± 1.3 and 8.8 ± 0.4 Nkg-1 for elite rugby union forwards and
backs, and 8.1 ± 0.8 and 8.2 ± 1.0 Nkg-1 for corresponding rugby league players. These F0-val-
ues are on par with the present male soccer players. Interestingly, although rugby players are
generally heavier than soccer players, they do not produce higher F0 when normalized for
body mass. As an athlete gets heavier, the energy cost of accelerating that mass also increases,
as does the aerodynamic drag associated with pushing that wider frontal area through the air
[30]. “Bigger” is therefore not necessarily better for sprinting, at least when there is no external
mass to push. Moreover, volleyball/beach volleyball were among the best sports in terms of F0
scores, while weight-/powerlifters produced clearly lower values, despite no substantial group
mean differences in body mass. This confirms previous findings that vertically-oriented and
heavy strength training of the lower limbs does not necessarily translate to higher horizontal
force production during accelerated sprinting [31].
The correlation between v0 and sprint performance was very large for 10-m sprint, and the
correlation values increased with increasing sprint distance (see Table 2). V0 was in the range
7.5–11 ms-1 for men and 6–9.5 ms-1 for women, equivalent to a mean sex difference of 11.9%.
Not surprisingly, sprinters obtained the highest scores. For comparison, the world’s fastest
male and female track sprinters reach peak velocities of ~12 and ~11 ms-1, respectively [10,
32]. The fastest male and female team sport athletes in our material approached/exceeded 10
and 9 ms-1, respectively, in line with previous reports [33–35]. Many practitioners would
argue that elite wide receivers, running backs and/or cornerbacks in American football are
even faster, but no studies to date have presented comparable data.
Metabolic energy turnover and efficient transfer to external power output underlies success-
ful performance in many sports. We observed perfect or extremely large correlations between
Pmax and sprint performance, depending on distance (Table 2). Sprint time improvement is
Sprint mechanical variables in athletes
PLOS ONE | https://doi.org/10.1371/journal.pone.0215551
July 24, 2019
9 / 14
not a linear function of power increase. Indeed, the change in velocity (Δv) is related to the
cube of the change in power (ΔP3), such that a 5% increase in velocity would require a nearly
16% increase in power [36]. The present data are consistent with this relationship. We
observed Pmax values in the range 13–25 Wkg-1 for men and 11–21 Wkg-1 for women. The
mean sex difference observed (21.9%), based on pooled data of all disciplines, corresponds well
with data for elite sprinters. Slawinski et al. [10] reported that Pmax in male and female world-
class sprinters was 30.3 ± 2.5 and 24.5 ± 4.2 Wkg-1, respectively, typically attained after ~1 s of
sprinting. The highest individual values were 36.1 Wkg-1 and 29.3 Wkg-1, representing cur-
rent upper limits in humans [37]. Athletes achieve two-to-three times higher Wkg-1 during
countermovement jump (CMJ) compared to sprinting [37]. However, because it is not possible
to assess a generic anaerobic maximal power, each anaerobic power test must be treated sepa-
rately, and comparisons of power values across modalities are meaningless.
SFV reflects the athlete’s individual balance between force and velocity capabilities. Morin &
Samozino [11] have suggested that athletes with velocity values lower for the sport could be
prescribed more maximal velocity sprinting, whereas athletes with horizontal force values
lower for the sport could be prescribed more horizontal strength training. The present results
show that SFV ranged from -0.75 to -1.10 Ns-1m-1kg-1 for men and -0.80 to -1.15 Ns-1m-
1kg-1 for women, corresponding to a sex difference of 2.4%. The group mean values of
national-level sprinters were similar to the SFV values observed in world-class sprinters [10]. In
the current study, women displayed lower slope values than men in all the analyzed sports
where both sexes were represented. If we follow the approach proposed by Morin & Samozino
[11], men should generally perform more force-oriented training than women. To our knowl-
edge, no previous studies have recommended differentiated sprint-training programs accord-
ing to sex. While young and untrained individuals tend to show improvements irrespective of
training methods [38], well-trained senior athletes only achieve annual improvements smaller
than typical variation [39].
Sprinting distance must also be considered if training prescription should be based on SFV
orientation. The correlation between sprint performance and SFV increased (towards more
velocity-oriented FV-profile) with increasing sprint distance (Table 2). This suggests that the
longer the sprint distance, the more velocity-oriented training should be prescribed. Force-ori-
ented sprint training (e.g., resisted sprints) is likely more appropriate for sports where the ath-
letes are required to perform brief sprints while moving an external mass (e.g., bobsleigh).
However, macroscopic Fv profiles derived from the simple method provide limited informa-
tion about the Fv relationship of the individual muscles involved. The fascicle shortening
velocity of the different muscles engaged do not necessarily change with increasing running
velocity, and this inconsistent relationship is explained by an augmented contribution from
elastic properties with increasing running velocity [40, 41]. Hence, running velocity is not a
proxy for muscle contraction velocity, and one cannot use Fv profiles derived from sprint tests
to determine training prescriptions for muscles in isolation.
RFmax reflects the proportion of the total force production that is directed in the forward
direction of motion at sprint start [9]. We observed a perfect correlation between RFmax and
Pmax (r = 1.0, Table 2). This is mechanically sound, as Pmax in the simple model corresponds to
the peak of the power curve (i.e., the maximal product of horizontal force and velocity), while
vertical force corresponds to body mass when averaged over one step. Hence, Pmax and RFmax
are two measures of the same capability. Within our material, RFmax ranged from 41 to 52% in
men and 37 to 48% in women. Rabita et al. [8] reported 71.6 ± 2.6% in male world-class sprint-
ers, but these values are not directly comparable due to methodological differences. In the
Rabita et al. paper [8], RF was computed from force platform data, where the y-intercept of the
extrapolated linear RF-velocity curve was defined as RFmax, that is; the theoretical maximal
Sprint mechanical variables in athletes
PLOS ONE | https://doi.org/10.1371/journal.pone.0215551
July 24, 2019
10 / 14
contribution of anteroposterior force to the total force produced over one contact phase at
zero velocity. For the simple method, where mechanical variables are calculated from anthro-
pometric and spatiotemporal data, RFmax does not correspond to the extrapolated value at
zero velocity, but to the value at 0.5 s (corresponding to the RF-value approximately at the first
step). Based on publicly available split time and anthropometric data of world-class sprinters
[42], we calculated RFmax values at 0.5 s as high as 56–57% in men and 52–53% in women,
using the simple model. However, the maximal possible value of RFmax (100%) is not optimal
for sprinting because a certain amount of vertical force is required to work against gravity.
DRF expresses the athletes’ ability to maintain a net horizontal force production despite
increasing running velocity. The more negative the slope, the faster the loss of net horizontal
force during acceleration, and vice versa. In the present dataset, the values ranged from -7 to
-10.5% among the men and -7.5 to -11% among the women. For comparisons, Rabita et al. [8]
reported −6.4 ± 0.3% for male world-class sprinters. In practical terms, DRF reflects the dis-
tance over which athletes are able to accelerate (i.e., distance to peak velocity). Previous
research has shown that the duration of the acceleration phase varies as a function of athlete
performance level. Team sport athletes [33, 34], students [43] and prepubescent children [44]
typically achieve peak velocity at ~ 25–30 m of maximal linear sprinting. National and interna-
tional 100-m track sprinters attain peak velocity after 40–50 and 50–80 m of sprinting, respec-
tively, but men peak ~20% further in distance than women [10, 32, 45]. The nearly perfect
correlation between DRF and SFV (r = 0.99, Table 2) is logical and expected, as the most veloc-
ity-oriented athletes are able to accelerate over a longer distance than their more force-ori-
ented counterparts.
Conclusion
In the present study, substantial differences in sprint mechanical properties were observed
across sports. Based on these findings, some may argue that the chronic practice of an activity
induces different Fv profiles in sprint running over time. However, the large spread within
each discipline, in addition to the large overlap across sports, indicate that such variables are
more individual than sport specific. Most sprint mechanical variables are strongly correlated
with sprint performance level, in line with the laws of motion. Indeed, when split times and
anthropometric data form basis for calculations of multiple variables, it is reasonable to expect
high correlations among them. Based on these considerations, practitioners may question the
practical relevance of such variables, as they are entwined, and in some cases, mere ‘different
explanations of the same story’. However, while split time data provide a basis for convenient
analysis on the field, sprint mechanical variables may provide deeper insights into individual
biomechanical limitations. The values presented here can be used by practitioners to develop
individual training interventions.
Supporting information
S1 File. Excel File containing sample size, group means and SD for all variables across
sports.
(DOCX)
Author Contributions
Conceptualization: Thomas A. Haugen, Stephen Seiler.
Data curation: Thomas A. Haugen, Felix Breitscha¨del.
Sprint mechanical variables in athletes
PLOS ONE | https://doi.org/10.1371/journal.pone.0215551
July 24, 2019
11 / 14
Formal analysis: Thomas A. Haugen, Felix Breitscha¨del.
Investigation: Thomas A. Haugen, Felix Breitscha¨del, Stephen Seiler.
Methodology: Thomas A. Haugen, Felix Breitscha¨del, Stephen Seiler.
Project administration: Thomas A. Haugen, Stephen Seiler.
Software: Felix Breitscha¨del.
Supervision: Thomas A. Haugen, Stephen Seiler.
Writing – original draft: Thomas A. Haugen, Stephen Seiler.
Writing – review & editing: Felix Breitscha¨del, Stephen Seiler.
References
1.
Rumpf MC, Lockie RG, Cronin JB, Jalilvand F. Effect of different sprint training methods on sprint per-
formance over various distances: A brief review. J Strength Cond Res. 2016; 30(6):1767–85. https://
doi.org/10.1519/JSC.0000000000001245 PMID: 26492101
2.
Haugen T, Tønnessen E, Hisdal J, Seiler S. The role and development of sprinting speed in soccer.
Brief review. Int J Sports Physiol Perform. 2014; 9(3): 432–41. https://doi.org/10.1123/ijspp.2013-0121
PMID: 23982902
3.
Petrakos G, Morin JB, Egan B. Resisted sled sprint training to improve sprint performance: a systematic
review. Sports Med. 2016; 46(3):381–400. https://doi.org/10.1007/s40279-015-0422-8 PMID:
26553497
4.
Fenn WO, Marsh BS. Muscular force at different speeds of shortening. J Physiol. 1935; 85(3): 277–97.
https://doi.org/10.1113/jphysiol.1935.sp003318 PMID: 16994712
5.
Wilkie DR. The relation between force and velocity in human muscle. J Physiol. 1950; 110(3–4):249–
80.
6.
Morin JB, Edouard P, Samozino P. Technical ability of force application as a determinant factor of sprint
performance. Med Sci Sports Exerc. 2011; 43(9):1680–8. https://doi.org/10.1249/MSS.
0b013e318216ea37 PMID: 21364480
7.
Morin JB, Bourdin M, Edouard P, Peyrot N, Samozino P, Lacour JR. Mechanical determinants of 100-m
sprint running performance. Eur J Appl Physiol. 2012; 112(11): 3921–30. https://doi.org/10.1007/
s00421-012-2379-8 PMID: 22422028
8.
Rabita G, Dorel S, Slawinski J, Sàez-de-Villarreal E, Couturier A, Samozino P, et al. Sprint mechanics
in world-class athletes: a new insight into the limits of human locomotion. Scand J Med Sci Sports.
2015; 25(5):583–94. https://doi.org/10.1111/sms.12389 PMID: 25640466
9.
Samozino P, Rabita G, Dorel S, Slawinski J, Peyrot N, Saez de Villarreal E, et al. A simple method for
measuring power, force, velocity properties, and mechanical effectiveness in sprint running. Scand J
Med Sci Sports. 2016; 26(6):648–58. https://doi.org/10.1111/sms.12490 PMID: 25996964
10.
Slawinski J, Termoz N, Rabita G, Guilhem G, Dorel S, Morin JB, et al. How 100-m event analyses
improve our understanding of world-class men’s and women’s sprint performance. Scand J Med Sci
Sports. 2017; 27(1):45–54. https://doi.org/10.1111/sms.12627 PMID: 26644061
11.
Morin JB, Samozino P. Interpreting power-force-velocity profiles for individualised and specific training.
Int J Sports Physiol Perform. 2016; 11(2):267–72. https://doi.org/10.1123/ijspp.2015-0638 PMID:
26694658
12.
Cross MR, Brughelli M, Samozino P, Morin JB. Methods of power-force-velocity profiling during sprint
running: A narrative review. Sports Med. 2017; 47(7):1255–69. https://doi.org/10.1007/s40279-016-
0653-3 PMID: 27896682
13.
Rakovic E, Paulsen G, Helland C, Eriksrud O, Haugen T. The effect of individualised sprint training in
elite female team sport athletes: A pilot study. J Sports Sci. 2017; 9: 1–7. https://doi.org/10.1080/
02640414.2018.1474536 PMID: 29741443
14.
Monte A, Nardello F, Zamparo P. Sled Towing: The optimal overload for peak power production. Int J
Sports Physiol Perform. 2017; 12(8):1052–1058. https://doi.org/10.1123/ijspp.2016-0602 PMID:
27967284
15.
Pantoja PD, Carvalho AR, Ribas LR, Peyre´-Tartaruga LA. Effect of weighted sled towing on sprinting
effectiveness, power and force-velocity relationship. PLoS One 2018; 13(10):e0204473. https://doi.org/
10.1371/journal.pone.0204473 PMID: 30289907 eCollection 2018.
Sprint mechanical variables in athletes
PLOS ONE | https://doi.org/10.1371/journal.pone.0215551
July 24, 2019
12 / 14
16.
Buchheit M, Samozino P, Glynn JA, Michael BS, Al Haddad H, Mendez-Villanueva A, et al. Mechanical
determinants of acceleration and maximal sprinting speed in highly trained young soccer players. J
Sports Sci. 2014; 32(20):1906–13. https://doi.org/10.1080/02640414.2014.965191 PMID: 25356503
17.
Cross MR, Brughelli M, Brown SR, Samozino P, Gill ND, Cronin JB, et al. Mechanical properties of
sprinting in elite rugby union and rugby league. Int J Sports Physiol Perform. 2015; 10(6):695–702.
PMID: 25310279
18.
Nagahara R, Morin JB, Koido M. Impairment of sprint mechanical properties in an actual soccer match:
A pilot study. Int J Sports Physiol Perform. 2016; 11(7):893–8. https://doi.org/10.1123/ijspp.2015-0567
PMID: 26791405
19.
Morin JB, Petrakos G, Jime´nez-Reyes P, Brown SR, Samozino P, Cross MR. Very-heavy sled training
for improving horizontal-force output in soccer players. Int J Sports Physiol Perform. 2017; 12(6):840–4.
https://doi.org/10.1123/ijspp.2016-0444 PMID: 27834560
20.
Winter EM, Maughan RJ. Requirements for ethics approvals. J Sports Sci. 2009; 27(10): 985. https://
doi.org/10.1080/02640410903178344 PMID: 19847681
21.
Morin JB, Samozino P. Spreadsheet for sprint acceleration force-velocity-power profiling. Research-
Gate 2017. https://www.researchgate.net/publication/321767606_Spreadsheet_for_Sprint_
acceleration_force-velocity-power_profiling?_sg=CuBBw_XwgEAtdCkL8QKaMMLUEFzmLp
EIkMDsHU8dJoYTgIEc2ajruKZRlXrNB7njaQPI-HyYZ5SJiJ5I9ViryIwEvCxzpweYnPHHcVjJ.FNC82J-
TAKYvWmcYRn5itbWhbTST_WIyHM9Z7VPv_bAQ-yvAm9WMnmdEyYOb7DccGEWA3g1_
GmcC3KlwRiQERQ
22.
Haugen T, Tønnessen E, Seiler S. The difference is in the start: impact of timing and start procedure on
sprint running performance. J Strength Cond Res. 2012: 26(2):473–9. https://doi.org/10.1519/JSC.
0b013e318226030b PMID: 22233797
23.
Haugen T, Tønnessen E, Svendsen I, Seiler S. Sprint time differences between single and dual beamed
timing systems. Technical report. J Strength Cond Res. 2014; 28(8): 2376–9. https://doi.org/10.1519/
JSC.0000000000000415 PMID: 24531428
24.
Haugen TA, Breitscha¨del F, Samozino P. Power-force-velocity profiling of sprinting athletes: methodo-
logical and practical considerations when using timing gates. J Strength Cond Res. 2018. https://doi.
org/10.1519/JSC.0000000000002890 PMID: 30273283 (forthcoming).
25.
Haugen T, Buchheit M. Sprint running performance monitoring: methodological and practical consider-
ations. Sports Med. 2016; 46(5):641–56. https://doi.org/10.1007/s40279-015-0446-0 PMID: 26660758
26.
Hopkins WG, Marshall SW, Batterham AM, Hanin J. Progressive statistics for studies in sports medicine
and exercise science. Med Sci Sports Exerc. 2009; 41(1):3–13. https://doi.org/10.1249/MSS.
0b013e31818cb278 PMID: 19092709
27.
Hopkins WG, Batterham AM. The vindication of magnitude-based inference. Sportsci. 2018; 22:19–29.
28.
Hopkins WG. A spreadsheet for combining outcomes from several subject groups. Sportsci. 2006;
10:51–53.
29.
Helland C, Haugen T, Rakovic E, Eriksrud O, Seynnes O, Mero AA, et al. Force-velocity profiling of
sprinting athletes: single- vs. multiple-run methods. Eur J Appl Physiol. 2018. https://doi.org/10.1007/
s00421-018-4045-2 PMID: 30519907 (forthcoming).
30.
Haugen T, Tønnessen E, Seiler S. 9.58 and 10.49: Nearing the citius end for 100-m? Invited commen-
tary. Int J Sports Physiol Perform. 2015; 10(2):269–72. https://doi.org/10.1123/ijspp.2014-0350 PMID:
25229725
31.
Loturco I, Contreras B, Kobal R, Fernandes V, Moura N, Siqueira F, et al. Vertically and horizontally
directed muscle power exercises: Relationships with top-level sprint performance. PLoS One 2018; 13
(7):e0201475. https://doi.org/10.1371/journal.pone.0201475 PMID: 30048538
32.
Graubner R, Nixdorf E. Biomechanical analysis of the sprint and hurdles events at the 2009 IAAF World
Championships in athletics. New Stud Athletics 2011; 26:19–53.
33.
Haugen T, Tønnessen E, Seiler S. Anaerobic performance testing of professional soccer players 1995–
2010. Int J Sport Physiol Perform. 2013; 8(2):148–56.
34.
Haugen T, Tønnessen E, Seiler S. Speed and countermovement jump characteristics of elite female
soccer players 1995–2010. Int J Sport Physiol Perform. 2012; 7(4):340–9.
35.
Schimpchen J, Skorski S, Nopp S, Meyer T. Are "classical" tests of repeated-sprint ability in football
externally valid? A new approach to determine in-game sprinting behaviour in elite football players. J
Sports Sci. 2016; 34(6):519–26. https://doi.org/10.1080/02640414.2015.1112023 PMID: 26580089
36.
Seiler S, De Koning JJ, Foster C. The fall and rise of the gender difference in elite anaerobic perfor-
mance 1952–2006. Med Sci Sports Exerc. 2007; 39(3):534–40. https://doi.org/10.1249/01.mss.
0000247005.17342.2b PMID: 17473780
Sprint mechanical variables in athletes
PLOS ONE | https://doi.org/10.1371/journal.pone.0215551
July 24, 2019
13 / 14
37.
Haugen T, Paulsen G, Seiler S, Sandbakk Ø. New records in human power. Int J Sports Physiol Per-
form. 2018; 13(6):678–86. https://doi.org/10.1123/ijspp.2017-0441 PMID: 28872385
38.
Tønnessen E, Svendsen I, Olsen IC, Guttormsen A, Haugen T. Performance development in adoles-
cent track and field athletes according to age, sex and sport discipline. PLoS ONE 2015; 10(6):
e0129014. https://doi.org/10.1371/journal.pone.0129014 PMID: 26043192
39.
Haugen T, Solberg PA, Mora´n-Navarro R, Breitscha¨del F, Hopkins W, Foster C. Peak age and perfor-
mance progression in world-class track-and-field athletes. Int J Sports Physiol Perform. 2018; 13
(9):1122–1129. https://doi.org/10.1123/ijspp.2017-0682 PMID: 29543080
40.
Lai A, Schache AG, Brown NA, Pandy MG. Human ankle plantar flexor muscle-tendon mechanics and
energetics during maximum acceleration sprinting. J R Soc Interface 2016; 13(121). https://doi.org/10.
1098/rsif.2016.0391 PMID: 27581481
41.
Miller RH, Umberger BR, Caldwell GE. Sensitivity of maximum sprinting speed to characteristic param-
eters of the muscle force-velocity relationship. J Biomech. 2012; 45(8):1406–43. https://doi.org/10.
1016/j.jbiomech.2012.02.024 PMID: 22405495
42.
Ferro A, Riveral A, Pagola I, Ferreruela M, Martin A, Rocandio V. A kinematic study of the sprint events
at the 1999 World Championships in athletics in Sevilla. 20th International Symposium on Biomechan-
ics in Sports, 2002.
43.
Babić V, Čoh M, Dizdar D. Differences in kinematics parameters of athletes of different running quality.
Biol Sport. 2011; 28:115–121.
44.
Chatzilazaridis I. Stride characteristics progress in a 40-m sprinting test executed by male preadoles-
cent, adolescent and adult athletes. Biol Exerc. 2012; 8:59–77. https://doi.org/10.4127/jbe.2012.0060
45.
Debaere S, Jonkers I, Delecluse C. The contribution of step characteristics to sprint running perfor-
mance in high-level male and female athletes. J Strength Cond Res. 2013; 27(1):116–24. https://doi.
org/10.1519/JSC.0b013e31825183ef PMID: 22395270
Sprint mechanical variables in athletes
PLOS ONE | https://doi.org/10.1371/journal.pone.0215551
July 24, 2019
14 / 14
| Sprint mechanical variables in elite athletes: Are force-velocity profiles sport specific or individual? | 07-24-2019 | Haugen, Thomas A,Breitschädel, Felix,Seiler, Stephen | eng |
PMC10108008 | Received: 19 April 2022 |
Accepted: 16 November 2022
DOI: 10.1111/cpf.12800
O R I G I N A L A R T I C L E
Detailed investigation of multiple resting cardiovascular
parameters in relation to physical fitness
Lars Lind1
|
Karl Michaëlsson2
1Department of Medical Sciences, Uppsala
University, Uppsala, Sweden
2Department of Surgical Sciences, Uppsala
University, Uppsala, Sweden
Correspondence
Lars Lind, Department of Medical Sciences,
Uppsala University, Uppsala 75185, Sweden.
Email: [email protected]
Funding information
Akademiska Sjukhuset
Abstract
Objective: Maximal oxygen consumption at an exercise test (VO2‐max) is a
commonly used marker of physical fitness. In the present study, we aimed to find
independent clinical predictors of VO2‐max by use of multiple measurements of
cardiac, respiratory and vascular variables collected while resting.
Methods: In the Prospective study of Obesity, Energy and Metabolism (POEM), 420
subjects aged 50 years were investigated regarding endothelial function, arterial
compliance, heart rate variability, arterial blood flow and atherosclerosis, left
ventricular structure and function, lung function, multiple blood pressure measure-
ments, lifestyle habits, body composition and in addition a maximal bicycle exercise
test with gas exchange (VO2 and VCO2).
Results: When VO2‐max (indexed for lean mass) was used as the dependent variable
and the 84 hemodynamic or metabolic variables were used as independent variables
in separate sex‐adjusted models, 15 variables showed associations with p < 0.00064
(Bonferroni‐adjusted). Eight independent variables explained 21% of the variance in
VO2‐max. Current smoking and pulse wave velocity (PWV) were the two major
determinants of VO2‐max (explaining each 7% and 3% of the variance; p < 0.0001
and p = 0.008, respectively). They were in order followed by vital capacity, fat mass,
pulse pressure, and high‐density lipoprotein (HDL)‐cholesterol. The relationships
were inverse for all these variables, except for vital capacity and HDL.
Conclusion: Several metabolic, cardiac, respiratory and vascular variables measured
at rest explained together with smoking 21% of the variation in VO2‐max in middle‐
aged individuals. Of those variables, smoking and PWV were the most important.
K E Y W O R D S
exercise test, physical fitness, pulse wave velocity, smoking, VO2‐max
Clin Physiol Funct Imaging. 2023;43:120–127.
120
|
wileyonlinelibrary.com/journal/cpf
This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium,
provided the original work is properly cited.
© 2022 The Authors. Clinical Physiology and Functional Imaging published by John Wiley & Sons Ltd on behalf of Scandinavian Society of Clinical Physiology and
Nuclear Medicine.
1
|
INTRODUCTION
Maximal oxygen consumption at an exercise test (VO2‐max) is
commonly used as a marker of physical fitness. VO2‐max has further
been shown to be related to all‐cause mortality in a dose response
fashion (Kodama, et al., 2009). VO2‐max is generally considered to
have a strong genetic component and twin studies report heritability
estimates of 0.5–0.7, although fitness is naturally also affected by
lifestyle habits (Maes et al., 1996).
There are sex differences in VO2‐max, and VO2‐max declines with
age (Amara et al., 2000; Serrano‐Sánchez et al., 2010) and increasing body
fat (Serrano‐Sánchez et al., 2010). Other important determinants or
consequences of low fitness are lung function at rest (forced vital capacity
[FVC] and forced expiratory volume at 1 s [FEV1]) (Laukkanen et al.,
2009; Mendelson et al., 2016; Nakamura et al., 2004) and smoking
(Bernaards et al., 2003; de Borba et al., 2014; Suminski et al., 2009).
As reviewed by Rost (1997), cardiac enlargement in athletes was first
described by Henshen in 1899 comparing cross‐country skiers with
sedentary controls (Henschen, 1899). Later studies have also evaluated
total heart size in physical fitness (Bouchard et al., 1977), but in most
other studies, the heart size has been divided into left atrial (LA) size, left
ventricular (LV) end‐diastolic diameter and LV mass using echo-
cardiography to give more detailed information. All of these indices of
heart size have been linked to cardiorespiratory fitness (Brinker et al.,
2014; Gidding et al., 2010; Lam et al., 2010; Rogers et al., 2020).
Regarding other cardiovascular parameters, impaired endothelial
vasodilatory capacity (Montero, 2015), increased aortic augmentation
index (AIx) (Binder et al., 2006), increased arterial stiffness (Augustine
et al., 2016; Boreham et al., 2004; Fernberg et al., 2017), poor LV
diastolic function (Brinker et al., 2014), low haemoglobin level
(Laukkanen et al., 2009) and carotid artery atherosclerosis (Rauramaa
et al., 1995) have all been associated with poor VO2‐max.
A major disadvantage with previous studies is that they mainly have
investigated a limited number of cardiovascular and lung function
parameters in the same individuals. Accordingly, no comprehensive
picture of the determinants of VO2‐max has been presented.
With the present study, we, therefore, aimed to measure multiple
cardiovascular and lung function parameters in the same individuals
and to relate those to VO2‐max. We used data from the population‐
based Prospective study of Obesity, Energy and Metabolism (POEM),
in which multiple cardiovascular and lung function parameters have
been measured in the same individuals at the age of 50. We included
all measured cardiovascular and lung function parameters in the
analysis to capture as many facets of cardiorespiratory function as
possible. The hypothesis tested was that we by this approach could
explain a great proportion of the variance in VO2‐max.
2
|
METHODS
In a population‐based study of individuals, all aged 50 years, in
Uppsala City, Sweden, the POEM (Lind, 2013), a random sample of
men and women were invited to participate 1 month following their
50th birthday. The inclusion in the study started in 2012 and was
stopped in 2017. The participation rate was 25%, and the inclusion
was stopped after 502 participants. The study was approved by the
Ethics Committee of the University of Uppsala, and the subjects gave
their written informed consent to participate.
The participants were asked how many times a week they
performed mild (such as walking) and harder (to produce perspiration,
like running) exercises for at least 30 min. Based on these data, four
groups were defined (see Lind et al., 2021 for details): sedentary (13%
of the sample), mild exercise only (24%), some harder exercise (33%)
and harder exercise (30%).
All individuals were investigated in the morning after an
overnight fast. An arterial cannula was inserted in the brachial artery
for blood sampling and was later used for regional infusions of
vasodilators. Lipid variables and fasting blood glucose were measured
by standard laboratory techniques. Height was recorded by a ruler
and body weight was measured on a scale (Tanita BC‐418).
Thereafter, multiple physiological tests were performed.
Endothelial function and arterial compliance/stiffness were both
measured by three different techniques: acetylcholine‐mediated
increase in forearm blood flow, flow‐mediated vasodilation (FMD)
and peripheral artery vasodilation (EndoPath). The carotid arteries
were investigated by ultrasound for anatomy (intima‐media thickness
and echolucency and blood flow. The myocardial LV was evaluated
by ultrasound for LV geometry (LV mass, end‐diastolic volume, wall
thickness), systolic (ejection fraction) and diastolic function (isovolu-
metric relaxation time, E/A‐ratio, Doppler e′/a′ ratio). Blood pressure
was measured by four different techniques (conventional, invasive,
derived central pressure, 24 h ambulatory). Basal energy expenditure
was measured by indirect calorimetry and heart rate variability (HRV)
was recorded for 5 min. Arterial compliance was measured by three
techniques (carotid‐femoral pulse wave velocity [PWV], carotid
artery distensibility and the stroke volume to pulse pressure ratio).
Radial artery pulse wave was recorded for the AIx and reflectance
index. Blood flow of the brachial artery was recorded at rest and
following 5 min of hyperaemia.
Total and regional body fat and lean mass were estimated using
dual‐energy X‐ray absorptiometry (DXA; Lunar Prodigy, GE Health-
care). To minimize the potential operator bias, all scans were
performed in the same room by one experienced nurse. Total fat
and lean mass had a precision error of 1.5% and 1.0%, respectively.
For analysis, automatic edge detection was always used; however, all
scans were thoroughly checked for errors and manually corrected if
needed.
On a separate day, close to the first investigations, the
participants returned to the nonfasted state to evaluate lung function
(FVC and FEV1) and to perform a maximal bicycle ergometer test
with gas exchange recordings. Also, the recoveries of the heart rate,
blood pressure and VO2 and VCO2 during 5 min were recorded.
Smoking was identified as current smoking.
All the physical investigations have previously been described
by Lind and Lampa (2019) and are given in detail in the Supporting
Information.
LIND AND MICHAËLSSON
|
121
FIGURE 1
(See caption on next page)
122
|
LIND AND MICHAËLSSON
2.1
|
Statistical analysis
All variables were checked for a normal distribution, and some
variables such as the E/A ratio, serum triglycerides, most HRV vari-
ables, were skewed to the right and therefore ln‐transformed to
achieve a normal distribution to be used in the models.
First, the relationship between VO2‐max and sex was investi-
gated by ANOVA (same age of all subjects). Second, the relationships
between VO2‐max (adjusted for lean mass) and the 84 hemodynamic
or metabolic variables were investigated one by one in sex‐adjusted
linear regression models. Third, the relationships between VO2‐max
and the 84 hemodynamic or metabolic variables were investigated
one by one with sex and fat mass included in the model. Fourth, the
interactions between sex and the hemodynamic or metabolic
variables were investigated one by one. Fifth, a multiple linear model
with VO2‐max as the outcome and sex together with eight other
hemodynamic
or
metabolic
variables,
which
were
Bonferroni‐
significant in the initial analyses, as independent variables were
evaluated. In this model, variables being closely related (correlation
coefficient > 0.3) to other more significant variables, such as FEV1,
and several blood pressure and heart rate measurements, were not
included in this multiple model due to the risk of co‐linearity. In the
second to fourth steps, the relationships between VO2‐max and the
84 hemodynamic or metabolic variables were investigated one by
one, and therefore, Bonferroni adjustment for these tests was
performed resulting in a critical p‐value of 0.00064. In step five, we
regarded p < 0.05 to be significant.
STATA14 was used for the calculations (Stata Inc.).
3
|
RESULTS
The median and interquartile ranges of measured variables are given
in Supporting Information: Table 1.
VO2‐max (alone and when adjusted for lean mass) was a normally
distributed variable. The mean for unadjusted VO2‐max was 2.79 (SD
0.54) L/min in men and 1.85 (0.40) in women (p < 0.0001). Sex
explained 49% of the variation in unadjusted VO2‐max. VO2‐max
adjusted for lean mass was 0.46 (SD 0.079) L/min/kg lean mass in
men and 0.39 (0.083) in women (p = 0.0043). Sex explained only 1.7%
of the variation in VO2‐max after adjustment for lean mass. In the
following analysis, VO2‐max adjusted for lean mass was used. Current
smoking was reported by 9.8% of the individuals.
When VO2‐max was used as the dependent variable and the
84 hemodynamic or metabolic variables were used as indepen-
dent variables in separate sex‐adjusted models for each hemo-
dynamic or metabolic variable, 15 variables showed associations
with p < 0.00064 (Bonferroni adjusted threshold, see Supporting
Information: Table 2 and Figure 1 for details). Vital capacity, FEV1
and high‐density lipoprotein (HDL) were positively related to
VO2‐max, while the pulse rate, pulse pressure, diastolic night‐
time dipping at 24 h ambulatory recording, BMI, fat mass,
triglycerides, office recordings of the pulse rate, diastolic blood
pressure and calculated central systolic and diastolic blood
pressure all were related to VO2‐max in a negative fashion. All
of these variables displayed p < 0.05 when additional adjustment
for fat mass was made. No interaction term between sex and any
hemodynamic or metabolic variable was significant following
adjustment for multiple testing.
Together
with
smoking,
eight
hemodynamic
or
metabolic
variables being Bonferroni‐significant in the initial analyses explained
21% of the variation in VO2‐max. This held true also after omitting
the variable sex, which was included in the first version of the model.
In this model with VO2‐max as the outcome, smoking and
PWV were the two major determinants of VO2‐max (explaining 7%,
p < 0.0001 and explaining 3%, p = 0.008, respectively). They were
followed by vital capacity, fat mass, pulse pressure and HDL‐
cholesterol, which all showed p < 0.05 in this multiple model (see
Table 1 for details). The relationships were inverse for all these
variables,
except
for
vital
capacity
and
HDL.
Sex
(p = 0.97),
triglycerides and the resting heart rate showed p > 0.05. Figure 2
displays some of these main relationships more in detail.
4
|
DISCUSSION
The present study showed that smoking and an increased PWV at
rest were most closely related to VO2‐max, but lung function, fat
mass, pulse pressure and HDL‐cholesterol were also related to this
commonly used marker of physical fitness.
FIGURE 1
Relationships between hemodynamic and metabolic variables and VO2‐max (adjusted for lean mass) when the hemodynamic and
metabolic variables were evaluated one by one. The regression coefficient and 95% CIs are given for the sex‐adjusted analyses. a, atrial
contraction transmitral filling velocity; AIx, aortic augmentation index; AMBP, ambulatory monitoring of blood pressure; BMI, body mass index;
CI, cardiac index; DBP, diastolic blood pressure; e, early transmitral filling velocity; EDV, endothelium‐dependent vasodilatation; EE, energy
expenditure; EF, ejection fraction; EIDV, endothelium‐independent vasodilatation; FEV1, forced expiratory volume at 1 s; FMD, flow‐mediated
dilatation; HDL, high‐density lipoprotein; HR, heart rate; HRV, heart rate variability; IM‐GSM, echogenicity of the intima‐media complex; IMT,
intima‐media thickness; IVRT, isovolumetric relaxation time; IVS, intraventricular thickness; LA, left atrial diameter; LF, low frequency; LF/HF
ratio, low‐frequency/high‐frequency ratio; LVEDD, left ventricular end‐diastolic diameter; LVESD, left ventricular end‐systolic diameter; LVMI,
left ventricular mass index; PP, pulse pressure; PW, posterior wall thickness; PWV, pulse wave velocity; RHI, reactive hyperaemia index; RI,
reflectance index; RQ, respiratory quote; RWT, relative wall thickness; SBP, systolic blood pressure; SDFV, systolic to diastolic blood flow
velocity; SI, stroke index; SV/PP‐ratio, stroke volume to pulse pressure ratio; TPRI, total peripheral resistance index; VC, Vital capacity; VCO2,
carbon dioxin production; VO2, oxygen consumption.
LIND AND MICHAËLSSON
|
123
4.1
|
Comparison with the literature
All of these variables have previously been shown to be related to
VO2‐max, as cited in the Introduction section. The novelty of the
present
study
is
that
we
by
the
measurements
of
multiple
cardiovascular and lung function variables in the same individuals
were able to compare these variables in terms of importance and
independence from each other.
We could not reproduce some other previous findings that
endothelial vasodilatory function (FMD) (Montero, 2015), a poor LV
diastolic function and a large LV end‐diastolic volume (Brinker et al.,
2014), a low haemoglobin level (Laukkanen et al., 2009) and carotid
artery atherosclerosis (Rauramaa et al., 1995) were related to
VO2‐max.
One major advantage of the present study is that we could
evaluate the independent contribution of indices reflecting different
aspects of physiology in the same model and found that several
different physiological pathways are determinants of VO2‐max. This
is not a surprise, since it is obvious that the heart, the lungs and the
skeletal muscles simultaneously all play important roles in the
determination of cardiorespiratory fitness.
Given that basic assumption, it was a surprise that no variable
reflecting myocardial function or structure was amongst the major
identified physiological indices. One explanation for this could be the
very strict Bonferroni adjustment applied to compensate for the
multiple statistical testing. It could be seen that both the s′ and e′ at
TDI, the e′/a′‐ratio at TDI, stroke index, LA diameter (inverse) and
relative wall thickness (RWT) (inverse) showed p < 0.05 (p = 0.054 for
RWT). Thus, if not using this strict adjustment for multiple testing, we
could replicate the findings of others that several myocardial indices
are linked to VO2‐max, although other factors might be more
important.
Only a small part of the variance in VO2‐max could be
explained by the evaluated variables despite that a great number
of cardiopulmonary variables were assessed. Several factors could
explain this finding. First, all variables have a certain lack of
precision and variability in measurements that could lower the
degree of explained variance, especially when several variables
seem to be of importance. Second, all variables were measured at
rest. It could be speculated that a better R2 for VO2‐max would be
obtained if the variables were measured during exercise instead.
Third, certain factors of particular interest were not measured.
One such very important feature is the mitochondrial function in
the heart and skeletal muscles during exercise. Another could be
diffusion capacity in the lungs. Yet another factor is skeletal
muscle composition, which is important for endurance capacity
(Hall et al., 2021). Fourth, it has been shown that genetic DNA
variations both at the global level (Gineviciene et al., 2022), as well
as at the mitochondrial level (Vellers et al., 2020), are important
determinants of VO2‐max. Fifth, we normalized VO2‐max for lean
mass measured at DXA. Most other studies have not performed
such rigorous normalization, and if no normalization would have
been performed, lean mass in itself would explain 60% of the
variance in VO2‐max.
4.2
|
Clinical perspectives
Apart from an increase in endurance training, smoking cessation
would be the single most important action to improve VO2‐max, as
suggested by the present findings. We could not however find any
intervention studies to support that assumption.
It might also be warranted to reduce arterial stiffness, although
the causality is less clear in this case. In a small placebo‐controlled
trial in postmyocardial infarction patients, treatment with a combina-
tion of a statin and an angiotensin‐receptor blocker reduced PWV
(Turk Veselič et al., 2018). In an open trial of the combination of an
ACE inhibitor and a calcium channel blocker in patients with
hypertension, an improvement in PWV was seen after 12 months
(Radchenko et al., 2018). It would be of interest to see if such
interventions that improve arterial stiffness would also have an
impact on VO2‐max.
TABLE 1
Relationships between VO2‐
max (outcome, adjusted for lean mass) and
sex and eight hemodynamic or metabolic
variables as independent variables
Variables related to VO2‐max
Beta
95% CI low
95% CI high
p Value
Sex
−0.004335
−0.2405681
0.2318981
0.971
Ambulatory pulse pressure
−0.1310065
−0.2354139
−0.0265991
0.014
Smoking
−0.2153444
−0.2965848
−0.1341041
0.000
Fat mass
−0.1095866
−0.1923139
−0.0268592
0.010
Resting heart rate
−0.0134249
−0.1198909
0.093041
0.804
Triglycerides
−0.0539096
−0.1287491
0.02093
0.158
Pulse wave velocity
−0.112147
−0.1944053
−0.0298887
0.008
HDL
0.0920998
0.0008397
0.1833599
0.048
Resting vital capacity
0.1385489
0.0300567
0.2470411
0.012
Abbreviations: CI, cardiac index; HDL, high‐density lipoprotein; VO2‐max, maximal oxygen
consumption.
124
|
LIND AND MICHAËLSSON
FIGURE 2
Relationships between VO2‐max
(adjusted for lean mass) and variables were found to
be of major importance to explain the variation in
VO2‐max (adjusted for lean mass). VO2‐max versus
current smoking is given in the upper panel. VO2‐max
versus pulse wave velocity (PWV) is in the middle
panel and VO2‐max versus fat mass is given in the
lower panel.
LIND AND MICHAËLSSON
|
125
Weight loss might also be a way to increase VO2‐max, and at
least in patients with class III obesity (BMI > 40 kg/m2), weight
reduction increased VO2‐max (Hakala et al., 1996).
4.3
|
Strengths and limitations
The major strength of the present study is the multitude of
cardiovascular and lung function variables measured at rest together
with VO2‐max in individuals of the same age. Since age is an
important determinant of VO2‐max (Amara et al., 2000; Serrano‐
Sánchez et al., 2010), standardization of age would remove the
impact of this very important variable on the variance in VO2‐max.
Another strength is that we could adjust VO2‐max for lean mass,
measured by the gold standard, DXA. As could be seen in our
analysis, this standardization removed most of the sex effect on the
variation in VO2‐max.
This is a cross‐sectional study, and as such causality can never be
proven and the directions of relationships are not clear.
A limitation of studying a homogeneous sample is that the
generalizability is low, so the present results have to be reproduced in
samples from other countries with other ethnical groups, as well as in
other age groups.
5
|
CONCLUSION
Several metabolic, cardiac, respiratory and vascular variables mea-
sured at rest explained together with smoking 21% of the variation in
VO2‐max in individuals aged 50 years.
ACKNOWLEDGEMENTS
The study was funded by the University Hospital of Uppsala, Sweden.
The funders had no role in study design, data collection and analysis,
decision to publish or preparation of the manuscript.
CONFLICT OF INTEREST
The authors declare no conflict of interest.
DATA AVAILABILITY STATEMENT
According to Swedish law, personal health data cannot be made
publicly available. Data from this study are however available upon a
reasonable request by other researchers.
ORCID
Lars Lind
https://orcid.org/0000-0003-2335-8542
REFERENCES
Amara, C.E., Koval, J.J., Johnson, P.J., Paterson, D.H., Winter, E.M. &
Cunningham, D.A. (2000) Modelling the influence of fat‐free mass
and physical activity on the decline in maximal oxygen uptake with
age in older humans. Experimental Physiology, 85, 877–885.
Augustine, J.A., Yoon, E.S., Choo, J., Heffernan, K.S. & Jae, S.Y. (2016) The
relationship between cardiorespiratory fitness and aortic stiffness in
women with central obesity. Journal of Women's Health, 25,
680–686.
Bernaards,
C.M.,
Twisk,
J.W.R.,
Van
Mechelen,
W.,
Snel,
J.
&
Kemper, H.C.G. (2003) A longitudinal study on smoking in relation-
ship to fitness and heart rate response. Medicine and Science in Sports
and Exercise, 35, 793–800.
Binder, J., Bailey, K., Seward, J., Squires, R., Kunihiro, T., Hensrud, D. et al.
(2006) Aortic augmentation index is inversely associated with
cardiorespiratory fitness in men without known coronary heart
disease. American Journal of Hypertension, 19, 1019–1024.
de Borba, A., Jost, R., Gass, R., Nedel, F., Cardoso, D., Pohl, H. et al. (2014)
The influence of active and passive smoking on the cardiorespiratory
fitness of adults. Multidisciplinary Respiratory Medicine, 9, 34.
Boreham, C.A., Ferreira, I., Twisk, J.W., Gallagher, A.M., Savage, M.J. &
Murray, L.J. (2004) Cardiorespiratory fitness, physical activity, and
arterial stiffness: the Northern Ireland Young Hearts Project.
Hypertension, 44, 721–726.
Bouchard,
C.,
Malina,
R.M.,
Hollmann,
W.
&
Leblanc,
C.
(1977)
Submaximal working capacity, heart size and body size in boys
8–18 years. European Journal of Applied Physiology and Occupational
Physiology, 36, 115–126.
Brinker, S.K., Pandey, A., Ayers, C.R., Barlow, C.E., DeFina, L.F., Willis, B.L.
et al. (2014) Association of cardiorespiratory fitness with left
ventricular remodeling and diastolic function. JACC: Heart Failure, 2,
238–246.
Fernberg, U., Fernström, M. & Hurtig‐Wennlöf, A. (2017) Arterial stiffness
is associated to cardiorespiratory fitness and body mass index in
young Swedish adults: the lifestyle, biomarkers, and atherosclerosis
study. European Journal of Preventive Cardiology, 24, 1809–1818.
Gidding, S.S., Carnethon, M.R., Daniels, S., Liu, K., Jacobs, D.R., Jr.,
Sidney, S. et al. (2010) Low cardiovascular risk is associated with
favorable
left
ventricular
mass,
left
ventricular
relative
wall
thickness, and left atrial size: the CARDIA study. Journal of the
American Society of Echocardiography, 23, 816–822.
Gineviciene, V., Utkus, A., Pranckeviciene, E., Semenova, E.A., Hall, E.C.R.
&
Ahmetov,
II.
(2022)
Perspectives
in
sports
genomics.
Biomedicines, 10(2).
Hakala, K., Mustajoki, P., Aittomäki, J. & Sovijärvi, A. (1996) Improved gas
exchange during exercise after weight loss in morbid obesity. Clinical
Physiology, 16, 229–238.
Hall,
C.R.E.,
Semenova,
E.,
Bondareva,
E.A.,
Borisov,
O.,
Andryushchenko, O.N., Andryushchenko, L. et al. (2021) Association
of muscle fiber composition with health and exercise‐related traits in
athletes and untrained subjects. Biology of Sport, 38, 659–666.
Henschen, S. (1899) Skilanglauf und Skiwettlauf: eine medizinische
sportstudie. Mitt Med Klin Upsala (Jena), 2, 15–18.
Kodama, S. (2009) Cardiorespiratory fitness as a quantitative predictor of
all‐cause mortality and cardiovascular events in healthy men and
women: a meta‐analysis. Journal of the American Medical Association,
301, 2024–2035.
Lam,
C.S.P.,
Grewal,
J.,
Borlaug,
B.A.,
Ommen,
S.R.,
Kane,
G.C.,
McCully, R.B. et al. (2010) Size, shape, and stamina: the impact of
left ventricular geometry on exercise capacity. Hypertension, 55,
1143–1149.
Laukkanen, J.A., Laaksonen, D., Lakka, T.A., Savonen, K., Rauramaa, R.,
Mäkikallio, T. et al. (2009) Determinants of cardiorespiratory fitness
in men aged 42 to 60 years with and without cardiovascular disease.
The American Journal of Cardiology, 103, 1598–1604.
Lind, L. (2013) Relationships between three different tests to evaluate
endothelium‐dependent vasodilation and cardiovascular risk in a
middle‐aged sample. Journal of Hypertension, 31, 1570–1574.
126
|
LIND AND MICHAËLSSON
Lind, L. & Lampa, E. (2019) Lifetime change in central and peripheral
haemodynamics in relation to exercise capacity. Clinical Physiology
and Functional Imaging, 39, 261–275.
Lind, L., Zethelius, B., Lindberg, E., Pedersen, N.L. & Byberg, L. (2021)
Changes in leisure‐time physical activity during the adult life span
and relations to cardiovascular risk factors—results from multiple
Swedish studies. PLoS One, 16, e0256476.
Maes, H.H.M., Beunen, G.P., Vlietinck, R.F., Neale, M.C., Thomis, M.,
Eynde, B.V. et al. (1996) Inheritance of physical fitness in 10‐yr‐old
twins and their parents. Medicine and Science in Sports and Exercise,
28, 1479–1491.
Mendelson, M., Michallet, A.S., Tonini, J., Favre‐Juvin, A., Guinot, M.,
Wuyam, B. et al. (2016) Low cardiorespiratory fitness is partially
linked to ventilatory factors in obese adolescents. Pediatric Exercise
Science, 28, 87–97.
Montero, D. (2015) The association of cardiorespiratory fitness with
endothelial or smooth muscle vasodilator function. European Journal
of Preventive Cardiology, 22, 1200–1211.
Nakamura, Y., Tanaka, K., Shigematsu, R., Homma, T. & Sekizawa, K.
(2004) Determinants of cardiorespiratory fitness in patients with
chronic obstructive pulmonary disease, focusing on activities parallel
to daily living. Respirology, 9, 326–330.
Radchenko, G., Mushtenko, L. & Sirenko, Y. (2018) Influence of fixed‐dose
combination perindopril/amlodipine on target organ damage in
patients with arterial hypertension with and without ischemic heart
disease
(results
of
EPHES
trial).
Vascular
Health
and
Risk
Management, 14, 265–278.
Rauramaa, R., Rankinen, T., Tuomainen, P., Väisänen, S. & Mercuri, M.
(1995) Inverse relationship between cardiorespiratory fitness and
carotid atherosclerosis. Atherosclerosis, 112, 213–221.
Rogers, R.J., Schelbert, E.B., Lang, W., Fridman, Y., Yuan, N. & Jakicic, J.M.
(2020) Association of fitness and body fatness with left ventricular
mass: the heart health study. Obesity Science & Practice, 6, 19–27.
Rost, R. (1997) The athlete's heart. Cardiology Clinics, 15, 493–512.
Serrano‐Sánchez, J.A., Delgado‐Guerra, S., Olmedillas, H., Guadalupe‐
Grau, A., Arteaga‐Ortiz, R., Sanchis‐Moysi, J. et al. (2010) Adiposity
and age explain most of the association between physical activity
and fitness in physically active men. PLoS One, 5, e13435.
Suminski, R.R., Wier, L.T., Poston, W., Arenare, B., Randles, A. &
Jackson, A.S. (2009) The effect of habitual smoking on measured
and predicted VO2(max). Journal of Physical Activity and Health, 6,
667–673.
Turk Veselič, M., Eržen, B., Hanžel, J., Piletič, Ž. & Šabovič, M. (2018)
Improving arterial wall characteristics in patients after myocardial
infarction with a very low dose of fluvastatin and valsartan: a proof‐
of‐concept study. Medical Science Monitor, 24, 6892–6899.
Vellers, H.L., Verhein, K.C., Burkholder, A.B., Lee, J., Kim, Y., Lightfoot, J.T.
et al. (2020) Association between mitochondrial DNA sequence
variants and VO2 max trainability. Medicine & Science in Sports &
Exercise, 52, 2303–2309.
SUPPORTING INFORMATION
Additional supporting information can be found online in the
Supporting Information section at the end of this article.
How to cite this article: Lind, L. & Michaëlsson, K. (2023)
Detailed investigation of multiple resting cardiovascular
parameters in relation to physical fitness. Clinical Physiology
and Functional Imaging, 43, 120–127.
https://doi.org/10.1111/cpf.12800
LIND AND MICHAËLSSON
|
127
| Detailed investigation of multiple resting cardiovascular parameters in relation to physical fitness. | 12-01-2022 | Lind, Lars,Michaëlsson, Karl | eng |
PMC6901428 | Vol.:(0123456789)
Sports Medicine (2019) 49 (Suppl 2):S199–S204
https://doi.org/10.1007/s40279-019-01164-z
REVIEW ARTICLE
Genetic Approaches for Sports Performance: How Far Away Are We?
Michael J. Joyner1
Published online: 6 November 2019
© The Author(s) 2019
Abstract
Humans vary in their ‘natural ability’ related to sports performance. One facet of natural ability reflects so-called intrinsic
ability or the ability to do well with minimal training. A second facet of natural ability is how rapidly an individual adapts
to training; this is termed trainability. A third facet is the upper limit achievable after years of prolonged intense training;
this represents both intrinsic ability and also trainability. There are other features of natural ability to consider, for example
body size, because some events, sports, or positions favor participants of different sizes. In this context, the physiological
determinants of elite endurance performance, especially running and cycling, are well known and can be used as a template
to discuss these general issues. The key determinants of endurance performance include maximal oxygen uptake ( ̇VO2max) ,
the lactate threshold, and running economy (efficiency in the case of cycling or other sports). In this article, I use these physi-
ological determinants to explore what is known about the genetics of endurance performance. My main conclusion is that at
this time there are very few, if any, obvious relationships between these key physiological determinants of performance and
DNA sequence variation. Several potential reasons for this lack of relationship will be discussed.
Key Points
‘Natural ability’ or talent is a widely appreciated feature
of many elements of sports performance.
The assumption is that key physiological elements of
talent are embedded in, or explained by, interindividual
differences in DNA sequence.
At this time, interindividual differences in DNA
sequence explain only a small fraction of the physiology
underpinning sports performance.
1 Introduction
Over the past 50 or so years, the key physiological deter-
minants of endurance exercise performance have emerged.
These include maximal oxygen uptake ( ̇VO2max) , the
lactate threshold, and efficiency. In the case of distance run-
ning, efficiency is typically referred to as running economy
because it is difficult to calculate efficiency in a strict engi-
neering context in running humans [1]. By contrast, it is
much easier for cycling.
Data on these three variables can be modeled to predict
performance, and there are field tests that incorporate several
of these variables that are also highly predictive of perfor-
mance. For example, in the early 1990s I took emerging
evidence that humans run the marathon at a pace similar
to their running speed at lactate threshold, and calculated a
theoretical upper limit, at least at that time, for the ‘fastest’
potential marathon performance by men [2]. This model also
reasonably predicted the performance of a given individual.
Likewise, so-called velocity at ̇VO2max was shown to be
highly correlated with running performance [3]. This latter
measure incorporates both ̇VO2max and running economy
into one metric.
The basic idea underpinning these factors is that they
interact in a predictable way. ̇VO2max can be seen as the
upper limit of aerobic capacity, the lactate threshold related
to the fraction of ̇VO2max that can be sustained for a dura-
tion longer than a few minutes, and efficiency or economy
related to the actual power output or speed during a race that
can be generated at a given V̇O2. Additionally, the physi-
ological determinants of ̇VO2max and the lactate threshold
* Michael J. Joyner
[email protected]
1
Department of Anesthesiology and Perioperative Medicine,
Mayo Clinic, Rochester, MN 55905, USA
S200
M. J. Joyner
are well understood. Less is known about the physiological
determinants of efficiency/economy. The question then is,
if the physiological determinants of ̇VO2max and the lactate
threshold are well understood, what is known about the con-
tribution of DNA variation to these factors?
Before I go on, I want to share two sets of assumptions
related to the physiology behind ̇VO2max and the lactate
threshold. First, for ̇VO2max , the primary physiological
determinants under most circumstances in most humans
are related to maximum cardiac output and stroke volume,
along with red cell mass or total body hemoglobin [4]. In
other words, the ability of the heart to pump large quanti-
ties of oxygenated blood to the contracting skeletal muscles
is absolutely critical. While this is not true in every case
and in every circumstance, for example chronic obstructive
pulmonary disease (COPD), where the lungs can become
limiting, it is true for the vast majority of situations. Second,
the lactate threshold reflects, in large part, some combina-
tion of skeletal muscle mitochondrial content and function
in the contracting skeletal muscles and perhaps capillary
density [5, 6]. Efficiency/economy is much more complex
and likely sport-specific. It also has an element of the com-
petitive medium that needs to be considered. Examples
include wind resistance during high-speed cycling versus
lower-speed running, or water resistance for sports such as
swimming or rowing [1].
Therefore, with this general perspective as a background,
I will next try to ask what is known about the genetic con-
tributions to the major physiological determinants of endur-
ance exercise performance. A key question then is what
constitutes ‘genetic’. One approach is to focus on the herit-
ability of key traits related to athletic performance. These are
typically statistical arguments based on the correlation of a
given trait between family members, most notably mono-
or dizygotic twins. If the correlation between monozygotic
twins is greater than the correlation between dizygotic twins
then the interpretation is that this similarity is due primarily
to greater similarities in the DNA of monozygotic twins than
dizygotic twins [7]. For ̇VO2max , the heritability can be very
high for monozygotic twins, consistent with the idea that
there is a major genetic component to this variable. Twin
(and other family) studies also indicate that there is a sig-
nificant genetic component to the increase in ̇VO2max seen
with a few months of fitness-type training [8, 9].
While the observations highlighted above suggest there is
a strong genetic component to training, specific DNA vari-
ants associated with ̇VO2max and how ̇VO2max responds to
training have been hard to find. While a number of small
effect size variants considered in concert seem related to
the rise in ̇VO2max with training, no variants alone or in
combination that are clearly linked to canonical biological
pathways likely to underpin cardiac output and red cell mass
have been identified [10–13].
The issue of limits of genetic ‘causation’ is also part of a
general trend in genomic research for complex human traits
that has accelerated in recent years following the completion
of the human genome project. In the late 1990s and early
2000s, it was generally assumed that a limited number of
gene variants would explain much of the risk of develop-
ing common non-communicable diseases. The idea was that
once these variants were identified, a host of new approaches
to diagnosis, prevention, and therapy would emerge. Unfor-
tunately, this vision has not been realized and hundreds of
gene variants with small effect sizes have been associated
with complex non-communicable diseases. Importantly,
their role in the diagnosis, prevention, and therapy for these
diseases remains obscure. These larger issues related to
genomics and complex disease-related traits have been dis-
cussed in detail elsewhere [14].
2 Oxygen Transport Cascade
Another way to think about endurance performance is via the
so-called oxygen transport cascade (Fig. 1). In this cascade,
the path of oxygen from the air to the tissues is considered.
Therefore, in addition to cardiac output and red cell mass,
factors such as the lung, capillaries, and skeletal muscle are
incorporated into this approach. Using this schematic, it is
possible to further summarize what is known about DNA-
based explanations for differences in other key steps in the
oxygen transport cascade.
2.1 The Lung
A number of genome-wide association studies (GWAS) have
been conducted in an effort to understand the role of DNA
variants in lung function. The vast majority of these have
focused on lung volumes, and there is little information on
diffusing capacity. The take-home message from these stud-
ies is that there are a large number of potential common
DNA variants that explain a tiny fraction of interindividual
differences in lung function. Additionally, when so-called
gene scores (composite values for a number of gene vari-
ants associated with a given phenotype) are constructed,
lung function values in individuals in the highest quartile
or quintile versus the lowest quartile or quintile frequently
differ by only a few percentage points. These differences
typically account for < 0.1 L of a given lung volume and
are within the limits of the test–retest validity of spirom-
etry [15]. Thus, there is no reason to believe that DNA vari-
ants explain any major difference in lung function in elite
athletes, or their extremely high ̇VO2max values. Of note,
individuals who have spent their entire life at high altitude
have increased pulmonary diffusing capacity, but this is an
S201
Limitations to DNA Sequence Explanations for Sports Performance
adaptive response and is not intrinsic to populations who
have lived at high altitude for generations [16].
2.2 Cardiac Output and Stroke Volume
After the lung, the next step in the oxygen transport cascade
is cardiac output. The right ventricle of the heart pumps
blood through the lungs where it is oxygenated and returned
to the left side of the heart, which delivers it to the systemic
circulation. A hallmark of elite endurance performance is
a high maximum cardiac output driven almost exclusively
by a very large stroke volume [17]. To date, no DNA vari-
ants have been described that explain the impressive levels
of stroke volume and cardiac output in elite athletes. Addi-
tionally, no DNA variants have been identified that explain
why some people’s ̇VO2max , and presumably cardiac out-
put, increases more in response to exercise training than
another’s. In the late 1990s and early 2000s, it was thought
that differences in the ACE (angiotensin-converting enzyme)
genotype might contribute to the high stroke volumes and
̇VO2max values seen in elite athletes, based on the potential
for these variants to influence cardiac hypertrophy, but that
seems unlikely at this time [18]. Additionally, the genetic
contributions to maximum heart rate also appear physiologi-
cally trivial—only a few beats per minute [19].
2.3 Red Cell Mass
In addition to cardiac output, red cell mass or total body
hemoglobin are also important physiological determinants
of ̇VO2max . A high cardiac output that pumps anemic blood
will not deliver much oxygen to the periphery. Thus, red
cells and hemoglobin are required, together with a high
cardiac output, to generate the impressive values seen in
elite endurance athletes. At this time, there are no obvious
genetic explanations for the high red cell masses seen in elite
athletes, and these may be more generally linked to plasma
volume expansion with exercise via the so-called critometer
concept; in addition, there are examples of individuals with
rare variants in their erythropoietin-related systems who
have both high hematocrits and high values for ̇VO2max
[20, 21].
2.4 Peripheral Circulation
Once the blood leaves the left ventricle and enters the
peripheral circulation it is delivered to the tissues. A key
determinant of ̇VO2max is the ability to generate very high
skeletal muscle blood flows. It is generally accepted that the
capacity of skeletal muscle to vasodilate exceeds the ability
(at least in humans) of the heart to sustain very high levels
of blood flow in a large mass of active skeletal muscles,
and also preserve blood pressure [22]. This is known as the
‘sleeping giant hypothesis’. Additionally, endurance exercise
training does increase capillary density in the trained skel-
etal muscles, and there are also adaptations at the level of
the resistance vessels and conducting vessels. As is the case
for cardiac output and red cell mass, there is no clear DNA
variant-based explanation for interindividual differences in
these adaptations, or for the very high level of capillary den-
sity that can be seen in some highly trained individuals. It
is also interesting to note that pharmacological blockade of
vascular endothelial growth factor (VEGF) does not elimi-
nate the vascular adaptations in animal models [23]. If at
least some training-induced adaptations can occur when a
key pathway is blocked, it seems unlikely that there might
be a major impact of small effect size gene variants on these
responses.
2.5 Mitochondrial Density
One of the fundamental adaptations to endurance exercise
training is the increase in mitochondrial density seen in
trained skeletal muscle. When this was initially observed
Fig. 1 Schematic representation of the oxygen transport cascade. The
features of the steps in the cascade associated with endurance exer-
cise performance are well known, as is how these steps respond to
training. The intermediate physiology is also well understood (e.g.
the determinants of cardiac output). However, DNA-based explana-
tions for the variability of key steps in the oxygen transport cascade
have been hard to identify, and, as a result of physiological redun-
dancy in adaptive responses, it is unclear whether the search for
DNA-based explanations for the key elements of human performance
outlined in the text will ever be able to tell a detailed deterministic
story
S202
M. J. Joyner
by John Holloszy in the mid-1960s, it was a revolution-
ary finding that initiated the era of exercise biochemistry
[24–26]. Subsequent studies in humans showed that highly
trained individuals with widely different ̇VO2max values
have similar levels of mitochondrial adaptations in their
skeletal muscles [6, 27]. As is the case for VEGF above,
when knockout animals missing so-called ‘master regula-
tors’ for mitochondrial biogenesis are trained, there are still
significant mitochondrial adaptations [28]. Again, if at least
some training-induced adaptations can occur when a key
pathway is blocked or absent, it seems unlikely that there
might be a major effect of small effect size gene variants on
these responses.
While twin studies show that skeletal muscle fiber type
is highly heritable, there is ongoing discussion about so-
called fiber-type transformation in humans in response to
prolonged intense training [29–31]. In this context, a study
in a unique set of identical twins highly divergent for physi-
cal activity over decades showed that muscle fiber type,
especially for ‘slow twitch’ fibers, may be far more plastic
than previously demonstrated (see Fig. 2) [32].
3 Limitations and Potential Objections
to This Perspective
There are a number of potential limitations to the perspec-
tives outlined above. The most obvious is that very large
cohorts of subjects (perhaps numbering in the hundreds of
thousands) in conjunction with the phenotypes of interest
and DNA sequence information are simply not available for
the key steps in the oxygen transport cascade discussed in
this review. For this sort of cohort to be a reality, beyond
a blood test for genotyping, detailed measurements of gas
exchange at rest and during submaximal and maximal
exercise would be needed. Measurements of cardiac output
and red cell mass would also be needed, as would serial
measurements of blood lactate during graded exercise. Mus-
cle biopsies to assess fiber type, mitochondrial function, and
capillary density would also be essential. The financial and
logistical barriers to such a research program seem formi-
dable to say the least.
However, if such a cohort ever did emerge, it seems likely,
based on the data from other phenotypes, that very large
numbers of variants with very small effect sizes (relative
risks of 1.1–1.5 are typically reported) would emerge [33].
Additionally, any rare DNA variants found in smaller case-
control-like studies would likely show declining penetrance,
and thus explain less of the physiology in any larger cohorts
[34]. Importantly, the extent to which these variants would
be causally or ‘casually’ associated with the physiological
phenotype of interest would be uncertain, as would their
overall explanatory power. To address these limitations in
the studies of common disease risk, so-called polygenic gene
scores have been developed [35]. However, the predictive
utility of these scores is questionable for many complex phe-
notypes (e.g. obesity, diabetes, hypertension), and the overall
genetic contribution to the phenotype of interest is much less
than environmental and behavioral influences [36].
A final cautionary note is that for many complex human
phenotypes, genetic association studies can have reproduc-
ibility issues, and also require diverse ethnic cohorts. The
classic example of the reproducibility problem comes from
studies of depression where a recent report found essen-
tially no significant and reproducible genetic associations
for depression [37].
4 Conclusions
The above discussion of the oxygen transport cascade shows
that while there is evidence, based on family and twin stud-
ies, for a genetic component of ̇VO2max and its trainability,
it has been difficult to reconcile these observations with any
specific large effect size gene variants or combinations of
small effect size variants linked to key physiological path-
ways as a whole. Similar comments can be made about
peripheral adaptations in skeletal muscle, and the determi-
nants of efficiency are almost certainly complicated by bio-
mechanical and skill-related factors as much as they are by
genetic components. For considerations such as body size,
similar observations can be made, and even in the case of
ACTN3 variants associated with sprinting or power perfor-
mance, the effect sizes are tiny and there are examples of
elites with the ‘wrong’ genotype [38, 39]. Additionally, in
some sports such as swimming, the ACTN3 genotype does
not clearly segregate in sprinters versus endurance athletes
[40].
0%
20%
40%
60%
80%
100%
Untrained Twin
Trained Twin
Type I Fiber % Distribuon
Fig. 2 Marked differences in percentage slow-twitch fibers from
the vastus lateralis of monozygotic twins aged in their mid-50s who
were highly divergent for physical activity. The active twin had been
engaged in competitive endurance training and competition for dec-
ades [29]
S203
Limitations to DNA Sequence Explanations for Sports Performance
The obvious question is why? One emerging concept is
that there are many potential genetic pathways to a given
phenotype [41]. This concept is consistent with ideas that
biological redundancy underpins complex multiscale physi-
ological responses and adaptations in humans [42]. From an
applied perspective, the ideas discussed in this review sug-
gest that talent identification on the basis of DNA testing is
likely to be of limited value, and that field testing, which is
essentially a higher order ‘bioassay’, is likely to remain a key
element of talent identification in both the near and foresee-
able future [43]. While it is possible that more explanatory
DNA-based associations for complex exercise-related traits
might emerge if detailed physiological phenotyping of large
cohorts of humans is performed, there are many limitations
to this perspective. In this context, the advocates of ever-
bigger Ns should carefully review the limits of this approach
from studies of other complex phenotypes as they make the
case for a ‘more is better’ approach to future studies.
Acknowledgements This supplement is supported by the Gatorade
Sports Science Institute (GSSI). The supplement was guest edited by
Lawrence L. Spriet, who attended a meeting of the GSSI Expert Panel
in March 2019 and received honoraria from the GSSI, a division of
PepsiCo, Inc., for his participation in the meeting. Dr. Spriet received
no honorarium for guest editing the supplement. Dr. Spriet suggested
peer reviewers for each paper, which were sent to the Sports Medicine
Editor-in-Chief for approval, prior to any reviewers being approached.
Dr. Spriet provided comments on each paper and made an editorial
decision based on comments from the peer reviewers and the Editor-
in-Chief. Where decisions were uncertain, Dr. Spriet consulted with
the Editor-in-Chief.
Compliance with Ethical Standards
Funding This article is based on a presentation by Michael Joyner to
the GSSI Expert Panel in March 2019. Funding for attendance at that
meeting, together with an honorarium for preparation of this article,
were provided by the GSSI.
Conflict of interest Michael Joyner has no conflicts of interest relevant
to the content of this article.
Open Access This article is distributed under the terms of the Crea-
tive Commons Attribution 4.0 International License (http://creat iveco
mmons .org/licen ses/by/4.0/), which permits unrestricted use, distribu-
tion, and reproduction in any medium, provided you give appropriate
credit to the original author(s) and the source, provide a link to the
Creative Commons license, and indicate if changes were made.
References
1. Joyner MJ, Coyle EF. Endurance exercise performance: the physi-
ology of champions. J Physiol. 2008;586(1):35–44.
2. Joyner MJ. Modeling: optimal marathon performance on the basis
of physiological factors. J Appl Physiol. 1991;70(2):683–7.
3. Morgan DW, Baldini FD, Martin PE, Kohrt WM. Ten kilometer
performance and predicted velocity at VO2max among well-trained
male runners. Med Sci Sports Exerc. 1989;21(1):78–83.
4. Lundby C, Montero D, Joyner M. Biology of VO2max:
looking under the physiology lamp. Acta Physiol (Oxf).
2017;220(2):218–28.
5. Coyle EF, Coggan AR, Hopper MK, Walters TJ. Determi-
nants of endurance in well-trained cyclists. J Appl Physiol.
1988;64(6):2622–30.
6. Holloszy JO, Coyle EF. Adaptations of skeletal muscle to endur-
ance exercise and their metabolic consequences. J Appl Physiol
Respir Environ Exerc Physiol. 1984;56(4):831–8.
7. Bouchard C, Lesage R, Lortie G, Simoneau JA, Hamel P, Bou-
lay MR, et al. Aerobic performance in brothers, dizygotic and
monozygotic twins. Med Sci Sports Exerc. 1986;18(6):639–46.
8. Prud’homme D, Bouchard C, Leblanc C, Landry F, Fontaine E.
Sensitivity of maximal aerobic power to training is genotype-
dependent. Med Sci Sports Exerc. 1984;16(5):489–93.
9. Bouchard C, An P, Rice T, Skinner JS, Wilmore JH, Gagnon J,
et al. Familial aggregation of VO(2max) response to exercise
training: results from the HERITAGE Family Study. J Appl
Physiol. 1999;87(3):1003–8.
10. Sarzynski MA, Ghosh S, Bouchard C. Genomic and transcrip-
tomic predictors of response levels to endurance exercise training.
J Physiol. 2017;595(9):2931–9.
11. Joyner MJ, Lundby C. Concepts about VO2max and trainability are
context dependent. Exerc Sport Sci Rev. 2018;46(3):138–43.
12. Rankinen T, Fuku N, Wolfarth B, Wang G, Sarzynski MA,
Alexeev DG, et al. No evidence of a common DNA variant
profile specific to world class endurance athletes. PLoS One.
2016;11(1):e0147330.
13. Bouchard C, Sarzynski MA, Rice TK, Kraus WE, Church TS,
Sung YJ, et al. Genomic predictors of the maximal O2 uptake
response to standardized exercise training programs. J Appl Phys-
iol. 2011;110(5):1160–70.
14. Joyner MJ, Paneth N. Promises, promises, and precision medicine.
J Clin Investig. 2019;129(3):946–8.
15. van der Plaat DA, de Jong K, Lahousse L, Faiz A, Vonk JM,
van Diemen CC, et al. Genome-wide association study on the
FEV1/FVC ratio in never-smokers identifies HHIP and FAM13A.
J Allergy Clin Immunol. 2017;139(2):533–40.
16. Cerny FC, Dempsey JA, Reddan WG. Pulmonary gas exchange
in nonnative residents of high altitude. J Clin Investig.
1973;52(12):2993–9.
17. Ekblom B, Hermansen L. Cardiac output in athletes. J Appl Phys-
iol. 1968;25(5):619–25.
18. Rankinen T, Wolfarth B, Simoneau JA, Maier-Lenz D, Rauramaa
R, Rivera MA, et al. No association between the angiotensin-
converting enzyme ID polymorphism and elite endurance athlete
status. J Appl Physiol. 2000;88(5):1571–5.
19. Ramirez J, Duijvenboden SV, Ntalla I, Mifsud B, Warren HR,
Tzanis E, et al. Thirty loci identified for heart rate response to
exercise and recovery implicate autonomic nervous system. Nat
Commun. 2018;9(1):1947.
20. Montero D, Lundby C. Regulation of red blood cell volume with
exercise training. Compr Physiol. 2018;9(1):149–64.
21. de la Chapelle A, Traskelin AL, Juvonen E. Truncated erythropoi-
etin receptor causes dominantly inherited benign human erythro-
cytosis. Proc Natl Acad Sci USA. 1993;90(10):4495–9.
22. Joyner MJ, Casey DP. Regulation of increased blood flow (hyper-
emia) to muscles during exercise: a hierarchy of competing physi-
ological needs. Physiol Rev. 2015;95(2):549–601.
23. Lloyd PG, Prior BM, Li H, Yang HT, Terjung RL. VEGF recep-
tor antagonism blocks arteriogenesis, but only partially inhibits
angiogenesis, in skeletal muscle of exercise-trained rats. Am J
Physiol Heart Circ Physiol. 2005;288(2):H759–68.
S204
M. J. Joyner
24. Holloszy JO. Biochemical adaptations in muscle. Effects of exer-
cise on mitochondrial oxygen uptake and respiratory enzyme
activity in skeletal muscle. J Biol Chem. 1967;242(9):2278–82.
25. Hagberg JM, Coyle EF, Baldwin KM, Cartee GD, Fontana L,
Joyner MJ, et al. The historical context and scientific legacy of
John O. Holloszy. J Appl Physiol. 2019;127(2):277–305.
26. Hawley JA, Hargreaves M, Joyner MJ, Zierath JR. Integrative
biology of exercise. Cell. 2014;159(4):738–49.
27. Lundby C, Jacobs RA. Adaptations of skeletal muscle mitochon-
dria to exercise training. Exp Physiol. 2016;101(1):17–22.
28. Leick L, Wojtaszewski JF, Johansen ST, Kiilerich K, Comes G,
Hellsten Y, et al. PGC-1alpha is not mandatory for exercise- and
training-induced adaptive gene responses in mouse skeletal mus-
cle. Am J Physiol Endocrinol Metab. 2008;294(2):E463–74.
29. Komi PV, Viitasalo JH, Havu M, Thorstensson A, Sjodin B,
Karlsson J. Skeletal muscle fibres and muscle enzyme activities
in monozygous and dizygous twins of both sexes. Acta Physiol
Scand. 1977;100(4):385–92.
30. Simoneau JA, Bouchard C. Genetic determinism of fiber type pro-
portion in human skeletal muscle. FASEB J. 1995;9(11):1091–5.
31. Yan Z, Okutsu M, Akhtar YN, Lira VA. Regulation of exer-
cise-induced fiber type transformation, mitochondrial bio-
genesis, and angiogenesis in skeletal muscle. J Appl Physiol.
2011;110(1):264–74.
32. Bathgate KE, Bagley JR, Jo E, Talmadge RJ, Tobias IS, Brown
LE, et al. Muscle health and performance in monozygotic twins
with 30 years of discordant exercise habits. Eur J Appl Physiol.
2018;118(10):2097–110.
33. Weiss KM. Tilting at quixotic trait loci (QTL): an evolutionary
perspective on genetic causation. Genetics. 2008;179(4):1741–56.
34. Wright CF, West B, Tuke M, Jones SE, Patel K, Laver TW, et al.
Assessing the pathogenicity, penetrance, and expressivity of
putative disease-causing variants in a population setting. Am J
Hum Genet. 2019;104(2):275–86.
35. Saracci R. Epidemiology in wonderland: big data and precision
medicine. Eur J Epidemiol. 2018;33(3):245–57.
36. Said MA, Verweij N, van der Harst P. Associations of com-
bined genetic and lifestyle risks with incident cardiovascular
disease and diabetes in the UK Biobank Study. JAMA Cardiol.
2018;3(8):693–702.
37. Border R, Johnson EC, Evans LM, Smolen A, Berley N, Sullivan
PF, et al. No support for historical candidate gene or candidate
gene-by-interaction hypotheses for major depression across mul-
tiple large samples. Am J Psychiatry. 2019;176(5):376–87.
38. Sexton CE, Ebbert MTW, Miller RH, Ferrel M, Tschanz JAT, Cor-
coran CD, et al. Common DNA variants accurately rank an indi-
vidual of extreme height. Int J Genomics. 2018;2018:5121540.
39. Lucia A, Olivan J, Gomez-Gallego F, Santiago C, Montil M,
Foster C. Citius and longius (faster and longer) with no alpha-
actinin-3 in skeletal muscles? Br J Sports Med. 2007;41(9):616–7.
40. Ruiz JR, Santiago C, Yvert T, Muniesa C, Diaz-Urena G,
Bekendam N, et al. ACTN3 genotype in Spanish elite swim-
mers: no “heterozygous advantage. Scand J Med Sci Sports.
2013;23(3):162–7.
41. Boyle EA, Li YI, Pritchard JK. An expanded view of complex
traits: from polygenic to omnigenic. Cell. 2017;169(7):1177–86.
42. Joyner MJ, Boros LG, Fink G. Biological reductionism ver-
sus redundancy in a degenerate world. Perspect Biol Med.
2018;61(4):517–26.
43. Webborn N, Williams A, McNamee M, Bouchard C, Pitsiladis Y,
Ahmetov I, et al. Direct-to-consumer genetic testing for predicting
sports performance and talent identification: consensus statement.
Br J Sports Med. 2015;49(23):1486–91.
| Genetic Approaches for Sports Performance: How Far Away Are We? | [] | Joyner, Michael J | eng |
PMC9671438 | RESEARCH ARTICLE
The validity and reliability of wearable devices
for the measurement of vertical oscillation for
running
Craig P. SmithID*, Elliott Fullerton, Liam Walton, Emelia Funnell¤, Dimitrios Pantazis,
Heinz Lugo
INCUS Performance Ltd., Loughborough, United Kingdom
¤ Current address: Gymshark, Solihull, United Kingdom
* [email protected]
Abstract
Wearable devices are a popular training tool to measure biomechanical performance indica-
tors during running, including vertical oscillation (VO). VO is a contributing factor in running
economy and injury risk, therefore VO feedback can have a positive impact on running per-
formance. The validity and reliability of the VO measurements from wearable devices is cru-
cial for them to be an effective training tool. The aims of this study were to test the validity
and reliability of VO measurements from wearable devices against video analysis of a single
trunk marker. Four wearable devices were compared: the INCUS NOVA, Garmin Heart
Rate Monitor-Pro (HRM), Garmin Running Dynamics Pod (RDP), and Stryd Running Power
Meter Footpod (Footpod). Fifteen participants completed treadmill running at five different
self-selected speeds for one minute at each speed. Each speed interval was completed
twice. VO was recorded simultaneously by video and the wearables devices. There was sig-
nificant effect of measurement method on VO (p < 0.001), with the NOVA and Footpod
underestimating VO compared to video analysis, while the HRM and RDP overestimated.
Although there were significant differences in the average VO values, all devices were sig-
nificantly correlated with the video analysis (R > = 0.51, p < 0.001). Significant agreement
between repeated VO measurements for all devices, revealed the devices to be reliable
(ICC > = 0.948, p < 0.001). There was also significant agreement for VO measurements
between each device and the video analysis (ICC > = 0.731, p < = 0.001), therefore validat-
ing the devices for VO measurement during running. These results demonstrate that wear-
able devices are valid and reliable tools to detect changes in VO during running. However,
VO measurements varied significantly between the different wearables tested and this
should be considered when comparing VO values between devices.
Introduction
The availability and popularity of wearable sports technology for running has grown exten-
sively in recent years [1]. These devices provide users with feedback about a variety of
PLOS ONE
PLOS ONE | https://doi.org/10.1371/journal.pone.0277810
November 17, 2022
1 / 12
a1111111111
a1111111111
a1111111111
a1111111111
a1111111111
OPEN ACCESS
Citation: Smith CP, Fullerton E, Walton L, Funnell
E, Pantazis D, Lugo H (2022) The validity and
reliability of wearable devices for the measurement
of vertical oscillation for running. PLoS ONE
17(11): e0277810. https://doi.org/10.1371/journal.
pone.0277810
Editor: Bernard X. W. Liew, University of Essex,
UNITED KINGDOM
Received: April 21, 2022
Accepted: November 3, 2022
Published: November 17, 2022
Peer Review History: PLOS recognizes the
benefits of transparency in the peer review
process; therefore, we enable the publication of
all of the content of peer review and author
responses alongside final, published articles. The
editorial history of this article is available here:
https://doi.org/10.1371/journal.pone.0277810
Copyright: © 2022 Smith et al. This is an open
access article distributed under the terms of the
Creative Commons Attribution License, which
permits unrestricted use, distribution, and
reproduction in any medium, provided the original
author and source are credited.
Data Availability Statement: All relevant data are
within the paper and its Supporting Information
files.
biomechanical information, including running specific metrics such as speed and cadence.
The affordability and portable nature of wearable devices make them an attractive method for
measuring biomechanical features of running outside of the laboratory for runners [2],
researchers [3] and clinicians alike [4].
A running specific metric provided by several wearable devices is vertical oscillation (VO).
VO is the vertical displacement of the body during each stride measured at the centre of mass
(COM) [5], or proxy positions such as the pelvis [6]. VO has been linked to running economy
and injury prevention, with smaller VO of the trunk associated with improved running econ-
omy [7, 8] and a reduction in lower-limb injury risk factors such as vertical loading rate [9].
Running technique can be adapted to alter VO [9–11], such as increasing cadence to reduce
VO [12]. Providing a runner with real-time visual and auditory feedback that indicates when
their vertical oscillation is above or below a target level can allow a runner to manipulate their
vertical oscillation as required [13]. Therefore, wearable devices have the potential to provide
an accessible method for runners and coaches to obtain and utilise VO feedback for perfor-
mance gains.
There are a variety of wearable devices currently on the market which provide VO measure-
ments for running. These devices commonly utilise an inertial measurement unit (IMU) to
record body movement and derive a variety of biomechanical features during running. The
position of recording varies between wearables, with some devices positioned on the trunk at
the xiphoid process (Garmin Heart Rate Monitors), C7 vertebrae (INCUS NOVA), waistband
(Garmin Running Dynamics Pod), or on the dorsum of the foot (Stryd Running Power Meter
Footpod). The validity and reliability of VO measurements from wearable devices is essential
to determine whether the device can detect changes in VO or whether changes are the result of
measurement errors. However, few studies have focused on validating wearable devices for
VO measurements. VO recorded from Garmin heart-rate monitors with built in accelerome-
ters (HRM) have been compared to a video analysis method and found to be highly agreeable
[14, 15], as well as reliable between repeated measures [14]. The manufacturers of the Stryd
Running Power Meter Footpod (Footpod) report that the device measures COM VO with a
small average error of 3% when compared to a ground reaction forces method for deriving
COM VO [16]. These findings provide some evidence that wearable technology can be a valid
and reliable tool for measuring VO. However, the validity and reliability of VO for other
devices, such as the INCUS NOVA (NOVA) and Garmin Running Dynamics Pod (RDP) has
not been reported. Furthermore, it is not understood how VO measurements from different
devices compare, especially given they record at different locations on the body.
The aim of this study was to test whether wearable devices are reliable and valid tools for
the measurement of VO during running by comparing VO measurements from four wearable
devices (NOVA, HRM, RDP, and Footpod) to video analysis of a single trunk-based marker.
Based on prior research [14–16], it is hypothesised that the wearable devices will provide valid
and reliable VO measurements when compared to video analysis. However, because of the dif-
ference in body locations between devices, it is hypothesised that the VO measurements will
differ between devices, with the device in closest proximity to the trunk marker (NOVA) hav-
ing the most accurate VO measurements when compared to the video analysis measurements.
Materials and methods
Participants
Fifteen active runners (run for at least 1 hour per week) without any injury in the last 6 months
were recruited (7 females, mean ±SD age = 26.4yrs ±5.5, height = 174.4cm ±9.4,
weight = 71.1kg ±9.3). All participants gave written informed consent, and the experiment was
PLOS ONE
Vertical Oscillation Measured by Wearable Devices for Running
PLOS ONE | https://doi.org/10.1371/journal.pone.0277810
November 17, 2022
2 / 12
Funding: CPS, EF, LW, EF, DP, and HL were
funded by Innovate UK (project no. 00106514).
https://www.ukri.org/councils/innovate-uk/?_ga=2.
89826907.1149472773.1647884579-1155892482.
1640269449. The funders had no role in study
design, data collection and analysis, decision to
publish, or preparation of the manuscript.
Competing interests: I have read the journal’s
policy and the authors of this manuscript have the
following competing interests: The INCUS NOVA
wearable device used in the research article is a
license product of INCUS Performance Ltd. CPS,
EF, LW, EF, DP, and HL were employees of INCUS
Performance Ltd. at the time the research was
completed. This does not alter our adherence to
PLOS ONE policies on sharing data and materials.
conducted in accordance with the Declaration of Helsinki. Ethical approval for this research
was obtained from the Loughborough University Ethics Review Committee (ERSC22_27).
Apparatus
The participants ran indoor on a motorised treadmill (NordicTrack T8.5S, NordicTrack,
Utah, USA) in their usual running footwear at a 1% incline with a fan to mimic outdoor run-
ning [17] (Fig 1). During running, VO was measured using video analysis and four wearable
devices designed to measure VO during running: INCUS NOVA (INUCS Performance Ltd.,
Loughborough, UK), Garmin HRM (Garmin Ltd., Southampton, UK), Garmin RDP (Garmin
Ltd., Southampton, UK), and Stryd Footpod (Stryd, Colorado, US). The wearable devices were
worn, and data recorded, as per their instructions. The NOVA was worn in a purpose-built
harness positioned towards the top of the spine (C7 vertebrae) and paired with the INCUS
mobile application to start and stop data recording. The HRM was fitted using the accompa-
nying chest strap, with the device located on the xiphoid process and paired with the Garmin
Fig 1. Experimental setup. The illustration shows a participant aboard the treadmill and the positioning of the
wearable devices on the body. The INCUS NOVA (NOVA) was worn in a ‘T-Strap’, with the device positioned at C7
vertebrae. The ArUco marker was fixed to the NOVA and video recorded from the rear. The Garmin HRM-Pro chest
strap (HRM) was worn with the device positioned at the xiphoid process. The Garmin Running Dynamics Pod (RDP)
was clipped to the waistband, aligned with the sagittal plane. The Stryd Running Power Meter Footpod (Footpod) was
clipped to the laces of the right trainer. Depictions of the wearable devices are for illustrative purposes only.
https://doi.org/10.1371/journal.pone.0277810.g001
PLOS ONE
Vertical Oscillation Measured by Wearable Devices for Running
PLOS ONE | https://doi.org/10.1371/journal.pone.0277810
November 17, 2022
3 / 12
Forerunner 945 watch (Garmin Ltd., Southampton, UK) worn by the participant on their right
wrist. The RDP was fitted to the rear of the participants waistband aligned with the sagittal
plane and paired with a separate Garmin Forerunner 245 watch (Garmin Ltd., Southampton,
UK) worn on the participants left wrist. Both watches were set to treadmill mode and the
recording of data from both Garmin devices was started/stopped via their corresponding
watches. The Footpod was clipped to the lower two laces of the participants right trainer and
controlled via the Stryd mobile application.
A video analysis system was used to provide a reference to test the validity of the VO mea-
surements of the wearable devices. An ArUco marker, a 5 x 5cm square with a black border
and an inner black and white binary matrix [18], was fixed to the NOVA device and a digital
single-lens reflex camera (Panasonic Lumix DMC-FZ330 Digital, Panasonic, UK) was posi-
tioned on a tripod to the rear of the treadmill at the same height as the marker. The marker
was video recorded at 200 FPS in 4K. It is possible to accurately measure movements of an
ArUco marker <2.2m/s using video recording [19]. Vertical trunk movement during treadmill
running was not expected to surpass this velocity limit under the conditions of this study and
therefore the video recording of an ArUco marker was deemed a suitable method for measur-
ing VO.
Protocol
Participants self-selected a range of five preferred running speeds of 1km/h increments (e.g.,
8–12 km/h) during a five-minute familiarisation period on the treadmill. The participants ran
for one-minute intervals at each speed, with a minute of slow walking (1 km/h) between each
interval. This was completed twice in two blocks, with three minutes of slow walking between
the blocks (Fig 2). Therefore, a total of 10 running intervals (2 blocks x 5 speeds) were com-
pleted by each participant. The order of the running speeds was randomised (pseudorandom
number generator) within each block and for each participant. Data recording from the wear-
able devices were started by the researcher one minute prior to the beginning of the first run-
ning interval and recorded continuously throughout the protocol. Video recording of the
marker was collected by a researcher during each running interval.
Data and statistical analysis
The data recording from the NOVA was downloaded via the INCUS mobile application. The
data from the watches were downloaded using the Garmin connect software for the HRM and
RDP. Footpod data was downloaded using the Stryd Power Center software.
Video recordings of the marker positioned on the NOVA device allowed for automatic
detection of the trunk (C7 vertebrae) and NOVA’s position throughout each trial. ArUco
marker detection was achieved using the open-source library OpenCV [20]. First, using a
checkerboard calibration recording [21], any distortion in the camera image was removed
from each frame. Each frame was then converted to a grayscale image to achieve accurate
marker detection and shorter processing times. The pixel co-ordinates of each corner of the
marker were detected and the location of the centre of the marker derived. Using the marker
centre location, a pixel to centimetre ratio was calculated using the arclength method to mea-
sure the marker perimeter which has a known size (5x5cm). The difference between maximum
and minimum position of the marker in the vertical axis was calculated for each stride and
converted to centimetres using the pixel to centimetre ratio to derive vertical oscillation.
Time synchronisation of VO values between measurement methods was achieved by find-
ing the start and end time of each of the ten running intervals within the VO values recorded
for each method. For each device the starts of the intervals were clearly visible as a rapid rise in
PLOS ONE
Vertical Oscillation Measured by Wearable Devices for Running
PLOS ONE | https://doi.org/10.1371/journal.pone.0277810
November 17, 2022
4 / 12
VO, which remained elevated until the end of the interval when VO would then rapidly fall.
Therefore, the intervals were defined by the rise and fall of VO above and below the mean VO
across the whole protocol for each measurement method. For each interval, the mean and
standard deviation of VO was calculated for the middle 30 seconds, and the corresponding
running intervals were compared between methods.
Differences in average VO (bias and 95% limits of agreement) between the video analysis
and the other methods was calculated and a 1 x 5 repeated measure analysis of variance
(ANOVA) was used to test for a main effect of measurement method on VO. Post-hoc paired
T-tests were then calculated to test for differences between each of the methods, therefore ten
comparisons in total. To reduce the likelihood of Type 1 errors when making multiple com-
parisons, the alpha level for the paired T-tests was Bonferroni corrected, therefore divided by
the number of comparisons (i.e. alpha = 0.05/10). To determine the strength of the relationship
between device VO and the video analysis method, repeated measures correlations [22] were
calculated between the video analysis VO and each of the wearable devices across all running
interval speeds. To determine the reliability of each method, VO measurements during the
Fig 2. Running intervals. An example of the running interval profile for a participant with a preferred running speed
range of 8–12 km/h. Participants ran for one minute at each selected speed with a one-minute break (walking at 1 km/
h) between each running interval. This was completed twice (Block 1 & 2) with the speed order randomised within
each block and a three-minute walking period between the blocks.
https://doi.org/10.1371/journal.pone.0277810.g002
PLOS ONE
Vertical Oscillation Measured by Wearable Devices for Running
PLOS ONE | https://doi.org/10.1371/journal.pone.0277810
November 17, 2022
5 / 12
participants mid-range selected running speed interval in the first block was compared to the
second block using Intraclass Correlation Coefficients (ICC3,1), and the standard error of mea-
surement (SEM) was calculated. To determine the validity of the devices, the VO measure-
ments recorded for the participants mid-range selected running speed interval were averaged
across the two blocks of trials and ICC3,1 were calculated to test the agreement between the
devices and video analysis. Statistical analyses were carried out using Python 3.0.0 library Pin-
gouin 0.5.0, with the alpha level set at 0.05.
Results
The mean (±SD) preferred running speed range was 8.9–12.9 km/h ±1.2. The average VO
measurements for each device across all running intervals are shown in Fig 3. There was a sig-
nificant difference in average VO between the methods of measurement (F(4, 56) = 39.70,
p < 0.001). Post-hoc pairwise comparisons between devices (Bonferroni corrected alpha
level = 0.005) were all significant (t(14) > = 4.32, p < 0.001) other than between HRM and
RDP (t(14) = 0.40, p = 0.692), and NOVA and Footpod (t(14) = 3.16, p = 0.007).
Fig 3. Average vertical oscillation for each method. The box plot shows the distribution of VO values for each device.
Median VO and quartiles 1 (Q1) and 3 (Q3) are show by the box, while lower and upper bars are Q1–1.5Inter-
Quartile Range and Q3 + 1.5Inter-Quartile Range, respectively. There were no outliers above or below the bars for
any method.
https://doi.org/10.1371/journal.pone.0277810.g003
PLOS ONE
Vertical Oscillation Measured by Wearable Devices for Running
PLOS ONE | https://doi.org/10.1371/journal.pone.0277810
November 17, 2022
6 / 12
The Bland-Altman plots (Fig 4), along with the pairwise comparisons reveal that the HRM
(10.8cm ±1.5) and RDP (10.7cm ±2.1) overestimated VO compared to the video analysis
(9.4cm ±1.8), while the NOVA (8.7cm ±1.7) and Footpod (8.0cm ±1.5) underestimated.
Although, there was a difference in the average VO between the devices and video analysis,
the NOVA (R = 0.84, p < 0.001), HRM (R = 0.73, p < 0.001), RDP (R = 0.80, p < 0.001),
and Footpod (R = 0.51, p < 0.001) were significantly correlated with the video analysis values
(Fig 5).
To test the reliability of the VO measurements, the VO values for the participants mid-
range running speed interval were compared between the first (block 1) and repeated (block 2)
measurement. All devices had significant reliability between the repeated measurements
(ICC3,1 > = 0.928, F(14,14) > = 26.87, p < 0.001) and standard error of measurements < =
0.5cm (Table 1).
The validity of each device when compared to the video analysis was also tested for the
mid-range running interval (Table 2). There was significant agreement between all the devices
and the video analysis method (ICC3,1 > = 0.731, F(14,14) > = 6.45, p < = 0.001).
Fig 4. Bland-Altman plots between video analysis vertical oscillation and wearable devices. Bland-Altman plot for
video analysis vertical oscillation values compared to INCUS NOVA (top left), Garmin HRM-Pro chest strap (HRM,
top right), Garmin Running Dynamics Pod (RDP, bottom left), and Stryd Running Power Meter Footpod (Footpod,
bottom right). Mean bias is indicated by the solid line. Dashed lines indicate 95% Limits of Agreement. All running
intervals (n = 10) and participants (n = 15) were included.
https://doi.org/10.1371/journal.pone.0277810.g004
PLOS ONE
Vertical Oscillation Measured by Wearable Devices for Running
PLOS ONE | https://doi.org/10.1371/journal.pone.0277810
November 17, 2022
7 / 12
Fig 5. Video analysis vertical oscillation versus wearable devices. Scatter plots of the video analysis vertical
oscillation measurements (VO) versus VO values from four wearable devices; the INCUS NOVA (top left), Garmin
HRM-Pro chest strap (top right), Garmin Running Dynamics Pod positioned on the waistband (bottom left), and the
Stryd Running Power Meter Footpod (bottom right). All running intervals (n = 10) and participants (n = 15) are
included. The diagonal line represents the line of unity for the video analysis measurements. Repeated measures
correlation R values between the video analysis and devices are shown (p < 0.05).
https://doi.org/10.1371/journal.pone.0277810.g005
Table 1. Reliability between repeated vertical oscillation measurements.
Block 1 Vertical Oscillation (cm)
Block 2 Vertical Oscillation (cm)
ICC [95% CI]
SEM (cm)
Video Analysis
9.5 +/- 1.9
9.5 +/- 1.8
0.928 [0.80, 0.98]
0.5
INCUS NOVA
8.7 +/- 1.8
8.8 +/- 1.8
0.956 [0.87, 0.98]
0.4
Garmin HRM-Pro
10.9 +/-1.8
11.1 +/-1.8
0.948 [0.85, 0.98]
0.4
Garmin RDP
10.7 +/- 2.2
10.9 +/- 2.1
0.968 [0.91, 0.99]
0.4
Stryd Footpod
8.2 +/- 1.5
8.2 +/- 1.4
0.954 [0.87, 0.98]
0.3
Mean (±SD) vertical oscillation for the mid-range running speed interval across all participants for the first and second block of trials. ICCs show agreement between
the blocks for each method (p < 0.05). Standard Error of Measurement (SEM) indicates the amount of variability between the repeated measures due to measurement
error.
https://doi.org/10.1371/journal.pone.0277810.t001
PLOS ONE
Vertical Oscillation Measured by Wearable Devices for Running
PLOS ONE | https://doi.org/10.1371/journal.pone.0277810
November 17, 2022
8 / 12
Discussion
The agreement between the wearable devices and the video analysis reference, along with the
high reliability values between repeated measures, indicate that the wearable devices are valid
and reliable tools for measuring VO of the trunk during running. As hypothesised, the NOVA
measurements had the highest agreement and lowest average bias compared to the video anal-
ysis. However, the absolute VO values differed between the devices, with the NOVA and Foot-
pod underestimating VO compared to the video analysis, while the RDP and HRM
overestimated.
All four wearable devices had VO measurements which significantly agreed with the video
analysis. However, the strength of the agreement varied between devices. Furthermore, the
absolute VO values differed significantly between devices. The largest difference was between
the Footpod and HRM, with the average Footpod VO 26% lower than the HRM. Compared to
the video analysis, the NOVA had the highest correlation and ICC values of the four devices
(R = 0.84, ICC = 0.96), as well as the smallest average bias (0.7cm). The video analysis mea-
sured the vertical movement of a marker fixed to the NOVA, therefore measuring VO at the
C7 vertebrae. In contrast, the HRM recorded from the xiphoid process, the RDP recorded
from the rear of the waistband, and the Footpod from the foot. When compared to video anal-
ysis of a marker fixed to the HRM, the HRM has been found to have strong correlation coeffi-
cients (ICC > = 0.96) and minimal bias (< = 0.3cm) when compared with video analysis
measurements [14, 15], similar to the results for the NOVA in this study. This suggests that a
potential reason for the differences in VO found between the NOVA and HRM is that
although the devices are measuring VO referenced to the location of the device, real differ-
ences in VO between the measurement locations is the explanation for the difference found
between these devices. However, on average there was little difference in VO measurements
between the HRM and RDP, although these devices record from contrasting trunk locations.
This suggests recording location may not be the sole contributor to differences between the
trunk-based devices and that the differences are likely due to a combination of both location
and the device itself. Further research comparing device VO to video analysis at each device
location will help to understand if biomechanical factors contribute to VO measurement dif-
ferences when recording at different locations on the trunk.
Although positioned on the foot, the Footpod reports to measure VO of COM [16]. VO of
the COM is commonly measured using 3D motion capture and a segmental model of the body
is applied to locate COM displacements during running [23]. Measuring VO of COM with
either video analysis of a single marker [24] or a single IMU [15, 25] has proven difficult, with
both methods overestimating COM VO. A linear correction of IMU VO to infer COM VO
has been proposed, although this method is susceptible to overfitting on the sample tested and
Table 2. Validity of vertical oscillation measurements from wearable devices compared to video analysis.
Vertical Oscillation (cm)
ICC [95% CI]
Bias (cm)
95% Limits of Agreement
Video Analysis
9.5 +/- 1.8
INCUS NOVA
8.8 +/- 1.7
0.963 [0.89, 0.99]
0.7
[-0.3, 1.6]
Garmin HRM-Pro
11.0 +/-1.8
0.745 [0.39, 0.91]
-1.5
[-4.1, 1.1]
Garmin RDP
10.8 +/- 2.1
0.858 [0.63, 0.95]
-1.3
[-3.4, 0.8]
Stryd Footpod
8.2 +/- 1.4
0.731 [0.37, 0.90]
1.3
[-1.1, 3.7]
Mean (±SD) vertical oscillation across all participants for the mid-range speed interval. Intraclass correlation coefficients (ICCs) between video analysis and the devices;
INCUS NOVA, Garmin HRM-Pro, Garmin Running Dynamics Pod (RDP), and Stryd Running Power Meter Footpod (Footpod) (p < 0.05). Mean bias (±SD) and
95% Limits of Agreement between video analysis values and the devices are shown.
https://doi.org/10.1371/journal.pone.0277810.t002
PLOS ONE
Vertical Oscillation Measured by Wearable Devices for Running
PLOS ONE | https://doi.org/10.1371/journal.pone.0277810
November 17, 2022
9 / 12
requires validation in different cohorts [15]. The overestimation of COM VO when measured
by a single marker or IMU on the trunk may explain why the trunk located devices had higher
VO values compared to the Footpod which indirectly measures COM VO from measurements
taken at the foot.
Although further research is required to understand the mechanisms for the VO differences
between devices, this finding demonstrates that caution must be taken when using devices
interchangeably. This is an important consideration for users, who may discover significant
changes in their VO values when moving from one device to another. An artificial increase in
VO measurements could lead a user to unnecessarily adapt their running technique to reduce
their VO (e.g., increase cadence) with a negative impact on overall performance. On the other
hand, an artificial decrease in VO measurements could result in the user incorrectly believing
their VO has improved, preventing them from benefiting from improved running economy
[8] and reduced injury risk factors [9] associated with an actual reduction in VO. However,
when interpreting the VO feedback from a single device in isolation, this study has found that
wearable devices can provide a valid and reliable method for the measurement of VO, which is
important for user confidence. The ability to measure VO via a wearable device has the bene-
fits of being unobtrusive and affordable compared to the traditional method of video analysis.
Therefore, wearable devices provide a broader range of runners the opportunity to incorporate
VO feedback into their training.
In this study, VO was measured during treadmill running. Running on a treadmill com-
pared to overground running could potentially increase VO due to flexion in the treadmill
running surface and should be considered when applying the results of treadmill VO studies
to outdoor running. Another external factor known to effect VO is running footwear, with evi-
dence that running barefoot reduces VO compared to shod running [26]. In this study, partici-
pants wore their own choice of running footwear, therefore footwear type was not controlled
for. However, the effect of footwear type on VO was likely minimal considering the effect of
barefoot running has been reported to be a 7% reduction in VO [26].
Conclusions
Wearable devices provide a valid and reliable method for measuring changes in VO during
running when compared to a video analysis method. Therefore, such devices give runners an
accessible option to track changes in their VO with potential performance and injury related
benefits. However, absolute VO values differ between devices, therefore caution must be taken
when using devices interchangeably for VO measurements.
Supporting information
S1 Dataset.
(CSV)
Acknowledgments
The authors would like to thank Emily Codd and Spencer Patmore for their assistance in data
collection.
Author Contributions
Conceptualization: Craig P. Smith, Elliott Fullerton, Liam Walton, Emelia Funnell, Heinz
Lugo.
PLOS ONE
Vertical Oscillation Measured by Wearable Devices for Running
PLOS ONE | https://doi.org/10.1371/journal.pone.0277810
November 17, 2022
10 / 12
Data curation: Craig P. Smith, Elliott Fullerton, Liam Walton, Emelia Funnell.
Formal analysis: Craig P. Smith, Elliott Fullerton, Heinz Lugo.
Funding acquisition: Dimitrios Pantazis, Heinz Lugo.
Investigation: Craig P. Smith, Elliott Fullerton, Emelia Funnell.
Methodology: Craig P. Smith, Elliott Fullerton, Liam Walton, Emelia Funnell, Heinz Lugo.
Project administration: Craig P. Smith, Liam Walton, Heinz Lugo.
Supervision: Dimitrios Pantazis, Heinz Lugo.
Validation: Craig P. Smith.
Visualization: Craig P. Smith.
Writing – original draft: Craig P. Smith.
Writing – review & editing: Craig P. Smith, Elliott Fullerton, Liam Walton, Dimitrios Panta-
zis, Heinz Lugo.
References
1.
Janssen M, Scheerder J, Thibaut E, Brombacher A, Vos S. Who uses running apps and sports
watches? Determinants and consumer profiles of event runners’ usage of running-related smartphone
applications and sports watches. PloS one. 2017; 12(7). https://doi.org/10.1371/journal.pone.0181167
PMID: 28732074
2.
Strohrmann C, Harms H, Tro¨ster G, Hensler S, Mu¨ller R. Out of the lab and into the woods: kinematic
analysis in running using wearable sensors. Proceedings of the 13th international conference on Ubiqui-
tous computing; 2011 Sep 119–122, Beijing, China. New York: Association for Computing Machinery;
2011.
3.
Henriksen A, Haugen Mikalsen M, Woldaregay AZ, Muzny M, Hartvigsen G, Hopstock LA, et al. Using
Fitness Trackers and Smartwatches to Measure Physical Activity in Research: Analysis of Consumer
Wrist-Worn Wearables. J Med Internet Res. 2018; 20(3): e110. https://doi.org/10.2196/jmir.9157 PMID:
29567635
4.
Willy RW. Innovations and pitfalls in the use of wearable devices in the prevention and rehabilitation of
running related injuries. Phys Ther Sport. 2018 Jan; 29:26–33. https://doi.org/10.1016/j.ptsp.2017.10.
003 PMID: 29172095
5.
Anderson T. Biomechanics and running economy. Sports Med. 1996 Aug; 22(2):76–89. https://doi.org/
10.2165/00007256-199622020-00003 PMID: 8857704
6.
Williams KR, Cavanagh PR, Ziff JL. Biomechanical studies of elite female distance runners. Int J Sports
Med. 1987; 8: S107–118. https://doi.org/10.1055/s-2008-1025715 PMID: 3692651
7.
Heise GD, Martin PE. Are variations in running economy in humans associated with ground reaction
force characteristics? Eur J Appl Physiol. 2001; 84(5):438–42. https://doi.org/10.1007/s004210100394
PMID: 11417432
8.
Tartaruga MP, Brisswalter J, Peyre´-Tartaruga LA, Avila AO, Alberton CL, Coertjens M, et al. The rela-
tionship between running economy and biomechanical variables in distance runners. Res Q Exerc
Sport. 2012 Sep; 83(3):367–75. https://doi.org/10.1080/02701367.2012.10599870 PMID: 22978185
9.
Adams D, Pozzi F, Willy RW, Carrol A, Zeni J. Altering cadence or vertical oscillation during running:
effects on running related injury factors. Int J Sports Phys Ther. 2018 Aug; 13(4):633–642. PMID:
30140556
10.
Dallam GM, Wilber RL, Jadelis K, Fletcher G, Romanov N. Effect of a global alteration of running tech-
nique on kinematics and economy. J Sports Sci. 2005 Jul; 23(7):757–64. https://doi.org/10.1080/
02640410400022003 PMID: 16195026
11.
Tseh W, Caputo JL, Morgan DW. Influence of gait manipulation on running economy in female distance
runners. J Sports Sci Med. 2008 Mar 1; 7(1):91–5. PMID: 24150139
12.
Schubert AG, Kempf J, Heiderscheit BC. Influence of Stride Frequency and Length on Running
Mechanics: A Systematic Review. Sports Health. 2014; 6(3):210–217. https://doi.org/10.1177/
1941738113508544 PMID: 24790690
PLOS ONE
Vertical Oscillation Measured by Wearable Devices for Running
PLOS ONE | https://doi.org/10.1371/journal.pone.0277810
November 17, 2022
11 / 12
13.
Halvorsen K, Eriksson M, Gullstrand L. Acute effects of reducing vertical displacement and step fre-
quency on running economy. J Strength Cond Res. 2012 Aug; 26(8):2065–70. https://doi.org/10.1519/
JSC.0b013e318239f87f PMID: 22027846
14.
Adams D, Pozzi F, Carroll A, Rombach A, Zeni J Jr. Validity and Reliability of a Commercial Fitness
Watch for Measuring Running Dynamics. J Orthop Sports Phys Ther. 2016 Jun; 46(6):471–6. https://
doi.org/10.2519/jospt.2016.6391 PMID: 27117729
15.
Watari R, Hettinga B, Osis S, Ferber R. Validation of a Torso-Mounted Accelerometer for Measures of
Vertical Oscillation and Ground Contact Time During Treadmill Running. J Appl Biomech. 2016 Jun; 32
(3):306–10. https://doi.org/10.1123/jab.2015-0200 PMID: 26695636
16.
The Stryd Team. How to lead the pack: Running power meters and quality data [White paper]. Stryd.
2017. Available from: https://storage.googleapis.com/stryd_static_assets/stryd-metric-validation.pdf.
17.
Jones AM, Doust JH. A 1% treadmill grade most accurately reflects the energetic cost of outdoor run-
ning. J Sports Sci. 1996 Aug; 14(4):321–7. https://doi.org/10.1080/02640419608727717 PMID:
8887211
18.
Garrido-Jurado S, Muñoz-Salinas R, Madrid-Cuevas FJ, Marı´n-Jime´nez MJ. Automatic generation and
detection of highly reliable fiducial markers under occlusion. Pattern Recogn. 2014 Jun; 47(6): 2280–
2292.
19.
Wieczorek B, Warguła Ł, Kukla M, Kubacki A, Go´recki J. The effects of ArUco marker velocity and size
on motion capture detection and accuracy in the context of human body kinematics analysis. Technical
Transactions. 2020; 117(1).
20.
Bradski G. The OpenCV Library. Dr. Dobb Journal of Software Tools. 2020; 120: 122–125.
21.
Zhang Z. A flexible new technique for camera calibration. IEEE Transactions on pattern analysis and
machine intelligence. 2000 Nov; 22(11):1330–4.
22.
Bakdash JZ, Marusich LR. Repeated Measures Correlation. Frontiers in Psychology. 2017; 8: 456.
https://doi.org/10.3389/fpsyg.2017.00456 PMID: 28439244
23.
Folland JP, Allen SJ, Black MI, Handsaker JC, Forrester SE. Running Technique is an Important Com-
ponent of Running Economy and Performance. Medicine and science in sports and exercise. 2017; 49
(7): 1412–1423. https://doi.org/10.1249/MSS.0000000000001245 PMID: 28263283
24.
Napier C, Jiang X, MacLean CL, Menon C, Hunt MA. The use of a single sacral marker method to
approximate the centre of mass trajectory during treadmill running. J Biomech. 2020 Jul 17; 108:
109886. https://doi.org/10.1016/j.jbiomech.2020.109886 PMID: 32636000
25.
Gullstrand L, Halvorsen K, Tinmark F, Eriksson M, Nilsson J. Measurements of vertical displacement in
running, a methodological comparison. Gait Posture. 2009 Jul; 30(1):71–5. https://doi.org/10.1016/j.
gaitpost.2009.03.001 PMID: 19356933
26.
Moore IS, Jones A, Dixon S. The pursuit of improved running performance: Can changes in cushioning
and proprioception influence running economy and injury risk? Footwear Science. 2013 5:sup1 S61–
S62.
PLOS ONE
Vertical Oscillation Measured by Wearable Devices for Running
PLOS ONE | https://doi.org/10.1371/journal.pone.0277810
November 17, 2022
12 / 12
| The validity and reliability of wearable devices for the measurement of vertical oscillation for running. | 11-17-2022 | Smith, Craig P,Fullerton, Elliott,Walton, Liam,Funnell, Emelia,Pantazis, Dimitrios,Lugo, Heinz | eng |